Guides · Free tool

Keeper ratio: three numbers tell you if your cull over-selects

The clearest published reference we know: in a December 2024 wedding review (mariess.co.uk), the photographer's real pick was 580 keepers out of 4,800 frames — about 12%. The automated pass they tested kept 1,500 of the same frames — nearly three times their actual selection.

Your last gallery has the same three numbers. Nothing you type leaves this page.

Frames shot Frames kept after the cull Frames you actually plan to deliver

Editing attention per kept frame: 30 seconds Your pace, not ours — slide it to what a frame really costs you across editing, retouching, second looks.

Your keeper ratio

Over-selection factor

What the extra frames cost

BestTake culls recall-first: tight picks that still catch the moment. Bring your own photos to see it on a real set — run the in-browser Culling Audit → or see how BestTake works →

Method

Keeper ratio = frames kept ÷ frames shot. Over-selection factor = frames kept ÷ frames you plan to deliver. The hours figure is plain multiplication: extra frames beyond your plan × the per-frame pace you set on the slider. No hidden weighting.

Reference points come from one published December 2024 review (mariess.co.uk): a real pick of 580 out of 4,800 frames (~12%), an automated pass keeping 1,500 of the same 4,800 (~31%), and the reviewer's stated working range of 500-1,000 delivered images from 3,000-5,000 shot (~10-33%). One documented wedding and one photographer's published practice — a reference, not a statistic.

Nothing you type is sent or stored; only anonymous usage stats (page viewed, calculation shown) are measured.

Further reading

• The Culling Audit — drop real photos, see burst groups and soft frames (nothing uploads)
• Why do AI culling tools over-select wedding photos?
• How long should wedding photo culling actually take?
• All BestTake guides

Educational tool: the ranges quoted are dated, published references, not delivery standards — your genre, market and clients set your own. Content produced with AI assistance and reviewed by the publisher before release.