Ad creative testing without an ad budget: organic posts as a test bed
Can organic posts test ad creatives before you spend on ads? What 31,973 posts of versioned videos say about noise, sample size and calling a winner.
By Abondy ResearchPublished Updated 8 min read
In short: You can use organic posts to test ad creatives before you put money into ads, but only for big differences. In our data one post tells you almost nothing: the first post of a video on an app landed at 10 times that video's median there or more 12.8% of the time, and under a tenth of it 21.6% of the time. Between two videos made for the same campaign, the one ahead after 10 posts was still ahead on later posts only 53.7% of the time, close to a coin toss, and a 2-times lead after 10 posts held up only 57.1% of the time. When two videos' medians over all their posts (40 or more each) were at least 2 times apart, the odd and even halves of their posts picked the same winner 81.3% of the time. So post each creative 20 times or more on the app you plan to advertise on, compare medians, and call a winner only when the gap is large.
Ad creative testing usually means paying for impressions on each version until one wins. Organic testing swaps that ad spend for posts: you put each creative out as normal posts and let the feed decide. It costs less, but organic views are noisier than paid delivery, so you need more posts per creative than you'd guess. The numbers below show how much noise to expect. For the paid side, what Meta and TikTok recommend per ad set is in how many creatives per ad set.
How noisy is a single organic post?
Very. These are the 272 uploads in our data with at least 20 posts whose views we read. An upload is one video a brand gave us; each post is one version of it on one app.
| One upload's posts | Value |
|---|---|
| Posts read per upload (median) | 100 |
| Versions per upload (median) | 43.5 |
| Median views of a post, middle upload | 59.25 |
| Best single post, middle upload | 2,100 |
| Best post ÷ median post, middle upload | about 53× |
| Uploads with at least one post past 1,000 views | 91.5% |
| Uploads with at least one post past 10,000 views | 11.0% |
The same video, cut into versions and posted on different accounts, spread from a few views to thousands. If you judged a creative by its best post, almost every creative would look like a winner. If you judged it by its first post, you'd be wrong a lot: on uploads with 30 or more posts we read on an app, 12.8% of first posts were at least 10 times the upload's median on that app, and 21.6% were under a tenth of it.
Where does the spread come from?
We split the spread in views (on a log scale, so a jump from 10 to 100 counts the same as 100 to 1,000) into parts, one app at a time.
| App | Share of spread between uploads | Share between poster accounts |
|---|---|---|
| YouTube Shorts | 13.4% | 40.5% |
| TikTok | 9.0% | 64.4% |
| Instagram Reels | 14.3% | 18.0% |
| 12.5% | 11.8% |
The first column is closest to what you're trying to measure in a creative test: how much views differ from one upload to the next. It's 9.0% to 14.3%, and it also carries anything else that differed between uploads, such as campaign or timing. The second column is how much views differ from one poster account to the next, measured on accounts with 10 or more posts, and on three of the four apps it's bigger, by far on TikTok (64.4%). TikTok moved most when we left out posts whose views we never managed to read: they were bunched in some uploads, and counting them as 0 had made uploads look more different than they were (25.9% in an earlier version of this page). The two were worked out separately, so they don't add up to a total. In the first column, the rest is spread between posts of the same video.
Two practical points follow. Spread each creative across many accounts, so no creative wins just because it landed on a stronger account. And test on the app you'll advertise on. The same version's first post on two different apps barely lined up: the rank correlation was 0.074 between TikTok and YouTube, 0.050 between TikTok and Instagram and −0.017 between YouTube and Instagram. A post that does well on TikTok tells you little about Shorts.
How many posts does a creative need?
We checked how well a short test predicts later posts, in two ways.
| Test | YouTube Shorts | TikTok | Instagram Reels | |
|---|---|---|---|---|
| Rank correlation, first 10 posts vs later posts, all uploads | 0.489 | 0.195 | 0.511 | 0.337 |
| Same, ranked only against uploads from the same campaign | 0.260 | 0.136 | 0.522 | 0.368 |
| Same campaign: leader after 10 posts still ahead later | 54.4% | 45.3% | 55.8% | 43.5% |
| Same campaign: leader after 20 posts still ahead later | 59.3% | 41.7% | 61.7% | 67.4% |
A rank correlation of 1 means the short test orders uploads exactly as later posts do, and 0 means no link. Across all uploads, 10 posts sorted them only moderately on YouTube (0.489) and Instagram (0.511), and poorly on Facebook (0.337) and TikTok (0.195). Inside one campaign, 10 posts sorted them poorly except on Instagram, and the leader after 10 or 20 posts stayed ahead only 42% to 67% of the time, depending on the app. An earlier version of this page had TikTok sorting uploads well (0.778); that came from never-read posts counted as 0, which pulled the same uploads down in their first 10 posts and in their later ones. The TikTok column now rests on 70 uploads with 30 or more posts we read.
