Twelve out of 23. That is how many accounts had a static image, not a video, as their cheapest ad with meaningful spend. The figure comes from Biddyco’s analysis of 34 Meta accounts spending $10.2 million, and it makes a video-only creative policy hard to defend.
It does not make a static-first policy the new rule. In the tighter 23-account cut, statics took 38% of spend among the top 10 ads in each account, at a median cost per result 10% below their own account averages. Video took 62% at 4% above. Those are observations about selected ads that got delivery, not instructions to move 38% of your budget to statics.
I used to tell clients video would win if we just gave it enough variants. I was wrong often enough that I stopped saying it. This dataset is enough to challenge a video-only production policy, but not enough to prescribe a static-versus-video split. The hiring question is whether an agency sets the format mix from your account evidence or the format it prefers to sell you.
Start with the sample, not the budget recommendation
Biddyco ranked ads by spend in 34 Meta accounts over September 28 to October 27. Ranking by spend filters out the $40 test that got lucky and keeps the ads Meta actually fed. Comparing each ad with its own account average matters, too: a supplement purchase and a software demo need not have the same dollar value for both ads to be read as a percentage above or below their respective baselines.
| What Biddyco sampled | Figure | What it lets us read |
|---|---|---|
| Accounts in the full sample | 34 | Results across accounts rather than one brand’s case study |
| Combined Meta spend | $10.2 million | Ads with real delivery behind them |
| Observation window | September 28–October 27 | A bounded snapshot, not a forecast |
| Comparison method | Percent versus each account’s average | Relative performance without averaging unlike result costs in dollars |
The limit is selection. Spend-ranked ads tell us what got budget, not whether every submitted creative received an equally useful test. Nor does one window settle what will work next quarter. But a universal rule has a lower bar to clear: if static ads repeatedly survive delivery and produce inexpensive results, “serious brands only make video” needs better evidence.
Biddyco says it calls something a pattern only when three or more unrelated accounts agree, and anonymizes the brands. That does not turn the analysis into a controlled experiment. It does make its observations more useful than a portfolio slide about one client’s best month.
The same analysis offers a warning about the metric agencies like to put in the biggest font. Higher-CTR ads tended to convert worse after the click, with a correlation of -0.24. Ads beating their account averages by 20% or more on cost per result converted 52% better after the click despite 9% lower CTR. A cheaper click does not rescue an offer or page that fails to convert. Ask what happened after the click before praising the click.
The 38% figure is a spend share, not a target mix
Here is the tighter cut that tends to become a budget recommendation by the time it reaches a sales deck. Biddyco examined the top 10 ads by spend in 23 accounts. The table describes those selected ads, not all ads or all spend across the full 34-account sample.
| Format in the selected top ads | Share of top-ad spend | Median cost per result versus own account average |
|---|---|---|
| Static | 38% | 10% below |
| Video | 62% | 4% above |
Static ads held more than a third of this selected spend while the median static ran below its account baseline. That is a reason to keep the format in contention. It is not a reason to set next month’s static budget at 38%. Observed spend share is not an optimal allocation.
Spend share is not inventory share, either. An account could upload many more videos than statics and still have statics take substantial spend. As AdGenz describes Meta’s early delivery behavior, budget can move toward early promise after a few hundred impressions while other ads get little exposure. The top-10 table therefore shows survivors. It cannot tell us whether every ad left out of that table got a fair chance, or how the mix would look under a different test design.
That distinction matters when you hire an agency. A shop should be able to explain how it gave each format room to prove itself before it cites spend share as proof that one format deserves more production.
Ten percent below average is not ten percent cheaper than video
The median static in Biddyco’s 23-account cut ran 10% below its own account-average cost per result. The median video ran 4% above its own account average. These are account-relative figures, not a head-to-head price comparison. Do not subtract them and announce that statics beat video by 14%. The table did not run that test.

Nor does cost per result settle whether an ad made money. Cost per result divides spend by the event an ad set is optimized to produce. That event could be a purchase, a lead, a view or a ThruPlay, under that account’s performance goal and attribution window. Rows from different accounts can therefore represent different kinds of results. A lower number is not automatically a better customer.
Say you are spending $20k a month and an agency reports a lower cost per result from statics. Ask what those results were, how they converted after the click, and what they returned in first-purchase ROAS. Until then, you have a delivery fact, not a business decision. Keep the median next to its baseline, and make the agency connect it to the result you actually buy.
Twelve cheapest workhorses were static
This is the line I would put in front of a founder being pitched a video-only retainer: in 12 of the 23 accounts, the cheapest ad with meaningful spend was a static. Not a barely delivered test that happened to get a lucky purchase. An ad that took enough spend to matter. A video-only rule has to explain those 12 accounts.

