Inconsistency between age buckets OVER_18 and OVER_21 in reach estimation

[udrzalv]

When I send request for reach estimation of males interested in Travel/Air Travel in the age bucket AGE_OVER_18, I receive count: { min: 106805, max: 160207 },

When I send the same request, but targeting AGE_OVER_21, the response is count: { min: 111726, max: 167589 }

How is it possible that there is more males over 21 years than males over 18 years?

[udrzalv]

Bump. Can I get response on this?

[JBabichJapan]

The reach estimate endpoint has some known issues around this sort of illogical response and we are tracking internally as a bug. Please note that the same behavior should be exhibited on Ads UI and it’s a problem related to how the data is estimated, the way it currently behaves the reach estimate itself should be considered an estimate especially for things like these interests (compared to for example tailored audiences which should be basically exact precision).

[Alula]

I’m still seeing similar discrepancies in age ranges. Is there a solution forthcoming?

In addition, could you provide some insights into how users are assigned to different age buckets? It would help our reporting to understand how/why these inconsistencies are occurring. Thanks.