How we measure follower retention

The single number

Retention is reported as one figure: the percentage of followers from a delivered batch still attached to the target account thirty days after delivery completed.

30-day retention (%) = (followers remaining at 30 days / followers delivered) × 100

We deliberately do not report 24-hour retention as the headline. Every panel passes 24-hour retention; the meaningful test is whether followers are still there a month later.

How cohorts are formed

Each test order from the delivery-speed protocol doubles as a retention cohort once it completes. The cohort identifier ties together:

• The exact follower-IDs delivered to the target account (recorded at completion).
• The completion timestamp — used as t = 0 for every subsequent retention check.
• The target account & service tier — so retention can be sliced by platform and size band, not just reported as one site-wide number.

Sampling cadence

We re-check each cohort at four checkpoints. Every checkpoint is captured because the shape of the drop curve matters as much as the 30-day endpoint — a cohort that drops half its followers in the first week is qualitatively different from one that holds steady and drops 5% slowly.

Refill trigger

A refill is automatically queued when delivered count for an order falls below 95% of the originally-delivered figure. The 5% threshold is wide enough that ordinary platform churn (deactivations, suspensions) does not constantly re-fire the pipeline, but tight enough that real attrition is detected and corrected within a single audit cycle.

The audit cycle runs every 24 hours for the full first year of an order's life — the refill window. Within that window there is no cap on refill count: any drop below the 95% threshold is corrected until the 1-year window closes.

Industry context

Most SMM panels offer a 30- to 60-day refill window. Likes.io's refill window is 1 year from delivery: within it, any drop below the 95% threshold is refilled at no cost; beyond it, the order is considered settled. We can hold to a full year because our 90-day cohort data shows the overwhelming majority of genuine attrition surfaces inside the first month — so a 1-year window costs little beyond month one to honor, while covering the rare late drops that shorter guarantees quietly exclude.

This is also why we can claim a window this long: extended guarantees only work when the underlying retention is already high. A panel offering a 2-year refill on followers with a 60% 30-day retention is signing up to ship the same followers four times — a promise that quietly becomes hollow. On cohorts that hold in the 90s, the same promise is cheap to keep.

Out of scope

• Retention on accounts the customer subsequently grows organically (a +10% organic gain followed by a -2% drop is not a refill event — only the originally-delivered cohort is tracked).
• Drops caused by the target account being suspended or shadowbanned by the platform — by the time the platform has acted, the audit can no longer reach the public follower count.
• Like and view services — both are non-removable artefacts on the platform side, so retention is structurally 100% and not a meaningful metric.