AI culling vs outsourcing: the cost per wedding

The choice is not software versus a person. It is what each one costs on one wedding, how much reviewing you still do afterward, and what happens when the handoff goes wrong on a week you cannot afford it. Here are the published prices, the arithmetic that turns them into a per wedding number, and the tradeoffs that price alone will not show you.

You are comparing two different shapes of cost

Selecting is deciding which frames move forward: the blinks, the misfires, the six versions of one burst, the ones worth your editing time. It is not color work, retouching, album design or sequencing. Keep those separate or the comparison falls apart, because most outsourcing companies sell editing and treat selection as an extra line.

The second thing that breaks the comparison is the unit. A human culler quotes per job. Software charges per month. Those two numbers cannot be put side by side until both are expressed as dollars per wedding, and that is a division you have to do with your own booking count. We worked that method through for the whole stack in your culling stack, priced per wedding, and this article applies it to one decision.

The third thing is that neither option is the finished job. Both hand you a proposed selection that you still check before a gallery goes out under your name. So the honest comparison is vendor price plus your remaining hours, on both sides, for the same wedding. If it is your editing rather than your selecting that is drowning you, we costed that separately in outsourcing your editing.

What each side actually publishes

Start with real prices, not remembered ones. On the software side they are public and stable. On the human side they mostly are not, which is itself a finding.

Prices checked August 2026 from each vendor's own pricing page.

Two conclusions fall straight out of that table. A subscription is priced per month regardless of volume, so its cost per wedding is entirely decided by how many weddings you shoot: $120 a year is $12 a wedding at ten weddings and $4 at thirty, same invoice. A per image rate does the opposite. It scales with the shutter, so a 3,000 frame day at six cents a frame is about $180 before you have opened a single file, and the busy August wedding costs more than the quiet one in February.

FilterPixel belongs in this category too, but its published pricing moved across four of its own pages this month, so trust the checkout screen rather than any figure quoted anywhere, including here. The same caution applies to any culling company's numbers for a rival product.

Two formulas, your numbers

Write both columns for the same wedding, in dollars per job, with no monthly figures left in the sheet.

Then multiply every time line by the hourly figure you already use for business decisions and add it to the vendor cost. That hourly number is the whole game, so derive it rather than guess: the method is in the business case for culling software, and what those hours cost you today is in the true cost of manual culling.

Do not borrow a worked example, including from this article. The result swings on three personal things: how many weddings share the subscription, how fast your upload really is at 11 PM, and what an hour of your attention is worth mid season. Two photographers with the same gear can honestly reach opposite answers.

Add one line at the bottom for scale. The Knot's 2025 study of more than ten thousand couples puts the average US wedding photographer at $3,000 per wedding, quartiles $1,500, $2,800 and $4,700. Neither option will be the largest number on that invoice, which is a reason to choose on fit and risk rather than a few dollars.

The line both columns get wrong

Neither route takes the review away, and any calculation that assumes it does is wrong by several hours. What changes is the size of the pile. Before, you looked at three thousand frames and made three thousand decisions. After, you check a proposed set and correct it, which is faster, is still work, and carries its own kind of fatigue.

Measure it rather than estimate it. Time five real jobs end to end: getting files in, the first pass, your corrections, and the export or handoff into editing. Log the passive minutes too, the ones where an upload crawls or previews build, because that is time you cannot spend on anything else even though you are not touching the keyboard.

Photographers get this wrong in both directions. Some overstate the savings, because they still open every frame afterward out of habit. Others understate it, because they have forgotten the hours they spent last August comparing a thousand near identical dance floor frames. Five timed jobs settle it for your studio. For the shape of the work, a UK survey of more than 300 wedding photographers reported by PetaPixel put selection at 11 percent of working time and editing at 55 percent, though that was 2020, before this generation of tools.

Turnaround, and where the handoff breaks

Price is what you pay when everything works. Risk is what you pay when it does not, and these two options fail in different places.

A machine pass on your own computer fails visibly: it is slow, or it proposes a set you disagree with, and you know within the hour. A cloud tool or a human service fails at the transfer. A wedding is commonly two thousand to four thousand RAW frames, and getting that many files to somebody else is a real event on an ordinary home connection, before you count the failed upload you discover at midnight.

Ask five questions before you commit. How long does a full wedding take to reach the vendor, measured on your real connection rather than the speed on your bill? Can you start reviewing the moment you are home, or are you waiting on a return date? What happens when a second shooter's cards arrive two days late? Does it survive a hotel connection? Can you change your preferences without starting again?

Deadline shape decides a lot of this. If you have promised a preview within 48 hours, an outside party in the loop is a risk you are taking on someone else's calendar, which is exactly the kind of promise that should be written carefully in the first place: see the sneak peek clause. For a gallery due in six weeks, a trusted human culler buying you back an editing day is a much easier trade. Match the option to the deadline in front of you, not to an average job.

What you actually get for the money

Stated plainly, because an honest answer helps more than a scoreboard.

A human culler brings context. Tell them the groom's grandmother is in a wheelchair by the door, that the couple asked for the dog in everything, that the second day of a three day celebration matters most, and a good editor will act on it. They can follow a brief that changes per job, and be told what a "keeper" means in your work. That briefing is also the cost: it happens every time, and a vague brief returns somebody else's taste.

Software brings consistency and immediacy. It applies the same criteria to frame one and frame three thousand at 2 AM without getting tired, it never has a queue, and it does not need to be told anything twice. What it cannot do is know why a slightly soft frame is the photo of the day. It reads sharpness, eyes, faces and near duplicates, and it does that tirelessly, which is a genuinely different thing from taste. The specific gaps are collected in what AI culling misses.

There is also a middle path worth naming: doing a rough pass yourself before anything leaves your hands, so you are not paying per image to have obvious throwaways looked at. That is a per image saving on the human route and a smaller review on the software route, and it is covered in culling before you outsource.

Confidentiality, uploads and what you tell clients

Wedding files contain children, private homes and family moments never meant for an audience. That is a business consideration and sometimes a contractual one.

If you send files to a person or a company, put it in writing: confidentiality, whether they can subcontract, who owns the output, how long they keep your files, and how the files are stored while they hold them. If you use cloud based software, read the same terms with the same attention. Local processing on your own machine sidesteps most of this, which is worth something even when it is not the cheapest option.

Clients rarely ask, but you should have a straight answer ready. "I use professional tools to sort and organize the shoot, then I personally choose and edit everything you receive" is accurate for either route, and it is reassuring precisely because it is specific. What you should not do is let a tool or a vendor quietly become the author of a gallery your brand promises comes from your eye.

The rule to write down, and the test that settles it

It does not have to be one or the other. The workable hybrid is a machine first pass on every job plus a human on the exceptions, with the exception defined in advance rather than at midnight. Write the rule down: book outside help when a wedding runs past a set frame count, when two weddings land in one weekend, or when you travel the week after. A rule made while calm beats a decision made while exhausted.

Then settle it with evidence instead of opinion. Take one wedding you have already delivered, so you know what the right answer looks like. Run it through a free trial of an AI tool, and if you are seriously considering outsourcing, pay one vendor for the same job. Aftershoot gives 30 days with no card, Narrative gives a full free trial of its top tier with no card, and Imagen gives two free culling projects, so the software half of that test costs nothing.

Score both on the same five lines: keepers surfaced, moments missed, near duplicates still in the set, technical errors that got through, and hours from cards to a selection you would edit. Pay attention to the disagreements. Where the tool or the editor rejected something you kept for emotional reasons, that is what this decision costs you. Where they kept something you would not, that is the reviewing you just signed up for.

FAQ