Detection: Single creative driving most conversions

Key: creative_concentration_risk Severity: High Confidence: 80–90%

What this detection looks for

We flag a campaign as creative-concentration-risk when all of these are true:

1. The campaign is active
2. It has at least 2 active ads with conversions in the audit date range
3. It has at least 20 total purchase conversions in the date range
4. One ad is responsible for ≥ 70% of the campaign's purchases

Why this matters

When 70% or more of a campaign's conversions come from a single creative, the campaign's performance is effectively the performance of that one ad. Three things will happen, in roughly this order:

• The audience that responds to the creative will saturate. Click-through rate drops first, then conversion rate, then ROAS.
• The creative will fatigue — frequency rises, novelty decays, performance degrades.
• Meta's delivery system will continue to push impressions toward the historical winner even after it begins to underperform, which extends the decline (see also: false_winner_meta_bias).

A campaign carrying a 70%+ dependency on a single creative is one creative refresh cycle away from collapse. The fix is to build creative depth before the decline begins, not after.

How we calculate confidence

How we calculate the estimated monthly cost

The dollar figure on this finding is an exposure, not waste happening today. We estimate the at-risk spend as 50% of the spend served through the concentrated portion of the campaign, projected to a 30-day month.

monthly_at_risk = (concentration_ratio × 0.5 × campaign_spend_in_range) × (30 / days_in_range)

In plain English: half of the dollars flowing through the dominant creative become inefficient when it fatigues. That is the dollar value we surface.

What would change our mind

This finding can be a false positive in a small number of cases:

• The top ad is brand new and the campaign just launched. A newly launched campaign will have an apparent concentration that is just a function of which ad was approved first. We require at least 20 purchases in the date range to reduce this risk, but it can still happen for a campaign that started in the last 7 days.
• The winning creative is evergreen and the brand is built around it. Some brands intentionally run a hero creative for months. If you have refreshed cuts, hooks, or formats of the same hero and they are tracked as separate ads, the concentration may be safe.
• You have a tested creative refresh ready. If you have already produced the next generation of creative and are ready to launch, this finding is informational rather than urgent.

How to fix it

1. Identify the top-performing ad and what makes it work (hook, offer, format, audience).
2. Produce 3–5 variations that test different hooks while preserving what works. Do not test radically new creative — test adjacent variants.
3. Launch the variations in the same ad set so Meta can compare them in the same auction context.
4. Once one or two variants reach within 70% of the original's performance, begin redistributing budget. Do not pause the top performer until at least one replacement has matched its conversion rate over 7+ days.
5. Set a calendar reminder to refresh creative every 30–45 days even if performance is steady.

What we look at to make this detection

• effective_status on the campaign and on each ad
• Purchase conversions per ad, summed across the audit date range. We use offsite_conversion.fb_pixel_purchase and offsite_conversion.custom.<your-purchase-event>
• Ad creative_id and creative_type for evidence and methodology attribution

Source

This methodology page is generated from apps/api/app/services/detections/creative_concentration_risk.py. The detection code is open for inspection. We do not have hidden rules.