Who Plays the Beta and What They Buy
Every market question in Forever comes back to who is standing in it. This is how the beta population is shaped by the test itself, how a class list turns into a shopping list, and which parts of that demand are worth projecting past the wipe.
Nobody has published a census, and that matters
There is no population report for the Forever beta. Blizzard has not released class or race counts, a test client has no public armory behind it, and the guides quoting shares of the population are counting their own readers rather than the server. Anything that looks like a percentage of players on a given class is, at this stage, somebody's sample of a chat channel.
That is a real gap rather than a small one, because demand in a fresh economy is almost entirely a function of composition. A world full of plate wearers buys different things from a world full of casters, and the auction house categories that stay thin are the ones nobody in the population needs. Without a count, the honest move is to describe what can be observed directly and to mark the rest as unmeasured.
3 things can be observed by any player on the beta without waiting for an announcement: who is visible in the capitals and in the open world, what is listed in each auction house category, and what sits unsold. The first is a population read, the second is a supply read, and the difference between them is where demand lives. None of that requires a number from us, which is the point.
That framing needs one correction. On 30 October Blizzard published a snapshot of the beta population: a total count of characters created on the test, the share held by each race, and the share held by each race and class combination. So the accurate statement is not that no census exists, but that there is a single official measurement and no running one. Everything written below about reading the population yourself still applies, because a snapshot taken on one day does not update itself as the test fills up.
Read that release for what it covers and no further. It counts characters rather than accounts, so every alt, every abandoned reroll and every name parked at level 1 sits in the same total as a main that is actually trading. It was published before the 4 November invite wave arrived, which is exactly the period when the starting zones and the bottom of the auction house filled with people who had come in on somebody else's invitation. And it says nothing about faction split, level distribution, playtime, or what any of those characters bought - the parts a seller actually needs. We are not restating the published shares here, because the usable thing is the ordering of races and class combinations rather than a figure copied out of a dated post and read as if it described the server you are standing on now.
A second source has since appeared, and it is not Blizzard's. A public armory sample of 100,000 characters, covering roughly the first 24 hours of the test, has been circulated with 3 breakdowns attached: the class mix across the sample, a faction split that does not come out the same on every ruleset, and the share of characters sitting at level 10. That is more detail than the official snapshot carried, and it is the first time the faction question has been answered with anything other than a guess. It is still a sample rather than a census, so the shares in it describe the characters the armory exposed on 1 day, not the server.
A third dataset has since replaced both of those as the best available read, and the heading above is now out of date. A public armory crawl covers roughly 965,000 characters and reports a median level of 21 and a mean level of 22, a faction split, a split across rulesets, a full class breakdown, a race breakdown, and the share of characters holding guild membership, which it puts at 95.1 percent. We are not reprinting the individual class, race, faction and ruleset shares here, because they move as the crawl refreshes and a figure copied into a guide stops matching the source within days. The ordering is the usable part, and it is worth checking against the live dataset rather than against a quotation of it.
What the aggregate numbers settle is the argument this guide was already making about the level ceiling. A median of 21 and a mean of 22 mean the population is not stacked at the cap and is not evenly spread either - it sits in a band well below the ceiling, which is where the buyer pool for mid-level consumables, bags, weapon upgrades and first-tier profession materials actually lives. The near-total guild membership cuts the other way: when almost every character is in a guild, a large share of gear, enchants and crafted upgrades moves through guild chat and never reaches a listing at all, so auction house volume understates real demand in exactly the categories a guild can cover internally. Price against the levels people are at, and assume the organised part of that demand is being served somewhere you cannot see.
Two of those breakdowns need reading with care before you price anything against them. The level 10 cluster is partly behaviour and partly method - characters do not appear in an armory until they pass a visibility threshold, so the bottom of the distribution collects both the people who logged in once and stopped and the people who had only just become countable. A large share parked there means low-level consumables and first-profession materials have a wide buyer pool and a shallow one at the same time. The faction skew is the more usable figure, because each ruleset runs its own auction house and its own faction balance: a minority faction on one ruleset is a thin supply side and a thin buyer side at once, and that is checkable in your own listings within an hour. Neither breakdown says anything about what those characters spent.
It is worth being precise about how large that skew actually is, because the word is doing more work than the figure. The most lopsided PvP ruleset in the circulated sample is reported at roughly 55/45, which is a lean rather than a collapse. Both factions on it have enough characters to run a working auction house, and the minority side is short of sellers and short of buyers in about the same proportion, so the gap shows up as thinner depth and wider spreads rather than as empty categories. The practical consequence is that a listing on the smaller faction sits longer and moves at a worse fill rate, not that the market is missing. Treat it as a reason to post fewer, larger stacks and to check the other faction's board before assuming a category is genuinely undersupplied.
The skew also has a regional shape, and that part is older than the beta. Players relaying comments attributed to Kaivax report Blizzard acknowledging that the North American realms lean Horde while the European realms lean Alliance, which matches what most people already assumed from their own capitals. We are passing that on as a player-sourced account rather than as a figure we can check, since no breakdown by region has been published alongside the character counts. What it does explain is why two sellers comparing notes across regions can describe opposite markets in good faith.
What turns that lean into something structural is the ruleset layout. Each ruleset runs as its own megaserver with its own auction house, and where a ruleset applies a faction lock the imbalance it starts with is the imbalance it keeps, because nobody can cross over to even it out. So the depth of a category is not one number for the beta - it is one number per faction per ruleset, and the minority side carries fewer listings, fewer competing sellers and fewer buyers at the same time. Price and post against the board you are actually standing in front of, and treat advice written from the other region or the other faction as a description of a different market.
Alongside the datasets there is forum talk, and it deserves a line because it is the only material that describes behaviour rather than character counts. The recurring claim is that the average character sits around level 22 and that low-level dungeons are hard to fill, with players reporting long waits or no group at all in the early instances. We have not verified either half of that, and neither is measurable from a listing page - treat it as a reported impression, not a figure.
If it holds, it is consistent with the level band the crawl describes and it points at a composition question the aggregate numbers do not answer: a population clustered in the same stretch of levels, moving through it quickly, and skipping the content that would normally consume early consumables and crafted gear. Empty early dungeons mean fewer repair cycles, fewer replacement bags and less demand for the first-tier potions and bandages that dungeon runs burn through. Open-world levelling buys a narrower list and buys it in smaller quantities. Before you stock for that, check it the way you would check anything else here - watch whether your own low-level stacks clear, and let your unsold listings settle the question rather than a thread.
There is a mechanical reason that skew stops widening, and it is worth naming because it changes how you read every faction figure on this page. When one side of a realm runs far ahead of the other, character creation on the crowded faction is closed, and new characters there cannot be made until the gap narrows again. Players arriving after that point do not choose the popular side - they roll on the thinner one or they wait. So the faction composition in any dataset is partly a record of what people wanted and partly a record of what the queue allowed on the day they logged in.
For a seller that has 2 practical consequences. The first is that the minority faction is not a static pool shrinking toward irrelevance: it is the side receiving the inflow while the lock holds, so its buyer count climbs even while a snapshot taken earlier still shows it behind. The second is that the new characters arriving under a lock start at the bottom of the level range, which concentrates them in low and mid-level consumables, bags and first-tier profession materials rather than in anything near the ceiling. If you are pricing against a faction share, check whether creation was open on that realm when the figure was taken, because a closed gate makes composition a policy outcome rather than a preference.