Then we asked whether there was a stable difference to find at all. For pairs of uploads in the same campaign and app, each with 40 or more posts, we split each upload's posts into odd and even posts and checked whether both halves picked the same winner.
| Same campaign, same app, 40+ posts each | All apps | TikTok |
|---|---|---|
| Pairs | 3,745 | 257 |
| Both halves pick the same winner | 63.3% | 52.9% |
| Pairs whose medians were at least 2× apart | 40.2% | 0.4% |
| Both halves agree, when 2× apart | 81.3% (1,502 pairs) | 1 pair, agreed |
When two creatives were close, halves of 20 or more posts each couldn't tell them apart: they agreed 51.2% of the time. When their medians were 2 times apart or more, the halves agreed 81% of the time. On TikTok, only 1 of 257 same-campaign pairs differed that much, and the halves agreed 52.9% of the time, so the TikTok test looks like a coin toss.
How to run an organic creative test
- Test big differences. A new hook, a different angle or a different format. Videos made for the same campaign were often close in our data: 59.8% of pairs were less than 2 times apart, and on TikTok 99.6% were.
- Post each creative 20 times or more, on the one app you plan to run ads on.
- Spread each creative over many accounts, so the account doesn't pick the winner.
- Compare medians and the share of posts past 100 views. Ignore the best post. One big post can't tell a good creative from a lucky one.
- Call a winner only at about 2 times apart. Treat anything closer as a tie, and keep both.
- Then spend on the winner. Organic views aren't clicks or sales, so check the winner in a small paid test before you scale it.
Making 20 or more different-looking versions of each creative is the slow part by hand. Our video model Flagship 1.2 can make hundreds of versions of one video, each with its own crop, layout, caption and sound, and real people post them on their own accounts. You can see versions it has made on Explore.
What does it cost?
Three creatives, 20 posts each, on one app, is 60 posts. On an Abondy plan a post costs 12 credits, about 16¢ to 20¢ a post, so the test is at most about $9.60 to $12 for the posts, if every post is charged. Since November 20, 2026, you're charged only for posts that reach 100 views within 24 hours; the credits for the rest come back. Each plan includes free videos to make versions from: 100 a month on Plus, 1,000 on Pro and 2,500 on Ultra. Plans start at $20 a month.
How we counted
- Sample: posts made from August 29 to November 12, 2026 (New York time). We pulled the counts on November 20, 2026, when every post was at least 7 days old. 31,973 of the counted posts were versions made by our video models, from 395 uploads.
- Exclusions: Abondy's own test accounts, posts whose views were confirmed as bought, and posts that were flagged, rejected or deleted.
- Campaigns: the same-campaign pair tests come from 207 upload-and-app groups with 40 or more posts we read, in 6 campaigns. Pairs share uploads, so they aren't independent. Read those rows as rough.
- Views: each post's public count when we last read it, not a lifetime total.
- Updated November 21, 2026: an earlier version counted posts whose views we never managed to read as 0 views. They're now left out (11.3% of the version posts in this sample), and every figure was recounted on November 21, 2026 from the 28,356 version posts we read. TikTok changed most: a short test no longer sorts TikTok uploads well.
- Instagram: read the Reels numbers as a floor. Some counts come from each post's public page, and in a 16 Nov 2026 check of 37 posts that page showed about a third of the real count.
- Associations, not proof: these numbers describe our posts. They don't promise how a creative of yours will do, on organic posts or in ads.
Questions people ask
Can I test ad creatives without spending on ads?
Yes, for big differences. Post each creative 20 times or more as organic posts and compare medians. In our data, when two creatives' medians over 40 or more posts each were 2× apart, both halves of their posts picked the same winner 81.3% of the time; for smaller gaps it was 51.2%.
How many posts does an organic creative test need?
More than 10. Inside one campaign, the leader after 10 posts stayed ahead 53.7% of the time. With about 20 posts each, gaps of 2 times or more were stable.
Does a winner on TikTok also win on YouTube Shorts?
Not post by post in our data. The same version's first posts on TikTok and YouTube had a rank correlation of 0.074, close to no link. Test on the app you plan to advertise on.
Is organic testing the same as a paid split test?
No. A paid test controls who sees each version and measures clicks and sales. Organic posts only show which creative the feed favoured, so confirm the winner with a small paid test.
Read next
- Playbooks · 17 Nov 2026 · 8 min read
How many creatives per ad set? What Meta and TikTok recommend
What Meta's and TikTok's own pages say about creatives per ad set or ad group, the hard limits, automated creative tools and how often to refresh.
- UGC distribution · 17 Nov 2026 · 8 min read
TikTok Promote vs paying people to post your video: what you get
TikTok Promote pays TikTok to show one of your posts to more people. Paying people to post gets many posts on other accounts. What each one buys.
- Playbooks · 17 Nov 2026 · 7 min read
Can you post the same video on two TikTok accounts?
What TikTok's rules say about one video on more than one account, how that differs from many people posting it, and what our post data can and can't show.
