Ironforge.pro's Classic trackers published a number this week that the Classic community has been arguing about ever since: characters raiding at least once on TBC Anniversary realms fell by roughly a third to two fifths in a single week — in the same week the WoW Forever beta opened, and with no announced date for the next phase of TBC content.

The numbers as they stand

Ironforge.pro counts characters that appear in raids uploaded to Warcraft Logs (plus PvP leaderboard characters). As of September 28, its pages report:

Community summaries of the same tracker give the previous week's TBC figure as 212,821 and put the Sep 15–21 week at 124,487 — a drop of 88,334 characters, or 41.5%. Depending on which week boundary and which realm filters you select, the same underlying tracker produces a fall of roughly 36% to 42%. That spread is itself the first lesson: this is a number you have to read with its window attached.

Why a 40% weekly drop is almost never 40% of players leaving

Ironforge.pro measures log-uploading activity, not people. Several things move that number at once, and none of them are visible in the headline percentage:

• Phase weeks inflate. A new raid tier pulls every guild into logs for one week, then normalises. A drop measured against a phase-open week is partly a return to baseline.
• Raid lockouts are weekly. A guild that raids Tuesday gets counted; the same guild skipping one reset disappears entirely from the metric.
• Uploads are voluntary. A guild that stops logging still plays. Public log uploads are a proxy for organised PvE, not for population.
• One server dominates. TBC Anniversary is a small, realm-concentrated population — Spineshatter alone accounts for a very large share of the total — so a shift in one realm's guild scene moves the global figure more than it should.

For a longer treatment of this exact trap — including a tracker that still listed a shut-down server at 61,000 concurrent players as its top result — see our private server population verification guide. The rules there apply unchanged to official Classic realms.

What is actually driving it

Three plausible causes sit on top of each other, and the sources agree on the first one:

1. No roadmap. The most-liked forum complaint of the week was not about the beta at all: "Guilds in both TBC anniversary and MoP Classic are disbanding because the assumption is these versions have reached their endpoint." Blizzard has not announced a date for Fury of the Sunwell, and MoP Classic has been sitting in its final patch since summer. When a version's next step is unknown, guilds stop scheduling raids.
2. The Forever beta as a distraction. The beta opened September 17 — the first day of the week in question — and the beta is explicitly pitched at Classic players. Continued demand problems and repeated player-cap increases make it the obvious place to spend a week's play time.
3. Ordinary late-expansion attrition. Anniversary TBC is deep in phase 3, with Zul'Aman and Sunwell still ahead and nothing to farm once Black Temple is on farm.

The counter-evidence: retail did not empty out

The "everyone left for Forever" reading has a problem, and it is measurable. Raider.IO's weekly figures, analysed by Icy Veins, show retail Mythic+ runs down 23.99% in week 5 — but that decline followed a dungeon-event week and a Timewalking week, and raid participation tracked its existing trend. Crucially, when Forever's beta opened 51 hours after the US reset, the hourly data shows no break at that point. US activity ended the week at −0.46%, EU at +0.10%, and China at +5.36%.

So the honest version of the story is not "the beta cannibalised WoW." It is narrower and more interesting: forever's beta is absorbing attention from the versions of WoW with no announced future, while retail — which has a live season, a roadmap and a patch schedule — held its shape.

What this means if you run or play on a private server

Official Classic numbers are the benchmark every private realm gets compared against, in marketing and in arguments. Two practical consequences:

• Do not quote these numbers as population. If you cite "TBC Anniversary dropped 88,000 players," you are quoting a change in log uploads. The same discipline we recommend for private-server trackers applies upward to Blizzard's own servers.
• Expect churn to keep moving downward and sideways. Private projects are already positioned to catch Classic-side drift — the Turtle WoW successor ecosystem, Emberveil's UE5 Vanilla realm, and Project Epoch's Reborn launch on October 23 all sit in the window where anniversary-server raiders are deciding whether to keep logging in.

How to read the next tracker drop

1. Check the window. Two days of drift between "Sep 15–21" and "Sep 16–22" moved the TBC total by more than 10,000 characters.
2. Check your filters. Region, realm type and faction toggles change the denominator dramatically. Ironforge shows the selected total at the top of the page; note it.
3. Compare like weeks. A phase-open week versus a lull week is not a trend, it is a calendar artefact.
4. Cross-check with a second proxy. Auction house volumes, PvP ladder activity, and per-guild upload counts move independently of the headline.
5. Wait for the next phase announcement. Phase 5 will re-inflate the same number that just collapsed, and the people now declaring the version dead will be quietly logging back in.

What to watch

• A Fury of the Sunwell date. It is the single piece of news that would reverse this trend, and it has been absent for months — the subject of our September 15 piece on the same wait.
• Whether Blizzard says anything about Classic's future at all. The absence of Classic content at BlizzCon 2026 is doing more damage than any single patch delay.
• The October attendance numbers. One week is noise; three weeks in the same direction is a trend, and the beta's next level-cap increase and name reservation window both land inside that window.