There is a second mechanism shaping that skew, and it is not player choice. Faction balance controls on the PvP ruleset have closed Horde character creation, so players who want to roll Horde there cannot, regardless of what the current split looks like. That changes how the ratio should be read: the smaller side is not simply unpopular, it is capped, and the gap between the two factions cannot close from new characters while the lock holds. Anyone planning stock around a minority faction on that ruleset is planning around a buyer pool that is frozen at its current size rather than one that fills in over the following weeks.
The second-order effect is on the population as a whole. Players blocked from creating the character they came for do not always reroll on another ruleset - part of that group has cancelled its subscription instead, and some of them say so publicly. We have no figure for how many, and nobody has published one, so treat it as a direction rather than a size. What it means for the auction house is that a closed faction removes demand from both sides of the server at once: the characters that would have been created are missing, and a share of the accounts behind them stops logging in at all. Composition reads taken before the lock was applied describe a population that no longer exists in the same shape.
The heading on this section should now be read as a record of how the question looked at the start of the test, not as a current statement. RaiderIO runs a statistics section for Forever that covers close to 3.1 million characters, with the population broken out by faction, by class, by race, by level and by ruleset. That is an order of magnitude past the armory samples described above, and it is maintained rather than published once, so the correct phrasing is that an external census exists and is refreshed, while Blizzard's own measurement remains a single dated snapshot. The practical effect for anyone pricing stock is that the composition questions this guide treated as unmeasured are now answerable before you list anything.
Use it for ordering and for cross-sections rather than for exact shares. We are not reprinting the breakdowns, because a crawl that refreshes will not match a figure copied into a guide, and the page itself is the only version that is current. The 2 cross-sections worth pulling first are ruleset against faction, since every ruleset runs its own auction house and its own balance, and level against class, since that tells you which armour type and which consumable tier the buyer pool is actually sitting in. The limits that applied to the earlier datasets still apply here: it counts characters rather than accounts, an abandoned reroll weighs the same as an active main, and nothing in it records what any of those characters spent. Composition is now observable; spending still has to be read off your own listings.
There is a second faction effect that the breakdowns cannot show, because it happens before a character exists. Creation on the crowded side of a PvP ruleset has been closed at points during the test, which means some people who paid for beta access logged in and found they could not roll the faction they came for. Part of that group asked for a refund and left. Part of it rolled on a PvE ruleset instead, where the lock did not apply. Neither group appears in a character crawl as a faction preference - the refunds appear nowhere at all, and the rerolls appear as PvE characters that were never meant to be PvE characters.
For a seller this matters in 2 ways. The locked faction stops taking new entrants at the bottom while the lock holds, so its starting zones, its low-level consumables and its first-tier profession materials lose the steady trickle of fresh buyers that the other side keeps getting, and the age profile of its buyer pool drifts upward. The PvE rulesets absorb that redirected demand without having asked for it, which shows up as deeper bidding on the same low-level categories than their own invite flow would explain. We have no figure for how many accounts were refunded or redirected, and no published one exists, so treat this as a direction rather than a size: check whether creation is open on the faction you are pricing for before you read a thin bottom end as a gap worth filling.
The related argument among players is about free transfers, and nothing about it is settled. No transfer policy for the test has been announced that we have seen, so every version of the discussion is speculation about a feature that may or may not exist. What makes it an economic question rather than a queue question is that a transfer is a wealth channel: characters carry their gold, their bags and their stock with them, so an open door between rulesets or factions moves accumulated capital, not just headcount. If it happens, the receiving side gets a sudden injection of both buyers and undercutting sellers in the same week, and the side people leave loses its deepest listings first. Until there is an actual announcement, the only sensible position is that the split you can measure today is the one to price against.
There is one more thing a dense population does to the bottom of the level range, and it shows up in the auction house before it shows up anywhere else. Respawn timers in the starting zones are fixed and do not widen because more people are standing on them, so quest mobs are contested from the moment an invite wave lands. Objectives that only need kill credit survive that pressure, because credit for a slain mob counts for everyone in the party and a group simply clears the camp faster. Objectives that need a dropped quest item do not divide the same way - the drop goes to 1 looter, so a party of 5 has to kill roughly 5 times as much to finish the same step. The predictable response is that players run the kill quests while the zone is busy and leave the collection quests for quieter hours or for a later pass.
The faction skew on PvP rulesets is not only a thing players react to, it is a thing the realm list reacts to. When one side runs far ahead of the other on a PvP realm, character creation for the heavier faction is closed on that realm, so new arrivals either roll the minority side there or take their group somewhere else entirely. That turns a temporary imbalance into a fixed one for as long as the block stands: the majority side stops growing at the bottom, the minority side collects the new characters, and the composition you read in the capitals today is close to the composition you will be selling into next week.
The second-order effect matters more for demand than the block itself. Large guilds recruit continuously, and a guild that cannot bring its own applicants onto its own realm and its own faction does not wait - it moves, and PvE rulesets are where it moves to, because nothing is closed there. What leaves is the organised part of the population: the raid rosters, the profession coverage inside a guild bank, the people who buy materials in bulk rather than in stacks of 5. A PvP realm that loses several of those keeps its levelling traffic and loses its top-end bidders, while the PvE realm receiving them gets a buyer pool that is older, better funded, and more likely to settle trades in guild chat than in a listing. Check which side of that transfer your own realm is on before assuming a category is thin because nobody wants the goods.
The developers have since confirmed, in their own words, that a large number of new players has arrived on the test. That confirmation came without a figure attached: no total, no rate, no comparison against the earlier waves. Treat it as a direction rather than a measurement. It tells you the buyer pool at the bottom of the level range is growing and that the starting zones and the cheap end of the auction house are the parts under pressure, and it tells you nothing you can price against. Anyone converting that statement into a percentage is filling in a blank the source left empty.
The second skew is the faction lock. Characters are tied to one side, the population does not sit at even halves, and the split reported from the armory samples runs at roughly 45/55. Each faction trades in its own auction house, so an uneven split is not a cosmetic detail - it is 2 markets of different size sharing the same economy. On the smaller side listings are thinner, undercut chains are shorter, and a single seller can hold a category for longer; on the larger side the same goods turn over faster and sit closer to their floor. That skew stacks on top of the level 30 ceiling rather than replacing it. The ceiling decides what people buy by keeping the whole population inside the mid-level bracket, and the faction split decides how many of them are standing in your auction house when you list it. Read your own faction's listings before you read anyone else's numbers.
That routing is what the low end of the market is actually reacting to. Time shifts out of running between objectives and into sustained fighting, which keeps steady pressure on food, drink and bandages rather than on anything situational, and it keeps the cheapest usable weapon or armour upgrade worth listing because a character spends more of its early levels in combat than a solo levelling route would imply. On the supply side the same behaviour floods the auction house with what falls off contested mobs - low-level cloth, skins and the first tier of trade goods - which is why those categories can look deep while the consumables that support the grinding stay thin. Neither effect is measurable from a character count, and both are visible in your own listings inside an evening.
