
Choosing which culling tools to shortlist is one job. Interrogating the one you are about to hand a season of client work is a different one, and it is the job people skip. These are twelve questions to ask the vendor and answer inside the trial, covering file support, migration, export, cost, support and the one nobody asks until it is too late: what happens to your ratings if you leave.
Working out which tools belong on your shortlist is a separate exercise, and we have already written it: five questions about when you actually do the work, whether you want a decision or faster evidence, whether your files can leave your machine, whether the pricing model fits your year, and what has to happen after the picks. That is how to choose culling software, and if you have not narrowed the field to two candidates yet, start there and come back.
This page assumes you have your two. What follows is the interrogation: the questions that decide whether you can live with a product through a full season, and walk away from it cleanly if you need to. A demo is designed to be admired. A busy August is not.
Ask these in writing, before the trial ends. How they answer is itself a data point. Specific answers with documentation links are a good sign. Cheerful answers that restate the marketing are not.
1. Which RAW formats and camera bodies are supported, and how fast is a new body added? Do not assume support because a similar model is on the list. If you are buying a new body next spring, ask what the usual lag is between a camera shipping and the software reading its files. Some vendors answer this precisely. That is worth knowing.
2. What happens with dual card, second shooter and mixed media jobs? Ask about duplicate handling across two cards, files from a second shooter's different camera system, HEIC, and any video or audio sitting in the same folder. A tool that grinds to a halt on a folder containing anything other than one camera's RAW files will cost you the time it saves.
3. What does the trial actually include, and which tier is it? This one catches people. Narrative's published pricing says every new account starts with a full free trial of Ultra, its top plan, which is generous and slightly dangerous: you can spend two weeks loving Scenes View and the Close-Ups panel, neither of which is on the $10 Lite tier. Aftershoot's pricing page offers 30 days with no card required. Imagen gives you two free culling projects. Whatever the offer, confirm the feature you are buying for is available on the tier you intend to pay for.
Test all three with your own difficult files, not the vendor's samples. A high-ISO reception, a backlit ceremony exit and one long burst will tell you more in ten minutes than a week of feature comparison.
4. Does it read the ratings, flags and labels already on my files? If you rate in camera, or if a job has been through a first pass already, you need to know whether the tool respects that work or ignores it. Ask specifically whether existing marks are read, preserved, or overwritten. Those are three different answers and only one of them is safe.
5. Does it create a second catalog, and does it touch my originals? The answer you want is that it reads a folder, writes markers, and changes nothing else. Ask where those markers are written: into a database inside the app, into XMP sidecar files next to the RAW files, or into your existing catalog. For RAW files, marks usually go into sidecars rather than modifying the original, which is why the sidecar question matters more than it sounds. Ask whether a rerun overwrites the marks you made by hand after the first pass.
6. Where are my files while this happens, and how long are they kept? Local processing keeps this simple. Cloud processing means an upload, a copy on someone else's disk, and a retention policy you should read before a client's wedding becomes the test case. Imagen's culling runs in a browser and the shoot is uploaded first. Aftershoot, Narrative Select and FilterPixel process on your own machine. If your contracts restrict where image files can be sent, that settles it before features do. The practical tradeoffs are in local versus cloud culling.
7. Which marker does it write, and can I choose? Flags, star ratings and color labels are not interchangeable. Plenty of photographers use flags for keepers and stars for client choices, and a tool that writes five stars onto everything it picked can quietly flatten a system you have used for years. Ask whether the mapping is configurable, then verify it on a trial job rather than trusting a screenshot.
8. Which editors does the handoff actually support? Ask for the list, not the logos on the homepage. If you edit in Capture One rather than Lightroom Classic, this question alone usually removes candidates. Then export a real job and open it in your editor before the trial expires. Our walkthrough of the Lightroom side is in getting selections into Lightroom cleanly.
9. What happens to the frames it did not pick? Some tools offer to move unpicked frames into a rejects folder, which is a setting you should find and think about before you point it at a wedding, not after. Others write a rating or a label and change nothing else. Neither is wrong, but you need to know which one you have bought, because a client asking about one specific frame six months later is a bad time to find out.
One related question is about judgment rather than plumbing: can you find a softer but emotionally right frame inside a burst, beside the sharp one the software preferred? If the interface hides what it did not choose, you are not auditing a shortlist, you are accepting one. The protocol is in how to test AI culling accuracy.
10. If I stop paying, what do I keep? The answer should be that your originals and every marker written to them stay usable without the software. Verify it rather than believe it: export a small trial project, quit the application, and open the folder in your normal editor. If your ratings and folder structure still make sense, you have an exit. If they only exist inside the vendor's database, you do not.
11. What does the vendor keep, or take away? This is where the fine print lives. Aftershoot's own pricing page states you can pause any subscription for three months, that the Select plan can be cancelled with no implications, and that cancelling Edit or Retouch deletes your custom AI profiles unless you pay $4 a month to store them. That is a perfectly reasonable policy, clearly published, and exactly the kind of clause you want to know about before you train a profile on two years of your own edits. For cloud tools, ask the same question about uploaded images: how long are they kept after cancellation, and can you require earlier deletion?
12. How do I get help on a deadline, and how do I get my money back? Ask what support looks like on a Saturday night when a gallery is due Monday, whether there is a refund window, and how breaking changes are announced. The trial is your support test: send one specific workflow question and judge the reply on precision and speed.
