Photo keep rate, explained plainly

Keep rate is one of the few numbers in a photography business you can calculate exactly and still misread completely. It is delivered images divided by captured images, and nothing else. Here is what it measures, what moves it, why comparing your number to another photographer's tells you almost nothing, and how to record it so your own number starts earning its place.

Keep rate in one line

Keep rate is the share of the frames you captured that end up in the client's gallery.

Keep rate = delivered images divided by captured images, times 100.

That is the whole formula. The difficulty is never the division. It is deciding what goes on each side of it, and then deciding the same way next time.

Before you calculate anything, answer four questions once and write the answers down:

• Do captures include a second shooter's frames, or do you keep the two counts separate?
• Do captures include test frames, misfires and the shot of your own foot? They should, if you want the number to describe the real job.
• Do delivered images mean the finished gallery only, or does it include sneak peeks and anything you sent unedited?
• Are you counting files or moments? Two crops of one frame are two files.

Record the raw counts next to the percentage, always. Twenty percent of 4,000 frames and twenty percent of 900 frames are the same number describing two completely different days, and in six months the percentage on its own will tell you nothing about which was which.

The only published numbers worth anchoring to

There is no serious public dataset of photographers' keep rates. What does exist is a set of volume ranges for weddings, published by platforms rather than researchers, and worth treating as typical ranges rather than measurement: roughly 2,000 to 4,000 RAW frames across an eight to ten hour wedding day, and roughly 400 to 800 edited images delivered, which works out at about 40 to 60 delivered photos per hour of coverage. The most commonly cited version of those figures is self published platform data, so treat it as a sense of the shape of the market, not as a finding.

Run the four corners of those two ranges through the formula and you get the honest answer to "what is normal".

Arithmetic from the published volume ranges above, not a study of keep rates. Nobody has measured those.

So a wedding keep rate anywhere between 10 and 40 percent sits inside the commonly quoted ranges. A band that wide is not a benchmark. It is a warning that the number is driven by things other than quality, which is exactly what the rest of this article is about. If you want the wedding case worked forward, from a frame count on the card to a delivery target, we did that separately in the shot to delivered ratio for weddings. Everything else we can source on the wedding market lives in our 2026 wedding photography statistics.

Why your rate and another photographer's are not comparable

Two photographers can hit the same keep rate doing entirely different work, and the same photographer can post two very different rates in one week without getting better or worse at anything.

The reason is that a keep rate is the output of four separate decisions, and only one of them is about how well you saw the moment.

How you shoot. A documentary shooter holding short bursts through a processional generates a different denominator to someone timing single frames. Neither habit is wrong. They just do not divide the same way.

What conditions you worked in. A dark reception, an unfamiliar venue, a new body whose autofocus you do not yet trust, a fast child, a running dog. Technical failures are conditions as much as craft.

How strict your final standard is. An editorial photographer delivering a tight, curated set and a documentary photographer delivering the full thread of a day are answering different briefs.

What you promised. If your package says a range and you are near the bottom of it, that alone changes the numerator before you have judged a single frame.

None of those four is a talent measurement. A high rate does not prove every frame was intentional, and a low rate does not prove you were wasteful. It is also worth remembering that the rates photographers mention publicly are attached to their best work rather than their average Tuesday, and that a headshot morning and a ten hour wedding produce numbers that should never be put in the same column. Genre differences are large enough that we treat them separately in photos delivered per hour by genre.

The four dials that actually move the number

If you want to change your keep rate on purpose, these are the only places to reach.

Read across that table and one thing stands out. Two of the dials are set before the shoot starts, by your package and your genre, and a third is set by the weather, the venue and the couple's timeline. Only the first is fully yours on the day.

Which is why chasing a target rate is backwards. If you want the number to move honestly, change a capture habit or write a clearer standard, then let the rate follow. Writing that standard down is its own job, and we walked through it in building your own selection criteria.

Record it so the number means something

A keep rate on its own is trivia. A keep rate with its context attached is a record you can actually diagnose from a year later. Keep these fields in the same place you already track jobs, a spreadsheet is completely sufficient.

The candidate count in row two is the field most people skip and the one that pays. It splits the story into two halves: how many frames looked usable after a fast pass, and how many survived the standard you applied afterward. A drop in the first half is a shooting or conditions story. A drop in the second half is a judgment story. Without it you only have one number and two possible explanations.

Then leave it alone for a season. A single job's rate is noise. A run of similar jobs is a pattern, and the pattern is the only part that was ever going to be useful.

How to read a change in your own number

Because comparing across photographers is close to meaningless, the only real use of a keep rate is comparing you to you, on similar work. When the number moves, it is asking a question rather than answering one.

When the rate drops

First look at conditions and capture. More bursts than usual, more focus misses, a difficult venue, a second shooter still learning your coverage. That is a denominator story and it is often nobody's fault.

Then look at the other end. A rate can also fall because your final pass got severe, which tends to happen late at night when everything looks the same. If the gallery feels thin, or a client asks where their aunt is, you were not being disciplined, you were being tired. The fix is a coverage checklist before the last pass, not a rule about keeping more.

• Were there more bursts, focus misses or exposure failures than usual?
• Did the timeline create long repetitive stretches with nothing new in them?
• Did I make a clear first pass shortlist, or did I judge everything at once?
• Does the gallery still cover people, moments, details and transitions?

When the rate rises

A high rate is healthy on a varied day and suspicious on a repetitive one. Open the gallery in a grid and look for three frames of the same step of the processional, three crops of the same portrait, every reaction from one speech. Each file can be good while the gallery is still tiring to look through, which is the argument in why over delivering costs you.

A rise can also mean you shot too few frames to protect a moment. Excellent timing and no alternates look identical in the arithmetic right up until someone blinks.

And sometimes a rise is bookkeeping. If one job counted proofs as delivered files and the next did not, or a whole card never made it into the denominator, you did not change, your definition did.

What a picking tool changes, and what it does not

The common assumption is that AI selection raises your keep rate. It usually does not, and it is worth being clear about why. These tools are good at the mechanical part: focus, closed eyes, near identical frames from one burst. That work sits in the first pass, where you narrow captures down to candidates. Your delivered count is still set by your standard, your package and the story the day gave you, which is why the numerator barely moves.

What changes is the cost of getting there. That matters more than it sounds. A UK survey of more than 300 wedding photographers, run in 2020 before these tools were common and reported by PetaPixel, put 11 percent of working time on culling and 55 percent on editing, against 4 percent actually shooting. Read those two together and the incentive is obvious: every extra file you deliver lands in the biggest slice of your week. A keep rate paired with hours tells you which slice to attack, and the shape of that trade changes with job size, which we broke out in culling time by shoot size and in how long culling actually takes.

Kepla is being built for that first pass. It picks your best shots and hands you a shortlist to approve, and it never deletes, moves or renames a single file, so nothing it passed over is more than a tap away. The iPhone, iPad and Mac apps are still in development. The booking page is live and free today while we build.

Whatever you use, the way to make a keep rate earn its place is to change one thing and watch. Cap your static detail frames, or write a rule for how many alternates a burst is allowed, then run it across four or five similar jobs and compare captures, keep rate, selection time and how confident you felt about the gallery. Keep the change if the process got easier without the coverage getting thinner. That is the entire job of this number: not a score, a prompt to look at something you would otherwise have missed.

FAQ