The short answer
When Facebook ads stop working, check in this order: tracking, then the offer and landing page, then the creative. Most creative problems are a lack of real variety. Winners on our board spread across five awareness levels with none above 34%, run a median 120 days, and lean on video lengths of 30 to 120 seconds.
Check in this order
When results drop, most teams change the ads first. That is often the last thing to touch. Work through the causes in the order that is cheapest to check and most likely to explain a sudden change:
- Tracking. Did purchases in Ads Manager drop while orders in your store didn't? Then the ads may be fine and the signal broke. See Conversions API and signal quality.
- Site and checkout. Speed, stock, payment errors, a broken discount code. Check the funnel by hand on a phone.
- Offer and price. A price change, a competitor's sale, a promotion that ended.
- Account changes. Did a significant edit restart learning? New ads added to the scaling ad set, budget jumps, paused ad sets. See the learning phase.
- Season and auction. Q4, big sale days and election periods raise costs across every advertiser.
- Creative. Fatigue on your top ads, or a set of ads that are all the same idea.
Symptoms and likely causes
| What you see | Likely cause | First check |
|---|---|---|
| Purchases down in Meta, orders steady in the store | Tracking | Events Manager, deduplication, event match quality |
| CTR steady, conversion rate down | Landing page, offer or checkout | Session recordings, checkout errors, price changes |
| CPM up across all ads at once | Auction or season | Compare with the same week last year |
| CTR down and frequency up on top ads | Creative fatigue | Ad-level frequency and CTR over four weeks |
| Spend stuck on one ad, new ads get nothing | Concepts too similar, or new ads in the wrong place | Are new ads different ideas? Do they have their own budget? |
| Ad sets stuck in Learning Limited | Too many ad sets for the budget | Consolidate; see account structure |
| Good results for days, then a sudden drop | A significant edit or a delivery shift | Change history for the ad set |
If you work down to the creative rows, the rest of this article is for you.
How your ads compare with $50k+ winners
Our October 9, 2026 snapshot has 2,443 live Meta ads from 749 brands, each past $50k in estimated spend. They are a useful benchmark, because they show what keeps getting funded. Use this table as a checklist against your own active ads.
| Winners on the board | The number | Check yours |
|---|---|---|
| Video share | 81% video, 19% image | Running only one format? |
| Video length | Median 64 s; 30 to 120 s is the strongest band | All under 15 s? Under-15 s videos reach the top tier 3.4% of the time |
| Reader awareness | Five levels used, none above 34% | Are all your ads written for the same reader? |
| Narrative angle | Root cause is 34%; seven other angles share the rest | Is every ad the same argument? |
| Proof | Strong proof on 4% of ads, but 16.3% of those reach the top tier | Do your ads show the product working, or only claim it? |
| Call to action | 95% ask for a purchase | Is the ask clear? |
| Landing page | 36% product page, 23% dedicated page; quiz, listicle and advertorial over-index | Does the page match the ad's promise? |
| Lifespan | Median 120 days live | Are you replacing ads before they tire? |
| Versions | 71% run as a single version | Are you diluting winners with dozens of near-copies? |
Top tier means the top 10% of the board by Ad Radar's winner score (live 21+ days); about 10% is the baseline.
Two rows deserve emphasis. Awareness and angle variety is where most struggling accounts fall short: every ad tells the same story to the same reader, so delivery has one option and burns through the people it fits. Our creative diversification playbook shows how to spread out. And lifespan: many accounts kill ads too early. If your best ad from two months ago is paused, try turning it back on before writing something new. The data on how long winners run is a useful reality check.
One myth the data doesn't support: hiding the brand. Among videos with a transcript, 157 name the brand in the first sentence, and 12.1% of those reach the top tier, against 9.0% for the rest.
Is it the hook?
If your creative audit points at weak openers, compare the patterns you use with what reaches the top tier on the board:
| Opening pattern | Ads | Share reaching top tier | Median estimated spend |
|---|---|---|---|
| Specific number opener | 411 | 12.9% | $274K |
| Warning or shock opener | 156 | 10.3% | $250K |
| First-person story | 438 | 10.0% | $231K |
| Problem/benefit opener | 714 | 9.1% | $209K |
| Offer-first opener | 255 | 8.2% | $183K |
| Question opener | 366 | 7.1% | $219K |
Every pattern here has hundreds of winners, so none of them is wrong. But if every ad in your account opens with a question or a discount, you are using the two patterns that reach the top tier least often. Openers with a concrete number or a specific warning give the viewer something to check, and they hold up better on the board. Our Facebook ad hooks guide has examples of each pattern.
Three winners that do what struggling ads skip
$2,300 for that.
- Est. spend
- $4.9M
- Days live
- 233
- Format
- Video, 57s
- Variants
- 2
Opens on the price of the alternative the viewer already knows about. Written for someone comparing options, not for everyone.
Excuse me, what's this strap on your phone?
- Est. spend
- $2M
- Days live
- 449
- Format
- Video, 51s
- Variants
- 2
A street-interview opener that shows the product in use before anyone explains it. Live for 449 days.
Hey everyone, it's Blaine with Seven Sons Farms.
- Est. spend
- $1.5M
- Days live
- 1036
- Format
- Video, 194s
The founder names himself and the brand in the first line, talks for over three minutes, and the ad has run nearly three years in the top tier.
A two-week recovery plan
Days 1 to 3: rule out the non-creative causes. Compare Meta purchases with store orders by day. Test the checkout. Read the account's change history for the last two weeks.
Days 3 to 5: audit the creative set. List every active ad with its angle, awareness level, format and length. Mark the ones that are the same idea. Our winning ad checklist helps score each one.
Days 5 to 7: bring back what worked. Reactivate paused ads that once performed, as new ads in a fresh testing ad set.
Days 7 to 14: add two or three real new concepts. Pick awareness levels and angles you don't cover. Brief them from winners in your niche rather than from scratch.
Ad Radar is built for that last step: filter the board by niche, see which angles and awareness levels win there, and turn a winning ad into a creative brief for your product. The features page shows how it works.
Figures marked as estimated spend come from Ad Radar's model of engagement on public Meta Ad Library ads. They are estimates, labeled as such, and are best used to rank ads against each other.