There is a second reason the faction split differs by ruleset, and it is not player preference. When character creation for one faction is closed on the standard PvP ruleset, the people who wanted that faction do not stop playing - they create the character on the RP-PvP ruleset instead, because it is the nearest ruleset that still accepts them. The result is a population on RP-PvP that is partly made of players who never chose a roleplay ruleset for its own sake and would have been somewhere else if the queue had let them in. The skew runs in 1 direction only: toward whichever faction is locked out next door.
For the auction house that matters more than it sounds. A ruleset that absorbs overflow gets both sides of its market moved at once on the receiving faction - more gatherers listing raw materials, more buyers bidding on the same consumables - while the other faction there stays at whatever size it already was. It also makes the composition unstable, because the inflow only happens while the lock is in place and stops when creation reopens. Nobody has published how many characters were made this way, so treat it as a reason to read your own listings on the ruleset you actually trade on rather than as a figure to plan around. The cross-check is simple: if the faction mix in your capital does not look like the mix the sample describes, overflow is the first thing to suspect.
The skew is sharpest on the PvP ruleset, and it is worth separating from the overflow effect described above. A PvP ruleset attracts players who pick a faction for the fight rather than for the city they will sell in, and one side ends up carrying noticeably more characters than the other. That is the point at which faction stops being a flavour choice and becomes a market condition, because each faction trades in its own auction house and those 2 houses are not the same depth.
On 11 November Blizzard acted on that skew directly. Character creation on the PvP rulesets is now being held against a balance target of 50/50, with restrictions applied once a ruleset drifts past roughly 55/45. In practice that means creating a Horde character on the NA PvP ruleset and an Alliance character on the EU RP-PvP ruleset became harder from that date, and the limits are described as adjusting with the balance rather than as a fixed, permanent block.
The market consequence is worth stating plainly: the composition of buyers in an auction house is set partly by who is allowed through the door. A faction under a creation restriction stops taking in new low-level characters, so the bottom of its market - first-profession materials, low-level consumables, starter gear - stops gaining fresh buyers while the unrestricted faction keeps gaining them. The restricted side does not shrink, it just stops growing at the same rate, and the gap shows up first in the categories that only new characters buy. None of this is visible in the earlier snapshot or the armory sample, both of which were taken before the limits were in place, which is another reason to read your own ruleset's listings rather than inherit a faction split from a dated breakdown.
What a shallow side looks like in practice: fewer listings per category, long gaps in the mid-level bands, and whole subcategories with nothing posted at all. Sales are slower because the buyer pool is smaller, but undercutting is also slower, so a listing can sit at its asking figure for a full posting cycle instead of being pushed down within the hour. The crowded side is the mirror image - stock in every band, faster turnover, and competition that resets prices several times a day. Neither side is better to sell on by default; they reward different behaviour, and the same item can be a steady earner on one and dead stock on the other.
You cannot read one faction's auction house from the other, and you cannot move goods between them except through the neutral auction houses, which carry their own cut and their own thin listings. So measure the one you are standing in: pick 3 or 4 categories you actually trade, count what is listed, and come back in 24 hours to see how much of it moved. That read is specific to your faction on your ruleset, which is exactly the resolution no published sample has.
The one official line on scale comes from Microsoft, which has described the Forever beta as the largest beta in the history of the game. That is a statement about size relative to earlier tests, not a measurement: no participant count was attached to it, and no breakdown by faction, class or ruleset was published alongside it. Treat it as confirmation that the buyer pool is wide rather than as a figure you can divide into anything.
For market purposes that still changes the default assumption. A test described in those terms is not a handful of invited guilds trading among themselves, so thin categories on the auction house are more likely to reflect what nobody wants than how few people are logged in. What it cannot tell you is which parts of the population are standing in your faction's capital, and that is the gap the direct observation below is meant to fill.
Alongside the armory sample, players on the forums have been reporting a faction split closer to 60/40 on the PvP rulesets. No source was attached to that figure and nobody has said which realms or which day it covers, so treat it as a shared impression rather than a measurement. It is worth noting only because it points in the same direction as the sampled data: on a PvP ruleset the 2 factions are not the same size, and the gap is wide enough that players notice it while standing in a capital.
What an uneven split does to the auction house is specific. The smaller faction has fewer gatherers feeding raw materials in and fewer crafters converting them, so listings in the bulk categories run shallow and refill slowly after someone clears them. It also has fewer buyers, which means a stack that would move in an hour on the larger side can sit for a full posting cycle. Those 2 effects pull against each other rather than cancelling: prices are less predictable, single listings move the visible price more, and a category can look empty because nobody is supplying it or because nobody is asking for it. The way to tell them apart is to post a small quantity and watch what happens, since the split itself does not say which side of the market is thin on your ruleset.
One number has been attached to scale, and it comes from the Chinese side of the game rather than from the beta. NetEase, which operates the game in China, has reported a peak of 1 million players there. That figure describes a live service in 1 region, not the test population, and it is not a count of anyone standing in the auction house you trade on. It is worth knowing because it is the only published player number of that size in circulation, and because almost every secondhand estimate of how many people are online traces back to it.
That is the caveat to carry into the rest of this note. Where the text below talks about a wide buyer pool or a thin category, it is describing what can be seen in listings and in capitals, not a measurement handed down by the developer. Any total-online figure you meet in coverage of the beta, including ours, is a retelling of someone else's published number or an inference from one. Treat the ordering of things as the usable part, and check the size of your own market by what sells rather than by a headline count.
Two further claims have been made in the same register. Blizzard has described the test as setting a record for peak concurrent players, and has separately described a record number of returning players coming back for it. Both were stated as records rather than as measurements: no concurrency figure was given, no returning-account count was given, and no comparison baseline was named beyond earlier tests of the game. They confirm direction and nothing more.
For a seller the useful part is the second claim rather than the first. A returning player is not a new account at level 1 with no gold and no bags - that profile arrives with old habits, knows which professions are worth levelling and which consumables are worth paying for, and starts bidding further up the level range than a guest who was handed a code by a friend. A population weighted that way buys deeper into the mid and upper brackets than the starting-zone traffic suggests. How large that share is, nobody has published, so treat it as a reason to check the upper categories yourself rather than as a number to plan against.
One detail about how people got in changes the way that population should be read. Access arrived as invite codes, and a code is not a pass that expires with the week it was handed out for. An account that redeemed one for the 4 November to 11 November window is not cut off when the window closes: from that point the test client sits behind an ordinary subscription, with no second code and no renewed invitation needed to keep logging in.
The practical effect is that the buyer pool does not reset on a date. Whoever is standing in the auction house during the invite week can still be standing in it afterwards, so the attrition you see is people losing interest in a wiped test rather than people losing access to it. That makes the composition read above worth repeating later rather than treating as a snapshot of one week - the same characters, classes and listings can be checked again once the first wave of curiosity has drained off, and the difference between the 2 reads is the part that tells you who actually intends to trade. How many invited accounts carried on is not something we can count, and nobody has published a figure for it.
The 4 November to 11 November window also brought in a second kind of participant. Alongside the redeemed codes, invited accounts were able to pass access to friends, so a share of the characters logged in during that week belonged to people who had not applied for the test at all and had no particular interest in it beyond a look. Those guest arrivals cluster in the first days after an invite goes out, which is exactly when the starting zones and the low end of the auction house are busiest. We have no count of how many accounts came in that way, and no breakdown of how long they stayed.