The finding breaks the policy argument that statics are legacy work and every production dollar belongs in motion. It does not prove statics win in your account. This cut cannot identify your placement mix, offer strength or retargeting share. It also cannot tell you what happens when Meta distributes spend differently.
There is corroboration, with the same limits. An earlier Biddyco analysis of 848 top ads across $14.9 million found statics were 40% of winners and finished within 4% of video on cost per purchase. MHI’s data on 74 DTC brands found image beat video on median CPA and first-purchase ROAS even while video took most spend. Neither analysis supplies a universal budget split. Together, they make a simpler decision defensible: keep statics in testing until your own delivery and return data say otherwise.
Ask what produced the result, not just which format won
An account-level read answers whether statics survived. It does not explain why. Was the winning static in prospecting or retargeting? Did it carry a stronger offer, lead to a cleaner page or run in a different placement mix? Those differences can travel with format in a report, then disappear when someone turns the report into “video wins” or “static wins.”
MHI offers a useful illustration of the tension: in its data, video CPM was $27.41 versus $19.42 for image, while video won link CTR 1.98% to 1.38%. More expensive impressions and more clicks can interact in ways a format label alone cannot explain. If you cannot separate delivery cost from what happened after the click, you cannot confidently copy the result.

Another dataset shows why I would ask for both spend and return. JetFuel’s read of 83 brands plus 10 cleanly tracked accounts found produced creative running about 4:1 video over static. Video won CTR in all 10 tracked accounts and took most budget in nine, yet the static and catalog layer matched or beat video ROAS in seven. Its examples included coffee at 2.2x static versus 1.9x video and wine at 8.1x versus 4.3x. That is not a clean isolated static-versus-video trial, particularly with catalog in the return comparison. It is a good reason not to confuse the format receiving the most exposure with the format producing the best return.
Creative variety deserves the same scrutiny. In the Andromeda discussion cited here, a new angle or persona matters more to creative diversity than putting the same photo in a different color. I would ask whether an agency tested a different idea or shipped five versions of one video and called it testing. A report that skips downstream quality, including who bought and at what margin, has not earned a production decision.
Make the agency show how it would set your mix
If an agency recommends video-only, I want the account math that earned that rule. The request is not for a prettier creative gallery. It is for a report I can inspect and a test the agency is willing to run even if the result means producing less video.
- Show the last 30 days of top ads by spend against each account’s average. Include format and the event behind cost per result. Do not use a winning ad from one account as the answer for another.
- Split prospecting from retargeting. Put CPM, link CTR and conversion rate after the click on the same row. Then I can see whether a claimed format win comes with different delivery costs or different behavior after the click.
- Follow the result toward money. Report cost per qualified lead or purchase, then first-purchase ROAS and margin where tracking permits. If the monthly report stops at CTR and spend share, it cannot tell me whether a production recommendation improved the business.
- Give both formats a test the agency will respect. I would keep two or three statics live against each new video, isolate variables, use equal budgets and wait for about 50 results per variant before treating the outcome as a production rule. If video earns the spend under those conditions, buy more video. If statics again hold the efficiency floor, you may have saved a shoot.
That last request is the cheapest way to expose a supplier preference. A shop that accepts the possibility of a static win is selling an allocation process. A shop that insists on video before seeing the account is selling a production plan. What the deck calls “synergy,” I call paying twice for the same guess.
The incentive is worth examining, too. A partner charging one flat fee of $1,499 a month for Meta, whatever you spend, can recommend either format without losing a video-production fee. A supplier paid per video produced cannot say the same. Pricing does not prove which creative will win. It does tell you which recommendation may be harder for a supplier to make.
That is the decision these numbers support. Twelve of 23 cheapest workhorses were static; statics took 38% of selected top-ad spend at a median 10% below account average. None of that sets your mix. Hire the shop that lets your account set it and proves the decision in rows you can audit. Fire the one that arrived with the answer in the treatment.

