The fastest way to lose money on client video work is to accept feedback without limit. One client asks for a small change, then another, then asks to "try something completely different", and a two-day job becomes a two-week job at a two-day price.

The fix is not to be stricter. It is to define what a round is before the first edit, so the boundary exists when you need it rather than being invented in an awkward conversation.

What a revision round actually is

A round is one pass of consolidated feedback, delivered once, and answered in one new version.

Three properties matter and all three are usually missing:

Consolidated. All the feedback arrives together, not in a trickle across four days.
Bounded. It addresses the existing cut, not a different video.
One version out. You deliver one new cut, not a menu of options.

If feedback arrives as a stream, you are doing a round per message, and you will do five of them.

Write the policy before the first edit

The number belongs in the brief or the quote, in plain language, with no hedging:

Two rounds of revisions are included. Additional rounds are billed at the hourly rate.

Two is a sensible default for short-form: one round to correct what you got wrong, one to refine taste. Three is generous but workable. Unlimited is a business model that only works if you charge for it up front.

The sentence also has to say what a round covers — the current cut, with consolidated notes — or the definition above becomes an argument later.

The difference between a revision and a new direction

A revision improves the video that exists: trim this pause, change that caption, swap the music, keep the opening but cut the third beat.

A new direction replaces the premise: use a different hook, lead with a different product, tell the story from the customer instead, cut it to eight seconds.

Both are legitimate requests. Only the first is a revision. The second is new work, and pricing it as a revision is how editors end up subsidising a client's second idea.

In practice the tell is simple: if the feedback cannot be expressed as a change to something in the current timeline, it is a new direction.

Handling "can we try something else"

The response that keeps the relationship and the boundary is to agree, then scope it:

Yes, that is a different approach rather than a tweak. I can cut that version as a new piece of work at X, or we can use the second revision round on it and drop the caption changes. Which suits you?

Offering a choice converts a demand into a decision the client makes. It also makes the cost visible without accusing anyone of anything, which is usually all that was needed.

The polite no that keeps the client

Most scope creep is not malice; it is a client who does not know where the line is because nobody drew one. So the no has to be about the work, not the person:

That is outside what we scoped, so I would want to quote it separately rather than rush it into this delivery.

Note that it is not a refusal. It is a reclassification, with the implication that quality matters more than speed. Clients accept that framing far more often than they accept a flat no.

Charging for the third round without a fight

When you reach the third round, quote it before doing it — a short message with the estimated time and the rate, and a request to confirm.

Two things make this land. First, the policy was already stated at the start, so this is not new information. Second, you quote the time, not a round number: "about an hour" is verifiable, and a client who has already asked for fourteen small changes will recognise that an hour is generous.

Keep the rate reasonable for the relationship. The point of charging is not the money, it is that the client starts consolidating their feedback, which is the actual goal.

Document every round

Keep one file per project listing each round, the date it arrived, what was in it, and what you changed. Two reasons: when a client says "I asked for that three times", you can point at what was in round one; and when you price the next project, you know how many rounds it actually took rather than how many you assumed.

That record is also the evidence base for turning the client into a retainer, where you are proposing a fixed monthly number and need to know the real shape of a month.

When to walk away

The pattern worth ending a relationship over is not pickiness. It is feedback that reverses itself: a change requested in round one undone in round two, repeatedly. That does not converge, no matter how many rounds you allow, and every hour after the second reversal is unpaid.

How to consolidate a round

If feedback arrives as a stream, ask once, early, for it to be collected: "Send everything in one pass by Thursday and I will cut it all together." That sentence is the difference between one round and five.

The request works better with the reason attached, because consolidated notes produce a better cut than sequential ones — changes interact. Fixing a caption in isolation and then discovering the music was also meant to change produces two mediocre versions instead of one good one.

The revision log

One table, one row per note: the timecode, what was asked, what changed, and any note that was declined, with the reason. Five columns.

It answers three questions that would otherwise become arguments: what was in this round, what happened to each note, and whether a request was new or a repeat. It also makes the next quote accurate, because you can count rounds instead of recalling them.

Fixed fee or hourly

A fixed fee is easier to sell and makes every extra round a direct loss, which is why scope discipline matters most on fixed-fee work. An hourly rate makes extra rounds painless to accept and easy to bill, and suits clients whose needs genuinely fluctuate.

The hybrid that works for most short-form work is a fixed fee for a defined scope with two rounds, plus an hourly rate beyond it. The client gets a number they can approve, and the third round has a price that nobody has to invent in the moment.

Whichever you choose, put the round count and the extra rate in the same document. Separating them is how editors end up explaining their own policy after the fact, which always sounds like a penalty rather than an agreement.

Finish the current delivery, then decline the next project without drama. The client can stay a client of somebody else, and the time goes to work that finishes.

Revision Rounds Without Scope Creep