One more mechanism shapes who is standing in the market, and it sits before the character ever loads in. On the PvP ruleset, Horde character creation has been blocked while the faction balance sits against it, and the block is a refusal rather than a queue. There is no position in line, no estimated wait and no notification when a slot opens, so the way players deal with it is to sit on the creation screen and retry, sometimes for hours, until an attempt happens to land in a window where creation is open again.
For a demand read that matters more than it looks. The players held at that screen are not absent from the test, they are delayed, and they arrive in clumps whenever the gate loosens rather than as a steady trickle. One faction's starting zones therefore fill on a different schedule from the other's, and the low end of its auction house lags behind - not because nobody wants the goods, but because the buyers who would bid on them have not been allowed to exist yet. Treat a thin category on the blocked side as a timing artefact first and a genuine gap second, and check it again after a batch of new characters gets through. How long the block stays in place is set by the balance it is correcting, and nobody has published a schedule for it.
There is also a limit on who can join which side at all. A faction balance restriction has been applied on the PvP and RP-PvP rulesets that blocks the creation of new Horde characters there, so a player who wants to roll Horde on those realms is turned away at the character screen rather than choosing something else. Public population trackers maintained by players show the same rulesets sitting on an Alliance majority. We are not quoting shares from those trackers, because they are unofficial counts assembled from scanned data and they move while the restriction is in place.
For a seller the consequence is more useful than the cause. The buyer mix on a locked ruleset is not the mix the population picked - it is the mix that was still available to pick, and the faction that is closed off stops growing while the open one keeps taking arrivals. That pushes the auction house on the blocked side toward a static set of established characters, and the open side toward a steady flow of new low-level traffic. Neither pattern is a preference, so treat both as a product of the restriction and expect the composition to shift again if it is lifted. How long it stays on, and whether it is applied to further rulesets, has not been stated.
For demand that matters more than the raw headcount. A player who joined on somebody else's invitation starts at level 1 with no bank, no bags and no gold, and buys accordingly: starter consumables, the cheapest usable weapon, light armour in the first few brackets, and whatever stack of cloth or leather a new profession needs. None of that touches the categories a committed tester trades in. If you read the auction house during that first week and project it forward, you are projecting a buyer mix that was heavier at the bottom of the level range than it will be once the curious have stopped logging in - which is the second reason to repeat the composition read later rather than trusting a single pass.
There is a third reason, and it is structural rather than seasonal. The beta runs on a single realm, and character creation on it has been closed at points during the test. While that lock is in place the faction mix is frozen in whatever shape it had when the gate shut: a player who logs in and finds one side thin cannot roll on the other side to balance it, and nobody outside can arrive to fill the gap. The population you are reading is not drifting toward an even split, it is held where it was.
That gate is not always a blanket one. On a PvP ruleset the block can apply to a single faction at a time: when one side runs ahead, creation on that side is throttled while the other stays open, and a player who tries anyway is turned away with a population imbalance message rather than a queue. The intent is to stop the larger faction growing further, but the side effect is that arrivals during a throttled period are sorted by what the gate allows rather than by what they would have picked. Anyone who wanted the crowded faction either waits, rolls on the other side, or does not roll at all.
For a seller that is a composition problem, not a fairness one. Each faction trades in its own auction house, so a throttle applied during an invite wave decides which of the 2 markets absorbs the new low-level buyers - starter consumables, first-bracket armour, cheap profession stacks - and which one keeps the buyer mix it already had. It also distorts the class read described above, since the races available to a player are set by the faction they were allowed to create on. When you repeat the composition check later, note whether creation was open on both sides at the time, because 2 reads taken under different gate states are measuring different populations.
Creation limits do not always arrive as a full stop. A realm that leans heavily to one side can have new characters throttled on the crowded faction while the thin one stays open, which is a tool for steering arrivals rather than for stopping them. Either version has the same effect on the auction house: the mix of buyers you are selling to was decided at the character select screen, not by what the market needed. If the side you rolled on was the one being held back, your faction keeps the population it already had while the other side takes in whoever turned up next.
That is worth checking before you read a thin category as a gap in supply. Each faction trades in its own auction house, so a listing that sits unsold on one side can be short on the other, and the only shared ground is the neutral houses in the goblin towns. A skewed realm produces exactly that pattern: the larger side has more competing sellers and faster undercutting, the smaller side has fewer listings and fewer bidders, and neither reads as a clean picture of what the test population wants. Which side is currently ahead, and whether creation is open on both, are things to confirm in the client at the moment you are trading - we have no published split to quote, and the state of the gate has changed more than once during the test.
For a seller that turns a composition quirk into a standing condition of the market. Each faction has its own auction house, so a skew in who is logged in is also a skew in who bids, and the thinner side stays thin in the categories its classes use - armour types, weapon classes, the reagents tied to whichever races and professions are over-represented. Listings that would clear on a busy faction can sit unsold across the lock simply because the buyers for them were never allowed onto the realm. Cross-faction trade through the neutral auction house is the only route around it, and it carries the usual cut and the usual wait. When creation reopens the mix can move again, so a read taken during a lock describes that window and not the test as a whole.
One external dataset does cover part of the gap, and it is worth naming for what it is. A community census addon, which logs every character its users walk past, has published a beta sample of 375,806 characters split 48.4 to 51.6 between the 2 factions - close enough to even that neither side reads as underpopulated. The same dataset ranks races and classes by how often they were seen, and puts 20.5% of the logged characters at level 20, the cap in force while that sample was being collected. The rankings are the usable part of it; we are not reproducing per-class shares here, because those figures shift as more clients report in and a number copied out today does not describe the sample a week later.
The collection method decides how far that goes. An addon census counts only the characters its users actually encountered, so it overcounts whoever stands in the capitals and on the busy early routes and undercounts anyone levelling in a quiet zone, parked offline, or playing at hours when few reporters are online. It is also a sample of people who install that addon rather than a roster of the server, and it inherits the same level ceiling as everything else on the test client. Read it as a rough shape - factions roughly balanced, most characters bunched at the cap, a visible ordering of races and classes - and keep using the direct observation below for anything you intend to act on.
There is a second way the population skews, and it has nothing to do with levels. Large streamer groups tend to arrive together and announce a side before they log in, and on a PvP ruleset that announcement usually lands on Alliance. Their viewers follow, roll the same faction on the same realm, and for as long as the stream runs that faction's capital, its contested zones and its auction house all look busier than the census would suggest. The characters are real and the bids are real, but the cohort has a shorter life than the test around it.
When the group moves on, the demand it carried goes with it, and the drop is not spread evenly. Consumables and cheap PvP-adjacent gear fall off first because the viewers buying them were levelling fresh characters, while the faction balance in open-world zones can flip inside a day or 2. If you build stock against what that side was bidding on at its peak, you are holding inventory for buyers who no longer log in. The read worth trusting is the one taken on an ordinary evening with no large group online, and it is worth asking whether a busy Alliance market is a steady population or a temporary one before committing materials to it. How many players arrive or leave this way is not something anyone has counted.
