Creator Cohorts: Building tiers and setting rules
Cohorts are how you organise the creators you work with. A cohort is a named group, scoped to your brand, that you can filter by, invite from, target with a sequence and use in rules. Most brands use them as performance tiers, from a top tier of best sellers down to a dormant tier.
You can assign creators by hand, in bulk, or by setting a rule that Unbox applies for you each month.
This article covers building your cohorts, getting creators into them, writing rules, and using the result to read the health of your creator community.
Two things to know first
A creator belongs to exactly one cohort at a time. Cohorts are tiers, not labels you stack. Moving a creator into a new cohort takes them out of their previous one. This is what makes cohort sizes add up to your whole community.
Cohorts are scoped to a brand. The cohorts you build for one merchant account are separate from any other, so different brands can run different schemes.
Building your cohort library
The Cohorts page in the sidebar is where you create and manage them.
Select Create cohort, give it a name, pick a colour and add a description. The colour is the colour of the cohort pill wherever that creator appears in the platform, and the description is a good place to record the rule in plain English for the rest of your team. Names are unique within a brand and are not case sensitive, so "VIP" and "vip" count as the same cohort.
From the same screen you can rename, recolour and delete. Deleting a cohort does not delete the creators in it, it leaves them unassigned.
Build the whole set of tiers before you assign anyone. Cohorts are most useful when they cover everyone, including a dormant tier for creators with no recent activity.
# Getting creators into a cohort
One at a time. In Creator Explorer, click the cohort cell on a creator's row and pick a cohort. It applies immediately.
In bulk. In Creator Explorer, select creators with the checkboxes, then use Assign cohort in the action bar.
By CSV import. Use this when your tiers already live in a spreadsheet. On the Cohorts page, select Import and download the template. The file needs exactly two columns with the header creator_handle | cohort. Upload it and Unbox shows a preview of what will change, including any rows it could not match, before you commit.
Two things trip people up on import. The cohort names in your file must already exist in your library, so create a new tier on the Cohorts page before importing creators into it. The @ on a handle is optional, with or without works.
Setting a rule on a cohort
A rule lets Unbox decide who belongs in a cohort instead of you assigning creators yourself.
Open the cohort, select Add rule, then choose a variable, an operator and a value. Add further conditions if the tier needs more than one.
The variables available are:
- GMV, the revenue the creator generated for your brand
- Unit sales, the number of units they sold
- Product IDs (PIDs) sold, the distinct product IDs they sold, or a specific product ID
All of them are measured over the last 30 days. The window is fixed rather than something you set per rule, so every tier in your scheme is judged on the same period and the tiers stay comparable to each other.
A rule for a top tier might read "GMV is at least 2000 and SKUs sold is at least 3". A rule for a dormant tier is simply "GMV is 0".
When rules run
On save. Saving a rule classifies your creators against it straight away, so you can check the result immediately rather than waiting a month to find out whether the threshold was sensible.
On the last day of each month. All of your cohort rules run together, looking back over the previous 30 days, and creators move between tiers based on that month.
Edits apply at the next run. Changing an existing rule does not reclassify anyone on the spot. Unbox shows you the date it will take effect when you save.
Between runs, membership is stable. That is deliberate, so your team can plan a month of outreach against a fixed set of tiers. You can still move an individual creator by hand at any time, and that assignment holds until the next monthly run reclassifies them.
Designing a scheme that does not overlap
Because a creator can only be in one cohort, your rules must not compete for the same creator. Unbox flags rules that overlap when you save.
The reliable pattern is a set of adjacent bands with no gaps and no shared ground. A common shape looks like this:
- VIP - GMV of £2,000 or more
- Top Creator - GMV of £1,000 and £1,999.99
- Active - GMV between £250 and £999.99
- Starting - GMV between £1 and £249.99
- Dormant - no GMV
Three habits keep a scheme clean.
Close your bands at both ends. "£250 to £999.99" behaves predictably. "More than £250" competes with every tier above it.
Put the more demanding rules at the top. Tiers that require both GMV and a spread of products should sit above the tier that only asks for GMV, so the stricter rule wins the creator.
Always include a zero tier. Without a dormant cohort, creators with no recent activity fall out of the scheme entirely and disappear from your view of the community.
There is no single correct scheme. A brand selling one hero product will band on GMV alone. A brand that wants creators selling across its range will add a product condition to the upper tiers. The thresholds should reflect what good looks like for you, not an industry benchmark.
Using cohorts elsewhere
Once your tiers exist they drive work across the platform.
Creator Explorer. Filter to include or exclude cohorts. Leaving the Include filter empty applies no cohort filter at all, it does not mean creators without a cohort.
Collaboration invites. Filter the candidate list by cohort when building an invite, so a campaign reaches the tier you intended.
Sequences. Link an outreach sequence to one or more cohorts. Steps can differ by cohort, and a cohort can be excluded from the sequence entirely.
Blocklist rules. Use cohort membership as a condition alongside the other rule fields.
Reading the health of your creator community
Your cohorts cover everyone and are recalculated on the same rules each month, so the size of each tier, and the way those sizes move between runs, is a straightforward read on how the programme is doing.
Watch your top tier for thinning. A shrinking top tier alongside a growing tier directly beneath it usually means your best creators dropped a band rather than left. That is a retention conversation, and it is more urgent than recruiting replacements.
Watch for flow upward. A healthy programme moves creators up over time. If your entry tiers stay full month after month but nobody graduates, sampling is producing content rather than building partners.
Watch dormancy against acquisition. If the dormant tier grows faster than your active tiers, you are recruiting over a leak. Reactivating a creator who has sold for you before is usually cheaper than finding a new one.
Because every run uses the same rules, the movement you see between months is a like-for-like comparison rather than a change in how you were counting. If you change a threshold, expect a step change at the next run and treat that month as a reset rather than a trend.
As a rough guide to what to do with each tier: protect the top, by prioritising sample allocation and keeping the relationship warm; grow the middle, with sequences and a wider set of products; test the entry tiers, giving the ones with signal a second collaboration; and reactivate selectively from dormant, starting with creators who performed before.
Common questions
Why is a creator in the wrong cohort?
If the cohort has a rule, the creator was classified on their last calendar months performance at the most recent run, which may not match your impression of them today. Move them by hand if you need to, and they will be reclassified at the next run.
I saved a rule and nothing changed.
Rules classify on save, so if you saw no movement, either no creators meet the threshold over the last 30 days or the ones you expected match a higher tier that takes precedence.
I edited a rule and the cohort looks the same.
Edits take effect at the next monthly run. The date is shown when you save.
Why did my CSV import skip rows?
Either the handle could not be matched, or the cohort named in the row does not exist in your library yet. Both are listed in the preview before you save. Check the header is exactly creator_handle | cohort.
Unbox says my rules overlap. What do I fix?
Two rules would claim the same creator. Usually the fix is to close the open end of the lower band, or to move the more demanding rule above the simpler one.
I deleted a cohort by mistake.
The creators are unassigned, not deleted. Recreate the cohort and reassign or reimport. If the cohort had a rule, recreating it with the same rule will repopulate it at the next monthly run.
Can a creator be in two cohorts?
No. One cohort per creator per brand. If you need overlapping groupings, use filters in Creator Explorer to cut the list a different way rather than trying to model it with cohorts.