Photo Approval When an Agency Sits Between You and Client

Why the middle position breaks ordinary approval

A direct photographer-to-client review has two parties and one handoff. Put an agency in the middle and you have three parties and, per round, at least four handoffs: photographer to agency, agency to client, client back to agency, agency back to photographer. Each handoff is a copy, and copies drift.

Three failures show up again and again:

• Relay loss. The client circles a hand in frame 31; the agency’s summary says “client wants some retouching on 31.” The photographer now has to guess which part.
• Premature exposure. The agency forwards the whole set, including frames it would never have shown, and the client falls for the wrong frame.
• Two yeses. The agency says “approved” and the client says “actually, can we see another option.” Nobody knows which one counts.

If you have already dealt with a client who keeps adding reviewers, the pattern will feel familiar; the approval chain fix applies here too, with one extra tier to design.

The two-tier model: who sees what

Treat the agency’s review and the client’s review as separate rounds with separate audiences. The agency tier is internal and wide: the art director, the producer and anyone who needs to judge the work against the brief. The client tier is narrow and curated: only frames the agency is willing to stand behind.

The crucial point is the last row. The agency’s yes is a gate, not an approval. It means the frame may go in front of the client. Only the client decider’s yes means the frame is approved. Write that sentence into the kickoff email, because the failure it prevents (agency says approved, client disagrees) is the most expensive one on this list.

For who should hold each decider seat, the sign-off order for brand photography has a RACI you can adapt; the agency version simply runs it twice.

Cost the relay before you design around it

Here is the arithmetic, with the assumptions shown so you can swap in your own. Suppose a round produces 40 notes, retyping or re-explaining a note takes two minutes, and the job runs two rounds with the agency in the middle.

These are illustrative assumptions, not measurements. The point is the shape: relay cost scales with notes times rounds, and it lands on the person whose hourly rate is usually highest. It also does not count the rework when a relayed note is wrong, which is the larger cost and the harder one to estimate.

The cheapest removal is to stop relaying. If the client’s comments land directly on the frame as markup the photographer or retoucher can open, the agency’s job shifts from transcription to judgment: reading the notes, resolving conflicts, and deciding what the client actually meant. That is the work they are paid for.

Let the client write on the frame, and let the agency see it first

Two design choices make the two-tier model work in practice.

Notes belong to a frame, not to an email. A comment pinned to the area it concerns carries its own context. Nothing needs summarising, so nothing is lost in summary. This is the same reason per-image comments beat whole-gallery feedback, and it matters more when a third party is in the path.

The agency gets an internal pass before the client does. The agency reviews the select set on its own link, culls and annotates, then releases a narrower set to the client. In Appruva, a share link grants access to the object shared (an image, a folder or a project’s images) and never to the studio itself, so one shoot can carry an internal link for the agency and a separate link for the client without either seeing more than it was given. Clients open the link in a browser with no account.

What the agency decides, and what it must not

A middleman is most useful when its role is explicit. Before the first round, agree on three things in writing.

1. The agency may veto a frame from the client set. It may not edit the client’s notes. The client’s words go to the photographer as written.
2. The agency resolves contradictions between its own reviewers before release, never after. The method is covered in resolving conflicting stakeholder feedback.
3. Silence has a defined meaning. If the client decider has not responded by the stated date, say in advance whether that is approval, a reminder or an escalation. Do not leave it to interpretation on the day.

That third item is the one agencies skip, and it is where schedules slip. A deadline with no stated default is a suggestion.

Hand the notes to the retoucher without a second relay

The relay problem has a second half: the point where approved notes go to whoever retouches. If notes are copied from the review into an email, then into a brief, then into the retoucher’s task list, you have rebuilt the middleman problem one step later.

Appruva has a Photoshop panel, a UXP panel for Photoshop 2023 (24.0) or newer on macOS or Windows, that reads the numbered notes inside the document, places the client’s markup on the file as a layer and delivers the retouched round back as a new version. A freelance retoucher needs no account for this: access is a share link created with retoucher access, and revoking the link ends the access. That keeps the agency out of the file-passing business. Whether or not you use it, the principle holds: notes should reach the person doing the work in the form the client wrote them.

A one-page kickoff you can send before the shoot

Send this as the first email on the job. It costs ten minutes and removes most of the arguments above.

• The two review tiers, and who sits in each.
• The named decider in each tier, and the sentence “the agency’s yes means safe to show, the client’s yes means approved.”
• The number of rounds included and what a further round costs (see how many revision rounds a contract should include).
• Round deadlines, and what happens when a deadline passes in silence.
• Where comments go: on the frame, in the review link, not in email.

The agency relationship is worth protecting because it brings repeat work. A clear process is a kindness to everyone, and it is part of why an agency will book you again.

A worked example: 120 frames, two tiers

Take a campaign shoot that produces 120 selects. The agency’s art director and producer review all 120 on the internal link and pass 30 to the client. That is 30 divided by 120, or 25 percent of the set, which is the figure to hold in your head: the client is choosing among a quarter of what you shot, and every one of those frames has already survived a creative judgment.

Round one goes to the client decider with the 30 frames. Suppose she marks 12 with notes and ticks 18 as approved. The agency reads the 12 notes, notices two that contradict each other because her colleague added a comment, and asks her which one she meant. That conversation happens inside the review, before anything reaches the photographer. Ten clean notes go to retouching and the two resolved ones are folded in.

Round two is the retouched set coming back as new versions of the same frames, so the client compares against what she saw rather than hunting for a fresh file. Nothing in this sequence required the agency to retype a comment. It required the agency to decide three things: which 30 frames, which of two notes stood, and when to release the next round.

Failure modes to watch for in the first month

Even with the model in place, the first few jobs surface predictable problems. Watch for these.

• The client link gets forwarded. A decider shares it with a colleague who comments freely. Agree in the kickoff that extra reviewers are added by name, and treat an unexpected commenter as a prompt to ask who they are, not as noise.
• The agency pre-edits the notes anyway. Habit is strong. If the account team keeps summarising, show them how much of the summary the photographer could not act on.
• The internal tier turns into a meeting. If three people from the agency must agree on every frame, you have rebuilt design by committee one level up. One agency decider means one agency decider.
• The photographer is left out of the gate. A photographer who never learns why 90 frames were dropped cannot improve the next set. Ask for a one-line reason on the vetoed frames; it is cheap and it sharpens your selects.

None of these is exotic. Each is a role that was never named, which is why the kickoff page matters more than any tool you choose.