That sample has since grown, and it is now large enough to quote as more than a hint. The NayKu census addon reports a beta dataset of 1,000,000 characters, split 50.4 to 49.6 between the 2 factions - even enough that neither auction house is trading against a thin buyer pool. Within each faction the most common race is human at 32.3% on one side and Undead at 33.5% on the other. Hunter is the most common class across the whole sample at 15.6%, and the Skyborne account for 18.7% of logged characters, with 31.7% of them rolled as druids. The collection caveats above have not gone away - this counts characters its users walked past, not a server roster - but at this size the ordering of races and classes is stable enough to plan purchases against.
The heading above this section is now out of date, and it is worth saying so plainly rather than leaving the older framing to stand. A census has been published for the test: the Forever Census reports 1,148,135 unique characters logged over the beta, split 48.6% Alliance to 51.4% Horde. On the Alliance side the ordering is human at 32.3%, Skyborne at 18.7% and night elf at 17.7%. On the Horde side it is Undead at 33.8%, orc at 18.8% and Skyborne at 18.6%. By class the top of the list is hunter at 15.3%, warrior at 12.2% and paladin at 11.7%, and the single combination worth calling out is the Undead paladin, which accounts for 6% of all logged characters on its own.
A second official figure has been attached to the test: Blizzard has said 8.8 million characters were created during the beta. That number has been doing the rounds on the forums as the basis for a headcount, with players dividing it by the character cap that applied during name reservation and arriving at estimates somewhere between 1 and 3 million accounts. The arithmetic is tidy and the conclusion is not, because the divisor does not describe the test.
The forum version of that figure has drifted, which is worth knowing before you borrow it. Threads discussing beta scale usually round it to 8 million created characters, and a second number travels with it: roughly 2 million logins, quoted by players as though it were published alongside the character count. It was not. Nobody has attributed the login figure to a Blizzard statement, and the people repeating it are repeating each other, so it belongs in the same bucket as the division arithmetic above - a plausible order of magnitude with no source behind the digits.
The argument that follows is about attrition rather than headcount, and it is the part that actually touches the auction house. One camp holds that most of those characters never leave the early zones, so the buyer pool at the level cap is a small fraction of any creation total. The other holds that a test advertised this widely pulls in players who intend to finish, and that the share reaching 60 is far higher than the usual free-trial pattern suggests. No figure has been published for either position, and the census samples described above cannot settle it, because they log the characters their users walk past rather than tracking accounts over time. For planning purposes, treat demand at the cap as something you read off listings and unsold stock in your own faction's auction house, not something you derive from a creation number.
One more thing shapes who ends up standing in each auction house, and it starts before anyone logs in to play. On the PvP ruleset, character creation for the over-represented faction is being blocked during the paid name reservation period: a player reports being unable to create a Horde character there at all, with the option closed rather than queued or delayed. This is a mechanical faction balance measure, not a bug report, and it applies while names are still being claimed.
The market consequence is that the buyer mix on that ruleset is partly decided in advance. Whoever gets in first fills the open side, the locked side stops growing until the lock lifts, and the auction house that opens at launch has a buyer pool whose faction split was set by creation rules rather than by player preference. If you are planning stock for a PvP realm, check which side is open when you reserve, because the faction you can roll is the faction whose consumables, armour types and weapon skills will be bid on in volume early. We cannot say how long any individual lock stays in place or which factions are affected on which realms - that is set per ruleset and has not been published as a list.
The reason that spread matters more here than it would on a normal realm is that the beta runs as a megaserver, with 1 auction house serving everyone rather than a separate market per realm. Every seller lists into the same pool and every buyer searches the same pool, so the gap between the low and the high end of those population estimates is a gap in how many people are refreshing the same search at the same moment. On a small realm an estimate being off by a factor of 2 changes how long a listing sits; on a shared market it changes whether a listing is ever the cheapest one on screen.
What could be observed directly during the test is more useful than the headcount argument. Underpriced stacks did not sit: anything listed below the standing price in its category tended to disappear within the first minute or 2, often before the lister had walked away from the auctioneer. That behaviour is consistent with a large buyer pool watching a single market, and it has a practical consequence for sellers - a listing priced to move does not reward you with a faster sale so much as it hands the margin to whoever is scanning, since the undercut is bought and relisted higher rather than consumed. It also means a thin category on a megaserver is a genuine demand signal, because there is no second realm where the buyers might be standing instead.
3 things break it. Characters created and then deleted still count toward the total, and a wiped test invites exactly that kind of churn - rolled, levelled a few hours, scrapped for a different class. The cap used in the division belongs to name reservation rather than to the beta itself, so an account that filled its slots during reservation was never limited to that many characters once the client opened. And an account with several alts is still 1 person standing in 1 auction house, buying 1 set of consumables at a time. Divide the created-character total however you like and the output is a range of account estimates, not a count of buyers, and the gap between those 2 things is where the demand read actually sits. Nobody has published an account figure, and we are not going to produce one by arithmetic.
Official totals have since been published as well, and they are larger than any addon sample by an order of magnitude. Blizzard puts the beta at 8,862,354 characters created, with 1,889,084 of them Skyborne and 612,096 rolled as Undead paladins. Those 3 figures line up with the ordering the community census was already showing - a popular new race and one class and race combination that far outruns the rest - so the direction of the sample was right even where its absolute numbers were not.
The one thing those totals do not do is count players. They are characters, not accounts: a single person who rolls a bank alt, a second faction and 3 abandoned level 5 starts appears 5 times in them, and the test actively encouraged that by wiping everything at the end. Nothing published states how many accounts those characters belong to, so the ratio between the 2 is unknown and cannot be estimated from what is on the record. Read the official figures as proof that the buyer pool is wide and that the Skyborne and the Undead paladin will be common on launch day, and keep using what is actually standing in your faction's capital for anything you intend to price against.
A census counts what people rolled on the test; it does not say which ruleset they intend to play once the wipe is behind them. The only numeric read on that question comes from a community poll on noob-club, where readers were asked which realm type they plan to start on. Of 87 votes, 50.5% picked PvP, 36.4% picked PvE, 6.1% picked hardcore, 4% picked RP-PvE and 3% picked RP-PvP. That is the shape of stated intent rather than of anything measured on a server.
Treat the weight of it accordingly. 87 responses is a small sample, it was collected on a single site in one language, and polls of this kind attract the people who already read about the game rather than a cross-section of everyone who will log in. What it is useful for is the ordering: PvP ahead of PvE by a clear margin, with hardcore and the 2 roleplay options splitting a thin remainder between them. If that ordering holds, the categories that sell on a PvP realm - consumables carried in bulk, cheap replacement gear, anything bought again after a death - matter more to plan around than the ones that only move on a quiet server. It is a hint about where the buyers go, not a count of them.
Read as a shopping list, that mix points the same way the observation method below does. Hunters and warriors between them cover more than a quarter of the population, so leather and mail levelling pieces, ammunition, weapon stock and cheap stat food move against a wide buyer pool. Paladins at 11.7%, with the Undead variant carrying more than half of that share by itself, put steady pressure on plate, shields, strength and intellect gear and anything that supports a hybrid levelling kit - a demand block that did not exist in older versions of this race and class table. The 2 factions sitting within 3 points of each other means neither auction house is the thin one by default. All of this describes the beta and only the beta: the characters behind these percentages are wiped, and the mix that shows up at release will be made by different people with a different cap in front of them, so treat the shares as a guide to what sells on the test rather than a forecast of the live market.
