You launch 15 new ads. You feel good about it. Then you check Ads Manager three days later and only 2 of them are spending money.
The other 13 are sitting there, doing nothing.
This happens to a lot of advertisers right now, and the reason usually comes down to one thing: the Andromeda Entity ID. Meta’s ad system no longer counts your ads the way you do. It groups similar-looking ads together and only gives that group one real shot at reaching people. If you don’t understand how this grouping works, your reach quietly shrinks and you never know why.
In this post, we’ll walk through three simple fixes for your Andromeda Entity ID setup, so your ads actually get a fair chance instead of fighting each other in the background.
What Is an Entity ID, in Plain Words
Think of an Entity ID as a fingerprint Meta gives each ad based on how it looks and sounds.
If two ads share a similar background, similar footage, or a similar structure, Meta treats them as basically the same ad even if the headlines are different.
So when you upload 10 versions of one video with different text, Meta might only see one real ad. The other nine are just noise competing for the same slot.
Why Andromeda Entity ID Matters for Your Reach
Reach depends on how many chances your ads get to enter Meta’s auction.
Every unique Entity ID gets its own ticket into that auction. Ads that share an Entity ID share one ticket between them.
So if your creative library is full of near-duplicates, you’re not actually running 20 ads. You might be running 3.
Fix 1: Stop Making Small Edits and Call Them New Ads
The mistake: Changing a headline, swapping a color, or trimming two seconds off a video and expecting it to count as a fresh test.
Why it fails: The Andromeda Entity ID system looks at the visual structure of your ad, not just the words. If the scene, setup, and delivery style stay the same, Meta still sees one concept.
The fix: Build ads that are genuinely different from each other. Change the format, the setting, or the way the message is delivered, not just the caption.
A Simple Example
Say you sell skincare cream. Instead of making five ads with the same model in the same lighting saying slightly different lines, try:
- One ad as raw, phone-shot customer testimonial
- One ad as a before-and-after style demo
- One ad as a calm, studio-style explainer with text on screen
These three will likely get three separate Entity IDs, because they don’t share the same visual pattern.
Fix 2: Check Your Andromeda Entity ID Similarity Score Before Blaming the Algorithm
Meta now shows advertisers a similarity score inside Ads Manager. Most people never check it.
If your similarity score is high, that’s your answer. It’s not that Meta stopped liking your brand. It’s that your ads are too visually alike to be treated as separate opportunities under the Andromeda Entity ID grouping.
| Similarity Score | What It Means | What to Do |
|---|---|---|
| Low (under 30%) | Your ads look distinct to the system | Keep testing this way |
| Medium (30–60%) | Some overlap, limited new reach | Adjust one major visual element |
| High (above 60%) | Ads are grouped, reach is capped | Rebuild with a new concept, not a new edit |
Check this score before you add more ads. Adding more of the same only makes the pile bigger, not more useful.
Fix 3: Give Each Andromeda Entity ID Room to Prove Itself
The mistake: Launching a new concept, then judging it and killing it after one day because it “didn’t perform.”
Why it fails: New Entity IDs need time to gather enough data. Cutting them early means Meta never learns who to show them to.
The fix: Let a new concept run for about 7 to 14 days with a steady budget before deciding to pause it. Judge it on trend, not on the first 24 hours. Every Andromeda Entity ID you launch deserves that same window before you decide it’s a loser.
That’s 5 uses of “Andromeda Entity ID” plus the standalone “Entity ID” mentions — in this section alone that’s already past 2% density for this word count. If you paste this into the full 1,000-1,200 word article, which already uses the keyword naturally elsewhere, you’ll likely land at 3%+ overall. That crosses from SEO-friendly into stuffing territory, which is a red flag for search engines, not a ranking boost.
Actionable Tips You Can Apply Today
- Pull up your active ads and group them by concept, not by campaign name.
- Check the Creative Similarity Score for your top-spending ad set this week.
- Pick 3 to 5 genuinely different concepts and build around those instead of dozens of small variations.
- Give new concepts a full 7-day window before judging results.
- Retire the weakest concept fully instead of tweaking its headline again.
Common Mistakes to Avoid
Mistake 1: Confusing ad volume with ad diversity. Uploading 30 ads feels productive, but if they all share a visual pattern, you’ve really only made one ad. Focus on distinct ideas, not distinct file names.
Mistake 2: Judging performance too early. Many advertisers pause a new concept after one bad day. This starves Meta’s system of the data it needs, and you lose a concept that might have worked with more time.
Mistake 3: Ignoring the similarity score. This score is right there in Ads Manager, yet most beginners never open it. It’s one of the fastest ways to understand why reach dropped.
Conclusion
our reach isn’t shrinking because Meta stopped liking your ads. It’s shrinking because too many of your ads are wearing the same fingerprint, and only one of them gets to compete at a time.
Fix this by building genuinely different concepts, checking your similarity score before adding more creativity, and giving each idea enough time to prove itself. Do these three things consistently, and your Andromeda Entity ID setup starts working with you instead of against you.
FAQs
What should I check before implementing this strategy?
Look at your current creative library first. Group your ads by visual style, not by name, and see how many actually differ from each other before you build anything new.
How can I tell if my results are improving?
Watch whether spend spreads across more of your ads, not just one or two. If more concepts start pulling budget over 1-2 weeks, that’s a sign your Entity ID setup is healthier.
Which mistakes cause the biggest losses?
Pausing new concepts too early causes the most wasted budget, because you never let Meta gather enough data to judge them fairly.
When should I avoid using this method?
If you’re running a very small daily budget, don’t launch too many new concepts at once. Spreading a tiny budget across many Entity IDs starves all of them of data.
What signs indicate it’s time to scale?
When a concept holds steady or improving results for 10+ days without rising costs, and its similarity score to other ads stays low, it’s a solid candidate to scale.
