A TikTok affiliate tier built on follower count is the first thing a commission cut destroys.
Most programmes get their first structure the same way: sort by followers, draw two or three lines, call the top slice tier one. It survives while commission is generous, because a generous rate makes almost any creator worth a sample. Then a margin decision lands — 20% down to 12% is the common shape — and inside three weeks acceptance has roughly halved. The creators who said yes at 20% are now comparing you against a brand offering 15%.
This is for sellers running 40 to 120 creators on TikTok Shop, one or two stores, usually beauty, apparel or supplements, with commission already moved at least once. The problem is not finding new creators. It is that the old ranking no longer tells you who to keep.
Why follower-count tiers collapse, and what replaces them
Follower count measures reach, and reach is only half of what commission buys. The other half is whether the audience converts. A 200,000-follower account that watches but does not buy is expensive at 12%; a 14,000-follower account selling three units per video is cheap at 12%.
At a high rate you never have to separate those two things. Everything is worth trying, so the ranking can stay crude. The moment the rate drops, every creator’s cost per order rises at once, and the ones whose reach never converted turn negative first. Tier one becomes the group you most need to trim — the opposite of what a tier structure is for.
The second failure is quieter. Follower count barely moves, so a follower tier looks stable — but that stillness is the absence of a signal, not evidence of one. A ranking that cannot report a change reads as good news.
Three figures do the job followers were doing badly.
GPM, gross revenue per thousand views. The most useful number here because it normalises for reach. A creator with 30,000 views and 300 in sales and one with 300,000 views and 3,000 in sales score identically. Rank on GPM and the expensive creators separate from the cheap ones immediately, at whatever rate applies.
Post-to-order rate. How many of her videos produce at least one order. A creator who posts four videos and produces orders on three is a different asset from one who posts ten and produces on one, even at identical total sales. The second is a lottery ticket being bought repeatedly.
Reply latency. Median hours to answer after a sample lands. Not a performance metric — a reliability one, and at a lower rate it matters more rather than less. Cheap creators who answer in a day keep campaigns on schedule. Cheap creators who answer in a week move your launch date without telling you.
Follower count stays useful as a filter; it just stops deciding who gets a sample.
A tier table that holds when commission moves
The table below is a starting structure, not a law. Each tier is defined by an action, so a rate change alters thresholds without invalidating categories.
| Tier | Entry condition | At 20% commission | At 12% commission |
|---|---|---|---|
| Core | GPM above catalogue median, 2+ posts per month, replies within 48h | Priority samples, first refusal on new SKUs | Unchanged. These stay profitable at any rate. |
| Proven | GPM above median but posts irregularly | Standard samples, quarterly check-in | Give a fixed monthly slot. Irregularity is the risk, not the output. |
| Trial | Under three posts, no GPM history | Sample one SKU, review after 30 days | Sample the cheapest SKU only. A wrong sample costs double here. |
| Parked | GPM below median after three posts, or no reply past 7 days | Low priority, still on list | Off the sample list. Keep the record, stop the spend. |
Two things make this survive a cut. Entry conditions are measured per post, not per account, so a smaller creator can enter Core and a large one can be Parked. And the tier names describe a relationship, so moving down is an operational decision rather than a judgement about the person.
DAMI’s creator filter covers the inputs this table needs — follower brackets, category, and the sales detail that produces GPM — so the sorting happens against real numbers instead of an export you rebuild by hand each month.
Migrating an existing list without losing your top ten
Do not re-rank everything on day one. A full rebuild of a 60-creator list produces an order you cannot yet trust, and usually drops two or three good creators because their history is missing.
Run the new numbers in parallel for a month. Keep the old tier in one column and the new one in another. Where they agree, nothing changes. Where they disagree, that is your review list, and it is usually shorter than it looks — disagreements cluster in the top decile and the bottom third.
Watch one pattern: creators the new ranking promotes from Parked to Core. Small reach with high conversion is exactly the profile a rate cut makes valuable, and losing them in a bulk rebuild is the most expensive mistake available here.
There is a limit worth stating. This table assumes at least three posts per creator. Under that, GPM is noise and post-to-order is a coin flip. New creators cannot be ranked yet, which is what the trial tier exists for. Reviewing a creator on GPM rather than followers walks through reading that first result without overreacting to it.
When this breakdown misleads you
Three cases where the table gets the wrong answer.
Launch periods. A new SKU with no history produces GPM figures unusable for the first two weeks. Ranking during a launch compares creators against a baseline that does not exist.
Brand-awareness campaigns. If the goal is reach rather than orders, GPM is the wrong axis and the ordering should invert. A brand paying for impressions should rank on views and audience fit, and accept that the top of that list looks expensive on every other metric.
Creators whose audience crosses into a second category. Someone whose audience buys beauty but who posts home goods will score badly on the product she was given and well on the product she was not. The tier reflects the assignment as much as the creator, and misreading that as creator quality loses good people.
DAMI does not decide which of these cases applies to you. The tier structure is a record of decisions, and the decisions still come from whoever owns the margin. What changes at a lower rate is that the cost of a wrong assignment goes up, which is why the assignment — not just the ranking — is where the outreach process has to become more deliberate.
If the list is long enough that re-ranking by hand each month is the bottleneck, the DAMI creator workspace keeps the tier, the GPM history and the sample state attached to one creator record instead of reconciled across three exports.
FAQ
Should I drop creators who fall below the GPM median?
Not on one month. GPM is volatile on small samples, and one weak video can drag a good creator under the line. Two consecutive months below median with three or more posts is defensible; one is not.
What commission should tier one get?
Whatever keeps cost per order inside your margin on that creator’s actual GPM — not a percentage copied from a competitor. Two creators at the same rate can have very different cost per order, and that gap is what to negotiate against.
Do I need a different table for each market?
The categories travel; the thresholds do not. GPM medians shift by market and category, so recalculate entry conditions per market while keeping the tier names and actions the same.