The small lean towards Horde has an explanation that does not need a survey behind it. The race and class table opened combinations on that side which no previous version of the game allowed, and the Undead paladin is the clearest example: it alone carries 6% of every character the census logged, which is roughly half the paladin population of the whole beta. A player choosing a side for the test is choosing between something they have not levelled before and something they have, and on the Alliance the headline picks - human, night elf, the familiar warrior and paladin pairings - are mostly reruns of characters people already own on live realms. That is enough to move a test population by a few points without anything else changing, and it explains why the gap is 3 points rather than a landslide: novelty shifts the undecided, it does not empty a faction.
For the auction house the effect is narrower than the headcount suggests. A new combination concentrates demand on the gear block its class needs rather than the one its race implies, so the Horde side of the market carries more plate, shield, strength and intellect pressure than a faction with that race list would normally produce, and it carries it from low levels upward because these are first characters rather than alts with a bank behind them. Alliance buyers spread across the older pattern - leather and mail levelling kit, weapons, ammunition, stat food - with more of them funded by players repeating a route they know and buying only where it saves time. Neither side is the thin one, so the practical read is about which categories clear quickly on each auction house rather than which faction has customers at all. Whether the same novelty holds attention after the wipe is not something the beta can answer.
The figure that actually reshapes the demand read is 6.8%, the share of logged characters sitting at the level cap. More than 9 in 10 characters in the sample are below it, which is the clearest confirmation of what this page argued before the data existed: the beta population is built out of short-lived alts, not mains. Spending follows that shape. Bags, low-level food and drink, first-tier materials and cheap levelling upgrades get bid on constantly because almost everyone in the market is in that bracket, while anything aimed at a finished character competes for a sliver of the population. A category that looks dead on the test is usually not unwanted - it is priced for buyers who were deleted at level 12 and rerolled. At launch the cap moves and that distribution inverts, which is why the thin end of the beta auction house is the part worth stocking for later rather than writing off.
The minority that does sit at the cap spends in a very different way, and one reported profile shows how. A tester describing 17 days of /played time on the beta says the end of that run is premade battlegrounds with twinked toons, on the stated grounds that once a character reaches 30 there is nothing left to do but PvP. That is one player's account rather than a measurement of the population, but it names a buyer the auction house meets often enough to plan for.
The question that follows is what this buyer pool looks like half a year after launch, and the honest answer is that nobody knows yet. Forum threads on the subject lean on the recent precedent: players point at TBC Anniversary and at MoP Classic, argue that both filled up at the start and thinned out once the first stretch of content was cleared, and project the same curve onto Forever. Specific drop-off figures get quoted in those threads, but they come from player estimates and addon samples rather than from any published count, so treat them as a claim about a pattern and not as a measured number. What the pattern implies for trading is simple enough: the categories that sell on a crowded server are not the ones that sell on a quiet one, and the levelling-bracket demand described above is the first thing to go when the opening wave drains off.
The counterargument is built into how Forever is supposed to run. The stated plan is a world that keeps growing - new zones and new raids added on a rolling schedule of roughly 24 months - rather than a fixed set of tiers that ends. If that holds, the population does not follow the single spike-and-decay shape of a Classic re-release, because each addition brings its own return wave and its own fresh set of materials, consumables and gear to trade. 2 practical conclusions sit in that gap: stock for the cap bracket rather than the levelling bracket once the first months are behind you, and treat any content addition as the point where thin categories become worth listing again. How large those return waves are, and whether the 24-month cadence is kept, is unproven until it happens at least once.
This buyer shops narrowly and repeatedly. The list is the best gear the bracket allows rather than the cheapest that works, enchants and other permanent upgrades on every slot that takes one, and a standing supply of consumables that gets burned off in a single evening of matches and bought again the next day. Low-level materials feed that demand indirectly, because the enchants and potions are made from them by somebody else. It is worth separating this spending from ordinary levelling traffic when reading listings: a levelling buyer stops bidding the moment the item is outgrown, while a capped PvP character keeps buying the same consumables for as long as the test lasts.
The useful part is that this habit does not end when the cap moves. A character built for a low bracket is not a character on its way somewhere: it is held at a fixed level on purpose so it can keep queueing against others in the same range, and it is kitted out as though it were finished. The shopping list is the best item the bracket allows rather than the cheapest one that works, a permanent enchant on every slot that accepts one, and a consumable bill that gets paid again after each session. At launch the ordinary levelling traffic through those same categories drains away as the population moves up, but this buyer stays in them.
That is worth keeping in mind before writing off a quiet low-level category as beta-only noise. Twinks are a small group next to the mass of levelling alts, so the volume they generate is nothing like what the test shows now, but they bid on a narrow, repeatable list and they are the ones willing to pay for a stat roll rather than a slot filler. Low-level materials, the recipes that turn those materials into permanent upgrades, and the handful of drops that are genuinely good for their level are the parts of this market with a buyer on the far side of the wipe. Everything else in the bracket is being bought by characters who will be deleted or outgrown.
One slice of that spending is easy to misread from the listings alone. A share of the low-level caster gear moving through the auction house is not bought by casters at all - it is a second set, intellect and +healing, picked up by hybrids who are levelling in a damage spec. A feral druid, an enhancement shaman or a retribution paladin quests in one set and swaps to the other the moment a group forms, because the test population runs dungeons constantly and the healer slot is the one that fills last. The buyer is not building a caster character; they are building a role they can switch into for an hour.
There is a second reason the same category can look busy in the morning and empty at night, and it has nothing to do with class mix. Players posting from Oceania report that they are testing on the same realm as everyone else, because the beta runs as a single large server with no separate Oceanic option. That is a player account rather than a published realm list, but it is consistent with what the auction house looks like: one market shared by regions whose play hours barely overlap.
Faction balance is not always left to the players either, and the tournament realm is the clearest case of that. New Horde character creation on the PvP tournament realm was closed off while the faction split there was skewed, so anyone arriving to test during that window had one side available to them regardless of what they intended to roll. We do not know how long the restriction stayed in place or on what basis it was lifted, and nothing here should be read as a rule that applies to the open beta realm.
The mechanism matters more than the single incident. A population shaped by what the client would let you create is not the same as a population shaped by preference, and a faction read taken from a realm under that kind of gate does not transfer anywhere else. It also moves the market in both directions at once: the closed side stops gaining new sellers and new levelling buyers, so its listings thin out and stay thin, while the open side absorbs everyone who showed up that day. If you are comparing auction house activity across realms, check first whether both factions were actually open to new characters on the realm you are reading.
The forum threads add one more observation that fits the same picture: the starting zones are crowded, and testers describe waiting on respawns and losing tags on quest mobs to other players working the same objective. That is anecdote rather than a count, but it is the kind of anecdote a census cannot give - it says where the population is physically standing, not just which classes it rolled. With most characters below the cap and a steady stream of new rerolls entering the same few zones every day, the congestion sits exactly where the sample says the players are.
