Why Meta Ignores Most of Your New Ads (and How to Fix It)

By Thisura, founder of Run1Ads · 2026-11-04

Quick answer

Meta usually ignores most of a new batch because it treats ads that look alike as one ad, so they share a single slot in the auction. Practitioners also point to early leaders getting most of the budget. Test fewer, clearly different concepts, give tests their own budget, and judge them on ROAS from real orders, not on who got the spend. Meta has not published a full explanation, so treat the details below as practitioner findings.

Why does Meta only spend on one or two of my new ads?

Nikhil Sharma, who consults for Shopify brands, wrote in an article updated on October 30, 2026 that Meta groups ads that look alike into a single "Entity ID", and that one Entity ID gets one entry into the auction. In his words, thirty near identical ads can behave like one. Entity IDs are not shown to advertisers in Ads Manager or the API, so this is an inference from how accounts behave, not a Meta statement.

The practical result is familiar. You upload a batch, two or three ads take almost all the spend, and the rest sit near zero. Sometimes those two or three really are better. Often the rest never had a fair chance.

Do different headlines and captions make ads different?

Mostly no. The same source says Meta looks at the visuals, the audio and the structure, not only the words. A new headline over the same image, or a new caption on the same video, probably still counts as the same ad.

A simple check: put the first frame of every ad side by side. If a stranger could sort them into two piles, you have about two ads, not twelve.

What does count as different: a new shoot or location, a different format (talking to camera, montage, static image, carousel), a different opening two seconds, different audio, and a different reason to buy.

How many ads should I test at once?

The same article suggests 8 to 12 clearly different concepts per campaign with 2 to 3 variations each, not 30 variations of two ideas. Jon Loomer, who tested Meta's creative testing tool, lets you test between two and five ads at a time and recommends making them differ in format, theme, customer type and message.

For a small store, five different ads with enough budget will teach you more than twenty similar ones. Match the number of ads to what your budget can fund. If each ad cannot get close to the roughly 50 purchases a week that Loomer uses as a rule of thumb, you will not get a clear answer.

Why does Meta pick a winner so fast?

Several practitioners say campaign budget optimization sends spend to early front runners. That is useful for scaling and bad for testing, because early data can be a few lucky purchases. Meta also does not promise that the ad which won your test will keep the most spend later. Loomer notes that after a test ends, Meta stops splitting spend evenly and may favor a different ad.

Two ways to give new ads a fair chance:

Put new concepts in their own test ad set with its own budget, so they do not compete with proven ads.
Use Meta's creative testing tool. It splits delivery evenly across two to five ads, recommends using no more than 20% of the budget, and defaults to 7 days according to Loomer. It needs the Highest Volume bid strategy.

How do I check whether ignored ads were actually worse?

Do not judge by spend alone. An ad with a little spend and a good result may be a winner that never got room.

List each ad with its spend, purchases and ROAS for the same dates.
Compare Meta purchases with real orders in your store for those days. If Meta shows 20 purchases and your store shows 14, fix tracking before you judge any creative. The post on Meta Pixel not tracking purchases shows how.
Work out ROAS from real order revenue divided by ad spend, per ad where you can.
Check whether 80% of spend sits on under 20% of your ads. That points to grouping, not preference.

Meta may also show a Creative Similarity Score and a Creative Fatigue Score under Analyze and Report, then Ads Reporting, then Account Insights. The article above says it is still rolling out, so it may not appear in your account, and it is only a proxy.

How do I check this automatically?

Run1Ads shows which of your ads bring sales, based on your Meta data and your store orders, and flags tired ads before ROAS falls. It reads your ad account through the official Marketing API, read only, and tells you the one fix that matters today. You can see it on your own store in the live demo.

When is an ad actually tired?

Do not turn an ad off because it had a bad day. Look for a steady fall in ROAS over several days together with rising frequency, not one weak result. Sharma says fatigue has become faster and recommends a refresh every two to three weeks, with about half new concepts, 30% improved versions of what works and 20% proven ads left running. That split is one practitioner's suggestion, not a Meta rule.

Key facts

Sharma reports that ads which look alike can be grouped into one Entity ID with one auction entry (article updated October 30, 2026). Meta has not confirmed this in documentation we found.
His suggested target is 8 to 12 distinct concepts with 2 to 3 variations each.
Same source: a similarity score above 60% is reported to trigger grouping, and he recommends staying under 40%. Entity IDs are not visible to advertisers.
Jon Loomer says Meta's creative testing tool takes two to five ads and recommends no more than 20% of the budget for the test.
Loomer says the test defaults to 7 days and needs the Highest Volume bid strategy.
Loomer notes the winning test ad may not keep the most spend once the test ends.

FAQ

Why is Meta not spending on my new ads?

Most often the new ads look too similar to each other or to an existing ad, so Meta treats them as one. Other causes are a budget too small to fund the whole batch, or early spend going to the first ad that looked good.

How many creatives should I test?

Start with a handful of clearly different ads, not a pile of similar ones. One practitioner suggests 8 to 12 concepts per campaign for larger accounts. Smaller stores can start with three to five.

Should I turn off ads with no spend?

Not right away. Check whether they are different enough from the ads getting spend, and test them in their own ad set. If they are near copies, replace them with new concepts.

Is there a tool that shows which of my Meta ads bring sales?

Yes. Run1Ads compares your Meta ads with your store orders, shows which ads bring sales, and flags tired ads before ROAS falls. It is read only and does not change your ad account.

Is something like this costing your ads sales right now?

Run1Ads is a live Meta ads consultant for e-commerce brands. It checks your ad account, Pixel and store every hour and tells you the one fix that matters today, with exact steps. We advise, you decide.

Try the live demo Start free

No card needed · Official Meta Marketing API