A perpetual license changes this whole section. Photo Mechanic's own pricing page lists $14.99 a month, $149 a year, or $299 once for a perpetual license, with Photo Mechanic Plus at $24.99, $249 or $399, a single user license covering up to two computers, and a 30-day trial. Buying outright means the exit question is largely answered on day one, which is the real argument in perpetual versus subscription.
Prices checked August 2026 from each vendor's own pricing page.
Paste this into the email. The right hand column is the part to watch.
| Ask | A good answer | A warning sign |
|---|---|---|
| 1. RAW formats and new bodies | A published list and a typical support lag | "We support all major cameras" |
| 2. Dual card, second shooter, mixed media | Specific behavior for duplicates and non-RAW files | No answer, or "put them in separate folders" |
| 3. What the trial includes | Named tier, named limits, no card | Trial of the top plan with no mention of tiers |
| 4. Existing ratings and flags | Read and preserved, with a documented rule | Silence, or "we start fresh" |
| 5. Catalogs, sidecars and originals | Reads a folder, writes markers, originals untouched | Vagueness about where marks are stored |
| 6. File location and retention | Local, or a written retention period | "Securely stored" with no time period |
| 7. Which marker it writes | Configurable flags, stars or labels | One fixed scheme that overwrites yours |
| 8. Editor handoff | A named list including your editor | Logos on a homepage and no documentation |
| 9. Frames it did not pick | Nothing moved unless you switch it on | Files relocated by default |
| 10. If you stop paying | Originals and markers usable without the app | Selections live only in their database |
| 11. What the vendor keeps | Published cancellation and retention terms | You have to ask twice |
| 12. Support and refunds | Stated response times, a refund window | A chatbot and a community forum |
Score each one out of five with a sentence of explanation. The sentence matters more than the score, because in three months you will remember that a tool felt good and forget that it could not map your flags.
The interrogation only works if the trial is run properly, and most trials are wasted on poking at features. Use a job you culled by hand six months ago, one where you still remember the frames you chose. You need a known answer to measure against, which is exactly what a fresh shoot cannot give you.
Then look only at the disagreements. Frames the software picked that you did not, and frames you picked that it passed over. Everything else is noise. Ten minutes in that pile tells you whether its taste is compatible with yours.
Time the whole loop, card to shortlist to editor, including the export. Vendors advertise the analysis step because it is the fast part. You are buying the loop. Run both candidates in parallel across two or three real jobs before you move a season across.
Check the genres you shoot least often, too. Adobe describes Assisted Culling in Lightroom as optimized for portraits and headshots, which is honest and a fair reminder that every tool here has a home genre. Ours on that feature is the Assisted Culling review.
It would be unfair to publish an interrogation and duck it, so here are our answers, including the unflattering ones.
The photo picking app, native for iPhone, iPad and Mac, is still being built. It has not shipped, so if you need an answer this Saturday, run a trial from your shortlist instead. The booking page is live and free today. Founding members get the picking app the day it ships.
On question 9, the one about frames it did not pick, our answer is fixed and it is the reason the product exists in the shape it does. Kepla marks which photos it picked. It never deletes, moves or renames a single file. Every original stays exactly where it was and anything not picked is one tap away, which means there is no setting anyone can turn on by accident that sends a wedding into a rejects folder.
On question 10, that principle is also the exit answer. Because nothing is ever moved and nothing is ever destroyed, walking away from us leaves your folders exactly as your camera wrote them.
On question 3, there is no trial to describe yet, and we are not going to pretend otherwise. What is live is the free booking page, which holds a date and takes a deposit, and which none of the culling tools on your shortlist do at all. If you want the wider picture of how the category is priced before you commit to anything, the full spec table sits in our culling software comparison.
Ask which RAW formats and camera bodies are supported, what the trial includes and on which tier, whether existing ratings and flags are preserved, where markers are written, whether files leave your machine and for how long they are kept, which marker the export writes, which editors are supported, what happens to unpicked frames, what you keep if you cancel, and how support works on a deadline.
It depends entirely on where the ratings were written. If the tool writes to XMP sidecar files or into your catalog, the marks stay readable in your editor with the software gone. If selections live only inside the vendor's own database, they leave when the subscription does. Test this during the trial by exporting a small project, quitting the app, and opening the folder in your normal editor.
Most write a rating, flag or label and leave the files where they are, but some offer to move unpicked frames into a separate rejects folder, and that setting is occasionally easy to switch on without noticing. Find the answer in the settings before you point the tool at real client work. Kepla's answer is that it marks its picks and never deletes, moves or renames a file.
One week is enough if you spend it properly. Run it on a shoot you culled by hand months ago so you have a known answer to compare against, look only at the frames where you and the software disagreed, and time the whole loop from card to editor rather than the analysis step alone. Also confirm the features you liked exist on the tier you plan to pay for.
It removes the exit problem, which is worth something. Photo Mechanic sells a perpetual license at $299, or $399 for Photo Mechanic Plus, against $149 a year for the subscription, so the break-even sits somewhere in year three. The tradeoff is that AI selection tools improve quickly and a perpetual license eventually stops receiving that improvement. Prices checked August 2026.
No, and no vendor worth buying from will tell you otherwise. Use the scores and groupings to direct your attention, then make the final call yourself with the client's story in mind. The value is in not having to look at every near-identical frame at full size, not in handing over the decision about which moment matters.
Kepla for Mac clears the obvious misses from a card, names the reason on every frame it sets aside, and leaves the choosing to you. Nothing is ever deleted, moved or renamed. Free through the private preview · the first hundred photographers keep it at $99 a year.