Crowding pushes spending toward the auction house rather than away from it. A player who cannot get a tag stops gathering their own hides and herbs and buys them instead, skips the contested quest chain in favour of grinding something quieter, and leans harder on food, drink and bag space to keep the run moving. The result is heavier bidding in the same low-level categories that the level distribution already points at, and it is a demand that renews itself: every reroll walks back into the same zone and buys the same short list again. Treat the congestion reports as supporting evidence for the low-level bias, not as a separate market.
The practical effect is that the listing pool refills and drains on a regional clock rather than a flat one. European evening, American evening and Oceanic evening each push their own wave of posts into the same 24 hours, and the hours between them are when undercut chains stall and stock sits. If you check prices once a day at a fixed time, you are reading one of those waves and treating it as the whole market. Checking the same category at 2 widely separated hours before deciding what something is worth costs nothing and removes most of that error, and it matters more for low-volume goods than for the levelling staples that get posted around the clock.
This is why cloth, leather and mail with intellect, spirit or healing on them can clear at the lower brackets while the equivalent pure damage pieces sit unsold. It also means the demand does not scale with the number of priests and mages in the population - it scales with the number of hybrids who want to tank or heal on demand without rerolling. Druids are the obvious case given how often they appear in the census ordering, but the same pattern covers every class that can heal in an off-spec. When stocking for that buyer, treat the healing statline as the product and the armour type as a secondary filter, not the other way round.
For a sense of scale there is one adjacent dataset, and it has to be labelled before it is used: it does not describe Forever. The same style of community census covering Season of Discovery megarealms logged roughly 70,538 characters on the US RP-PVP ruleset, 47,442 on the US PVE ruleset and 80,277 on the EU PVP ruleset, with the Alliance share running anywhere from 51% to 77% depending on which ruleset is being counted. Different mode, different client, different launch - it is the nearest published figure for how large a modern fresh-start market gets, and nothing more than that.
2 things in it transfer as shape rather than as quantity. Ruleset choice splits a player base into separate markets of very different size, so the same item can have a deep buyer pool on one and almost none on another; and faction balance is not guaranteed to be even, which matters because each faction trades in its own auction house. Both apply to the Forever beta, which also runs separate rulesets. The figures themselves move as more clients report in and as players transfer or stop logging in, so treat them as an order of magnitude and not as a count of anyone currently playing.
The same gap gets filled informally in community threads, and it is worth seeing how. A figure of roughly 170,000 players on the beta circulates in discussion, usually with a follow-on claim that the launch population will be around 3 times that. Neither number has a stated source behind it. No concurrency counter is exposed on the test client, no realm roster is published, and an addon census counts characters its users walked past rather than people logged in - so there is no published route from any available dataset to a headcount of players.
Treat those estimates as what they are: an expectation about scale, not a measurement of it. They are still useful in one narrow way, because they tell you what other sellers believe when they decide how deep to stock a category. A market where most participants expect the buyer pool to multiply at launch will see materials hoarded rather than listed, and that shows up as thin supply long before anybody confirms a single population number. Plan around the shape of the sample data above, and treat any quoted player count - including these - as unverified.
The test shapes the population before the players do
The beta caps characters at level 30 and deletes them on 4 November. Both facts push the population in the same direction: toward short-lived characters made to try something, rather than toward mains built for a season. People roll the class they intend to play at launch, run it far enough to feel the early rotation, and then roll another. The result is a population that is heavier on low levels and alts than any launch server will be, and one where a large share of characters never reach the brackets where real spending starts.
Access is filtered before any of that starts. Guaranteed entry to the beta comes with the Epic pack and the tiers above it; everyone else who registered is drawn by random selection, so a share of the interested player base is simply not on the client. That is a paid filter sitting in front of the population, and it skews the test toward people who committed money ahead of launch rather than toward the mix that will log in on day 1.
It matters for demand in 2 ways. A buyer who already paid for a higher tier tends to treat the test as a rehearsal for a planned main, which concentrates purchases on the classes and professions they intend to keep, and it thins out the casual end of the market that normally supplies steady low-value bids on bags, food and first-tier materials. Neither effect survives the wipe: at launch there is no entry tier and no draw, the population widens to everyone who buys the game, and the categories that look quiet on the beta get their buyers back.
That ceiling has not been fixed for the whole test. The beta opened with a cap of 20, and the move to 30 is listed among the changes coming with the next build; no date for that build has been published, so the 20 bracket is what a tester sees until it lands. We are taking the raise from the announced change list rather than from anything measurable on the client, and the rest of this page is written against the 30 figure because that is the state the test will spend most of its remaining time in.
10 extra levels redraw the purchase skew described above rather than remove it. Under a cap of 20 the whole population sits in the bracket where quest rewards cover nearly every slot and the auction house is reduced to bags, food and first-tier materials. At 30 a character has outgrown several starting pieces, carries more than the default bags hold, and reaches the point where a crafted weapon or a stat green is worth buying instead of waiting for a drop. Demand moves up one tier and widens, but it still stops well short of the brackets that matter at launch: the first mount's training cost, raid consumables and the late crafted sets all sit above 30 and stay invisible on the test client either way.
At the ceiling the behaviour changes shape again. A character parked at 30 with nothing left to level into turns to the group content inside the bracket and starts assembling a phase best-in-slot set out of 5-player runs, repeating the same instances for the pieces that beat quest rewards. That is what builds the purchase structure on the test client: gear comes from drops rather than from listings, so the auction house is asked for the things a run consumes instead of the things a run replaces. Food, bandage materials, bag space and the occasional crafted filler for a slot no instance covers are what a capped tester actually bids on.
Part of the player discussion around the test is a proposal to stage the level ceiling after launch rather than open it at once: 20 at the start, then 30, 40, 50 and 60 released a week at a time. The argument made for it is economic as much as it is social. A ceiling held in place for a week keeps the whole population inside one bracket, so the items that bracket uses keep selling for as long as the cap holds, instead of being outlevelled within a day by the people who push hardest.
That is the same mechanic the beta is already showing at a cap of 30, which is why the proposal is worth noting on this page rather than filing under launch speculation. Low-level crafted gear, the first-tier materials behind it and the consumables a 5-player run burns through all have a buyer pool on the test client for one reason: nobody can leave the bracket. Remove the ceiling and that pool drains upward in days. Whether the staged version is adopted is not something we know, and no schedule for it has been published - but anyone reading beta listings should treat the long life of low-bracket demand as a product of the cap, not as a property of the items themselves.
The level ceiling is not the only filter on who is standing in these markets. Access to the test is paid: testers report paying 60 dollars for entry and finding no cheaper tier, with parts of the content sitting behind a further separate purchase on top of that. We have not verified the exact tier structure ourselves, and the figure above is what players describe rather than a published price list, so treat it as the shape of the barrier and not as a quote.
What matters for demand is the shape, not the amount. A paywall of that size does not thin the population evenly - it removes the casual drop-in tester first and leaves behind people who have already committed money to playing at launch. That group levels with a plan, rolls the class it intends to keep, and spends on the things that carry forward into its own preparation rather than on whatever is convenient. It also skews toward players willing to buy convenience outright, which lifts the floor under bags, food and crafted fillers and leaves the cheap impulse categories quieter than a free test would. Read the class mix on the beta as the mix of people who paid to be there, and keep that in mind before projecting any of it onto an open launch where the entry cost is gone.
