A UGC test fails quietly when nobody writes the numbers down. The brand remembers that the video "did not work", the creator remembers that the brand never gave it a fair run, and the next brief repeats the same hook with a different creator. A tracker prevents that, and it takes twenty minutes a week.
One row per creative. One tab per month. Spend in Naira, no currency conversion mid-sheet.
The five numbers that decide a test
Hook rate is 3-second video plays divided by impressions. It tells you whether the first frame stops a scroll. On TikTok, Ads Manager reports 2-second and 6-second views, so divide 2-second views by impressions to keep the column comparable.
Hold rate is ThruPlay divided by 3-second plays on Meta, or 6-second views divided by 2-second views on TikTok. It tells you whether the body keeps the viewer who stopped.
CTR is link clicks divided by impressions. All-click numbers flatter a creative; a high all-clicks rate with a low link-click rate means people watched and did nothing.
CVR is purchases divided by link clicks. It tells you whether the landing page and price hold up after the ad does its job.
CPA is spend divided by purchases. ROAS is revenue divided by spend. Add CPM to see whether a weak result came from the creative or from an expensive auction.
The formulas, with a worked example
| Metric | Formula | Example |
|---|---|---|
| Hook rate | 3s plays ÷ impressions | 30,000 ÷ 120,000 = 25% |
| Hold rate | ThruPlays ÷ 3s plays | 9,000 ÷ 30,000 = 30% |
| CTR | Link clicks ÷ impressions | 1,800 ÷ 120,000 = 1.5% |
| CVR | Purchases ÷ link clicks | 60 ÷ 1,800 = 3.3% |
| CPA | Spend ÷ purchases | ₦60,000 ÷ 60 = ₦1,000 |
| CPM | Spend ÷ impressions x 1,000 | (₦60,000 ÷ 120,000) x 1,000 = ₦500 |
| ROAS | Revenue ÷ spend | ₦360,000 ÷ ₦60,000 = 6.0 |
The same ₦60,000 bought 60 sales at a ₦1,000 cost each. If the target CPA is ₦1,500, this creative has room to scale. If the target is ₦700, the hook is fine and the offer or the landing page is the problem.
The log
| Date | Platform | Creator | Concept | Spend | Impr. | 3s plays | Hook | ThruPlay | Hold | Clicks | CTR | Purch. | CPA | ROAS | Decision |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 12 Nov | Meta | A | Serum, price confession | ₦30,000 | 60,000 | 15,000 | 25% | 4,500 | 30% | 900 | 1.5% | 30 | ₦1,000 | 6.0 | Scale |
| 12 Nov | TikTok | B | Serum, bathroom shelf | ₦20,000 | 90,000 | 12,000 | 13% | 2,400 | 20% | 540 | 0.6% | 6 | ₦3,333 | 1.8 | New hook |
| 13 Nov | Meta | C | Serum, before and after | ₦30,000 | 55,000 | 14,000 | 25% | 5,600 | 40% | 1,100 | 2.0% | 44 | ₦682 | 8.8 | Scale |
Keep the creator column even when the same creator makes three variants. Over a quarter you will see which creators produce hooks that hold, and that changes how you brief.
For WhatsApp-first businesses, add a "chats started" column and count a chat as the conversion. Purchases happen off-platform, so CPA against website purchases will look like a failure when the ad is actually working.
Set the test up so the numbers mean something
Change one thing per test. If the hook, the creator, and the offer all change at once, the result tells you nothing about which one moved the number.
Use the same audience and the same daily budget across variants. A creative that got ₦30,000 and one that got ₦8,000 are not comparable.
Run for at least three days before judging. A ₦2,000 spend is noise, and Nigerian auction prices shift through the week.
Pick one attribution window and keep it. Seven-day click and one-day view is the common default on Meta; a comparison against a one-day click number will flatter whichever creative ran last.
Separate cold traffic from retargeting. Retargeting always wins on CPA, and mixing the two makes every cold-traffic creative look broken.
Name creatives the same way every time: [date]-[platform]-[creator]-[concept]-[hook]. "Video final 2" tells you nothing in three weeks.
Kill, iterate, or scale
| What you see | What it means | What to do |
|---|---|---|
| Hook under 15%, CTR fine | The first frame is losing people who would have clicked | Reshoot the hook, keep the body |
| Hook fine, hold under 20% | The promise in the hook is not paid off | Rewrite the middle, cut the setup |
| CTR fine, CVR under 1% | The ad works, the page does not | Check price, load speed, checkout |
| CPA above target by 50% after three days | The creative is not competitive | Kill it and reallocate spend |
| CPA below target | There is room | Duplicate and raise budget 20% per day |
| ROAS high, spend capped | Demand is being left on the table | Raise the cap before touching creative |
Two rules keep this honest. Judge a creative on the metric it was built to move, and never scale a winner past the point where CPA starts climbing. Most Nigerian accounts find that ceiling within a week.
The weekly review
Open the sheet every Monday and sort by CPA. Read the top three and the bottom three, then write one sentence on what the winner has that the loser does not. That sentence becomes the next brief.
Retire a creative after two weeks of flat performance, even if it was a winner. Frequency climbs, CTR falls, and the same video that carried the account in October costs double in December.
For the reporting layer above this sheet, read how to measure UGC ROI. For the brief that produces testable variants, use the UGC brief template. For Meta-specific setup, read UGC for Facebook ads.