How the audience quality score is calculated
Last updated
Glimpsr distills an Instagram audience into a single 0–100 score so you can track its health over time. The method is open — here is exactly how it works, so you can judge (or reproduce) it yourself.
score = round(100 × (0.45 × followBackRate + 0.35 × retention + 0.20 × balance))
Each factor is a value between 0 and 1, weighted by how strongly it signals a genuine audience. All three are computed only from your own follower and following lists — no engagement, demographic, or private data is used or required.
Follow-back rate (45%)
followbackrate = mutuals ÷ max(following, 1)
The strongest signal of an intentional, reciprocated audience. If the accounts you choose to follow tend to follow you back, your following is deliberate rather than a one-way spray of follows.
Retention (35%)
retention = 1 − (followers lost since last snapshot ÷ max(previous followers, 1))
Measures how well you keep the audience you earn. Heavy churn — people following then quickly leaving — usually points to follow-for-follow behaviour or content that doesn't match expectations. With only one snapshot this defaults to an estimated 0.85 and is flagged as estimated.
Balance (20%)
balance = min(followers ÷ following, following ÷ followers), clamped to 0–1
Rewards a healthy ratio between followers and following. Following far more accounts than follow you back is a classic low-quality pattern; a one-sided account (no followers or no following) scores 0 on this factor.
Limitations
- The score reflects audience structure (reciprocity, retention, balance), not content quality or reach. A great creator can have a modest score, and vice-versa.
- Retention needs at least two snapshots to be exact; with one it is an estimate.
- It does not detect bots directly — it infers audience health from ratios, so a padded account tends to score lower without naming individual ghost followers.
See the glossary for term definitions, or run your own audit.