Class quest rewards cut into that bidding wherever they land in the bracket, and the clearest case on the test client is the shaman chain that opens at 30. It ends in a choice of weapon - Lightning's Grasp, Tidebringer's Claw or Rage of the Storm - so a shaman who finishes it walks away with a main hand that no listing has to cover. The change list for the upcoming build also shortens the chain, trimming the travel and the collection steps that made it a multi-session job in the original version. We have not run the revised version ourselves, and the published notes do not say which steps were removed, so treat the shortening as announced rather than measured.
The market effect is narrow but worth naming. Every shaman who completes it is one fewer buyer in the weapon category for that slot, and on a test where a large share of the population is parked at the ceiling with the same chain available, the thinning shows up fast. A shorter chain only makes it thin sooner, because the reward stops being something a tester postpones. Weapons for the classes with no equivalent reward keep their buyers, and the consumables a capped character burns through are unaffected either way - that is the part of the demand picture a strong quest weapon does not touch.
The consequence for anyone reading the market is that weapon and armour tabs look quieter than the population would suggest, and the consumable and container tabs look busier. Neither is a verdict on launch demand - it is what happens when the only progression left is a loot table. Which instances sit under 30, and how many of them, has not been published for the beta, so treat the pattern as the useful part and the specific dungeon list as unknown.
That skews demand in a specific way. Purchases tied to the levelling middle and late game barely appear, because barely anybody is there. Nobody on the beta is saving toward training the first riding tier in earnest, nobody is stocking raid consumables, and the crafted gear that normally sells to characters in their forties has no buyer. What is left is the bottom of every category: starter weapons, bag space, first-tier profession materials, and food.
Class choice on a test client is also a declaration rather than a commitment. Someone trying a class for 2 hours to decide whether to main it buys nothing beyond what the trial needs, while the same player at launch would be a customer for 60 levels. So the beta tells you which classes people are curious about, which is useful, and how much those classes will spend, which is not.
There is a second source of the same distortion, and it is easy to miss because it looks like population. When a large stream brings a wave of viewers onto the test client at once, most of those characters are made in the same hour, run through the opening quests, and then stopped. They are never deleted, they never log back in, and they stay on the realm as a block of low-level characters that any count - an addon census, a chat impression, a crowded starting zone - reads as part of the faction. They buy nothing, list nothing, and gather nothing beyond whatever the first 2 or 3 quests hand them.
The practical effect is that the visible population and the spending population are not the same group, and the gap is widest at exactly the levels where the beta already has the most characters. A starting zone that looks busy may be busy with characters that were abandoned days ago, so crowding there is a poor proxy for competition on nodes or for buyers in the auction house. The check is movement rather than headcount: watch whether the characters around you are questing, gathering and repairing, and weight your read toward the ones that are. Whatever those accounts did earn is gone with everything else at the wipe, which is the same point as gold earned before the wipe.
One structural blank: nothing has been published about whether the Skyborne race is playable on the beta, and the 4 launch zones and 9 dungeons are content the test population may or may not be spread across. Until that is clear, any race-level read of the beta is describing a subset of the launch roster, not the roster.
Turning a class list into a shopping list
Composition converts to demand through 4 channels, and they behave differently from one another. Armour type decides which crafted and looted pieces have a buyer at all, so a population weighted toward one armour class leaves the other categories sitting. Weapon type does the same to the weapon tabs and to the vendors that feed them. Consumables cut across every class, which is why food and basic potions are the broadest demand in the game. Bags are demanded by literally everyone and are the one category where class composition does not matter at all.
The cross-class categories are the ones worth paying attention to on a test client, because they are the only demand that is not distorted by the level cap. Cooked food in Forever carries stats and a small experience bonus, which makes it useful to a character at any level and to every class, and it is made from materials a beta population is already walking over. Cloth behaves similarly from the other side: it drops for everyone, it is needed by the craft that makes bags and by first aid, and it is the category most likely to have both real supply and real demand inside a level 30 world.
Class-specific demand is narrower and more volatile. Reagent-style consumption, ammunition for the classes that need it, and the odd stat-specific green all depend on how many people rolled the class that wants them. In a population of test characters that number swings week to week, so a category can be starved on Monday and flooded on Friday without anything in the economy having changed. When you are judging whether a good is worth making, which trade actually pays is a separate question from which class buys the output.
The supply side of the same picture is covered by the gathering professions and supply: the free gathering quest around level 3 or 4 means most of the population arrives at the auction house as a seller of raw materials before it arrives as a buyer of anything. A market where nearly everyone gathers and almost nobody crafts is exactly the shape the beta is in, and it is why raw material listings outnumber finished goods so heavily there.
What survives the wipe, and how to check the rest yourself
Treat the beta as a source of behavioural information, not price information. Which classes people are trying, which zones they crowd into, which auction house tabs stay empty because nobody in the world needs that item type - all of that carries forward as a pattern. The gold itself does not: everything earned on the test client goes away with the character, which is the whole argument in why beta gold disappears.
A half-hour of deliberate looking produces more than any census would. Stand in a capital at a busy hour and note what people are actually playing rather than what the forums say they are. Walk the early gathering routes and count how many other gatherers you meet, which is a population read and a competition read at once - other gatherers in your copy change your yield more than node density does. Then open each auction house category in turn and look for the tabs that are thin on listings, because a thin tab with buyers is the only genuinely useful thing a test economy can tell you.
2 caveats before you act on any of it. The beta's 3 rulesets are separate populations with separate markets, so a category that is starved on one is not necessarily starved on another, and a read taken in one does not transfer. And the level cap means the whole upper half of normal demand is invisible; whatever you conclude about consumables, crafted gear or the late brackets is extrapolation, not observation.
What we cannot tell you: how many players are on the beta, how they split by class, race, faction or ruleset, and whether the mix resembles what turns up at launch. Those figures have not been published by anyone with access to them, and we are not going to invent them to fill a section. Everything above is a method for reading the population in front of you, and the arithmetic that follows from it sits with the rest of the Forever economy and its markets.
Questions
Is there an official class or race census for the Forever beta?
No. Blizzard has not published population figures for the test client, and there is no public armory for it. Any share or percentage you find is someone's sample of a chat channel, a poll or a guide's own readership, and should be read that way.
Does the beta class mix predict what launch will look like?
Partly. It shows which classes people are curious enough to try, which is real information, but it does not show how much those characters will spend, because the level cap and the wipe stop anyone from playing a full economic life. It also excludes any content and any race that only arrives at launch.
Which auction house categories are least affected by who is playing?
The cross-class ones. Bags are wanted by every character regardless of class, cooked food is useful to everyone and carries an experience bonus on top of stats, and cloth feeds both bag-making and first aid. Those categories hold their demand even when the population mix swings from week to week.
Can I use beta demand to decide what to stockpile for launch?
You can use it to decide what to learn, not what to hold. Nothing carries over from the test client, so stock is pointless, but knowing which categories stay thin and which routes are crowded is knowledge you keep. Treat the beta as a rehearsal of behaviour rather than of prices.
Sources
- Free gathering profession quest, camping objects from skill 20, the recipe additions and cooked food carrying stats plus an experience bonus — Wowhead's professions breakdown.
- Launch zones, dungeons, the Skyborne race and the content schedule the beta is testing against — Wowhead roadmap for Forever.