Migrating from a hand-curated influencer roster to cluster discovery

Most brands and agencies operate a hand-curated creator roster they spent years building. The instinct, when moving to a cluster-based platform, is to either (a) keep the roster and bolt the new platform on top, or (b) throw the roster away and start fresh. Both are wrong.

Here's the working migration playbook.

Step 1: Import the roster, don't replace it

Bring the existing roster in via CSV or API. The roster carries genuine institutional knowledge — past relationships, payment history, who delivered on briefs and who didn't. None of that is in the cluster yet, and you'd lose it by starting fresh.

Step 2: Let the cluster engine score the roster

Once the roster is imported, the cluster engine computes per-creator audience-signal scores against your priority verticals. You now have two data points per creator: your historical experience plus the platform's audience-fit score.

Step 3: Surface the agreement and the disagreement

Three buckets emerge:

• Both agree — creators you rated high AND the cluster scores high. These move straight into auto-invite for the relevant briefs.
• Roster high, cluster low — creators you've worked with successfully but whose current audience signal isn't matching your buyer profile. Either the audience has moved, or the creator pivoted topic. Investigate before next brief.
• Roster missing, cluster high — creators the cluster surfaces who aren't in your roster. The new top-of-funnel; brief them.

Step 4: Update the roster on every campaign

This is the compounding part. Each campaign's attribution data feeds back into the cluster. Creators move between buckets quarter by quarter. Your roster, once a static list, becomes a dynamic ranked surface.

Step 5: Retire the spreadsheet

Six months in, the cluster engine + the imported-roster history is doing the job the spreadsheet used to do, with better data. The roster spreadsheet retires; the cluster inherits its institutional memory plus everything the spreadsheet couldn't capture (audience signal, attribution, cohort behavior).

What to watch out for

• Roster bias — if your historical roster over-indexes on one creator type or region, your cluster scoring will appear biased against the rest. Look at coverage, not just agreement.
• Stale audience data — older roster entries may need a refresh before clustering. Run the scrape first; cluster second.
• Compensation history — bring rate-card and payment-terms data over; otherwise every brief becomes a renegotiation.

The cost of not migrating

Roster-only operating means every quarter looks like the last one. Cluster-driven operating compounds: the same creator-program team finds better creators in less time each quarter, and the cohort behavior they're delivering gets sharper. The cost of not migrating is opportunity cost, paid quietly, for several quarters.