Cost per ad creative: how to turn any quote into one number you can compare

01Why does a quote never tell you what an ad costs?

Because the seller picks the unit and your ad account picks a different one. Trend.io sells a video. Pencil sells a generation. Flocksy sells 42 hours a month. Vidpros sells two editor hours a workday. ManyPixels sells one daily output. Five sellers, five denominators, and not one of them is the ad you would run on Monday.

None of that is dishonesty. Every one of those units is a real thing the vendor controls, and none of them controls whether the output survives your review. A credit is a candidate. An hour is capacity. A daily output is a promise about a queue. The conversion into a finished ad happens on your side of the table, and your budget runs on the conversion.

One route on that list prints its own division. Trend.io does the arithmetic on the page, at $91.67 a video on Starter and $69.14 on Scale. Everybody else, us included, sells a thing and lets the buyer work out what an ad costs. That gap is where budgets get set wrong, and what ad creative costs in the US is the price list this arithmetic runs on. Which unit a monthly agreement should commit to in writing is what a creative retainer includes.

02Divide by what survives, not by what was made

In our production log, fewer than half of the films that clear every automated check survive the moment I sit down to watch them. The gate-pass-to-keep rate is 43%. We counted it here, on our own work. It is not a benchmark for anybody else and it is not a complaint about the tools. A surplus exists so that most of it can lose.

So we generate at least a twice-over surplus and pick from it. A bad take gets re-rolled, never repaired, and every re-roll is somebody's decision before it runs. The sister journal carries the full arithmetic in the hidden line item of what failed renders cost, and the rule itself in re-roll, never repair.

Three keeps, three different months

Three that survived

SaaS - studio demoHotel client - campaign reelFragrance client - campaign reel

Three frames that survived selection, from three different films, two of them client work and one a studio demo. Each came out of a batch at least twice this size, and the rest of that batch is the part no per-unit price on any vendor page accounts for.

A price per generation is a price per candidate. Nobody has ever run a candidate.

The line I use when a quote arrives

Push a keep rate through a per-unit price and the effect is arithmetic rather than opinion. A $100 unit at a perfect keep rate costs $100 an ad. At 43% the same unit costs $233. The vendor's price did not move. Your budget did, by a factor the pricing page never mentioned.

03What is the cost per finished ad on your quote?

Put your own numbers in. The calculator takes what you pay in a month, how many units that money buys, and the share you would actually put spend behind. It returns one figure you can set beside any other quote written in any other unit.

cost_per_finished_ad = monthly_total / (units_per_month * keep_rate)

monthly_total    everything paid inside one calendar month:
                 subscription + platform fee + seats + creator pay + rush
units_per_month  what that money buys, in the vendor's own unit
keep_rate        share of delivered work you put spend behind,
                 counted on your own last batch

worked           1000 / (16 * 0.45) = 138.89

Two warnings about that number. Your keep rate is a guess until you have counted it, so count it: watch a delivered batch cold, mark the ones you would put money behind, write the fraction down. And it moves. A brief that settles the story before anything renders moves it further than any change of tool I have watched anybody make.

The third output is the one that stings. Money spent on files you never ran does not feel like a cost, because nobody invoices it separately and the vendor delivered exactly what was promised. It surfaces later as a budget that keeps needing topping up, and as a quarter in which you shipped fewer ads than the plan said you would.

04How much does a 30 second video ad cost?

There is no published benchmark, and I went looking. The $342,000 median for a thirty-second US commercial that circulates on vendor blogs traces back to no publication by the body it is attributed to. What can be priced is the short-form ad a DTC brand actually runs: $69 to $375 a video with a creator in it, or about $31 an edit if you supply the footage yourself.

Thirty seconds is also the wrong question for most people asking it. Almost nothing a DTC brand runs on Meta or TikTok is thirty seconds long, and duration is a craft decision rather than a price band, which the sister journal argues in duration is the action's pace. Price the batch you will run, not the format television handed you.

05Three costs that never show up on a quote

The first is review time. Somebody on your side watches everything, decides, and writes it down, and that hour has a salary attached to it. The second is the round. Unlimited revisions also means unlimited waiting, while two rounds on a fixed batch means the work lands on a date, a trade laid out in what counts as a revision round. The third is the file you paid for and never ran.

All three are paid in time, and time is what a small team has least of. A quote can only price the part of the job the vendor does. The part you do scales with how much output you have to look at, so a route that doubles your volume doubles that cost even while it halves the price on the page.

One thing this arithmetic cannot do is tell you whether the ad works. We hold no outcome data on client campaigns, no hook rates and no return on ad spend attributed to our work, and a studio quoting you one should be able to show you the account it came out of. Cost per finished ad is a budgeting number. How many finished ads you need before you know anything is a separate question, answered in how many creatives do you need.

Every vendor in this market prices the attempt. What you buy is the keep, and until somebody publishes the ratio between the two, the only trustworthy number on a quote is the one you worked out yourself, on your own batch, after you watched it.