Your Meta Ads are probably not broken where you think they are
When an eCommerce brand comes to us and says “Meta stopped working,” the problem is rarely what they think it is. They assume the algorithm changed, or their audience got saturated, or Meta just “isn’t working anymore.” Those are symptoms. They are not causes.
The actual causes are almost always structural. They sit in the tracking layer, the measurement framework, the creative library, or the account architecture. And they are fixable.
We audit every new account by working through the full funnel, from tracking and data quality through to measurement, creative, audience structure, and nurturing. The problems almost always cluster into the same six categories. Here is what we find, in the order we check for it.
Fix 1: Your conversion tracking is broken or incomplete
This is the first thing we check on every account, and it is the most common root cause of underperformance. If Meta cannot accurately see your purchase events, every other optimisation is built on bad data.
The most frequent tracking issues we find:
Missing or misconfigured Conversions API (CAPI) Many accounts are still running on Pixel alone. Since iOS 14 reduced browser-level signal quality, accounts without CAPI active are working with a materially incomplete picture of their conversions. Meta’s algorithm optimises against the events it can see. If it can only see 60 to 70 percent of your purchases, it is making targeting and bidding decisions based on a partial dataset.
Implementing CAPI increases your event match rate, which means Meta sees more of your actual conversions. Higher match rates lead to larger, more accurate audience pools for both prospecting and retargeting. This is not a marginal improvement. On accounts where we have implemented CAPI for the first time, the impact on delivery quality and cost efficiency is immediate.
Duplicate or inflated events When both Pixel and CAPI are running without proper deduplication, Meta counts the same purchase twice. This inflates your conversion numbers and makes your reported ROAS look better than reality. If your Meta-reported ROAS seems too good relative to your actual Shopify revenue, check your event deduplication setup.
Wrong conversion event selected Some accounts optimise toward Add to Cart or Initiate Checkout instead of Purchase. This gives Meta a higher event volume to work with but trains the algorithm to find people who browse rather than people who buy. Unless your purchase volume is genuinely too low to sustain a Purchase optimisation (fewer than 10 to 15 per week), always optimise toward Purchase.
For the full CAPI and Pixel setup walkthrough, see our Meta Pixel and Conversions API setup guide.
Fix 2: You are measuring the wrong metric
This is the second most common problem, and it is the one brands are most resistant to hearing. If you are evaluating your Meta account on blended ROAS, you are almost certainly making bad decisions.
Blended ROAS combines revenue from new customers and returning customers into one number. A 4x ROAS looks healthy. But if 50 to 60 percent of that revenue is coming from returning customers who would have purchased anyway through email or organic, you are not actually acquiring new customers profitably. You are just taking credit for existing demand.
The fix: segment new versus returning customers
The metric that actually tells you whether Meta is growing your business is new customer acquisition cost (nCAC). To calculate it:
- Pull first-time customer purchase data from Shopify
- Isolate the spend allocated to prospecting (not retargeting)
- Divide prospecting spend by the number of new customers acquired
Now evaluate that nCAC against your customer lifetime value (LTV). If your nCAC is less than 30 to 50 percent of your first-year LTV, you have room to scale. If nCAC is approaching or exceeding LTV, you need to fix your acquisition economics before spending more.
This CAC to LTV relationship is the number that should drive every budget decision. Not ROAS. Not CPA in isolation. The ratio of what it costs to acquire a customer versus what that customer is worth over their lifetime.
For a full breakdown of nCAC calculation and attribution setup, see our guide to Meta Ads attribution: why your numbers don’t match GA4.
Fix 3: Your creative has gone stale
After tracking and measurement, creative is the next place we look. It is also the area where we see the most resistance to change. Brands often have one or two “winning” ads and keep running them until performance collapses.
The problem is not that those ads stopped being good. The problem is that Meta has shown them to everyone in your addressable audience who is likely to respond. Once frequency climbs above 4 to 5 over a 30-day window and your hook rate drops below 20 percent, the creative is exhausted. No amount of budget increase or audience adjustment will fix it.
The fix: creative diversification, not just creative refresh
Swapping in a new version of the same ad with a slightly different headline is not enough. What most underperforming accounts lack is creative diversity across multiple angles. If all your ads focus on the same value proposition (say, “premium quality” or “free shipping”), you are only reaching the subset of your audience that cares about that specific message.
We introduce new creative that touches different pain points and solutions:
- Problem-aware creative: Ads that name the specific frustration the buyer is experiencing and position your product as the resolution
- Social proof creative: Testimonials, reviews, and UGC from real customers speaking to their specific use case
- Educational creative: Ads that teach the buyer something about the product category before asking for the sale
- Comparison creative: Ads that position your product against the alternatives the buyer is already considering
Each angle reaches a different segment of your addressable audience. The result is not just better performance on individual ads but a larger total addressable reach within your prospecting campaigns.
For our full creative testing methodology, see the Meta Ads creative for eCommerce: what actually works in 2026.
Fix 4: No audience segmentation is creating a ceiling
Many eCommerce accounts run a single ASC campaign targeting everyone with no structural differentiation. ASC is powerful, but running it without segmentation means you have no way to understand where performance is coming from or where it is breaking down.
The fix: segment by customer type and funnel stage
At minimum, your account should separate:
- Prospecting (new customers who have never interacted with your brand): This is where ASC should do the heavy lifting with broad targeting and diverse creative
- Retargeting (people who have visited your site, viewed products, or added to cart): This should run in manual campaigns with content designed to educate and convert warm audiences
- Customer campaigns (existing buyers): Repeat purchase, upsell, and win-back creative should be separated so it does not inflate your prospecting metrics
Without this segmentation, you cannot tell whether your Meta spend is actually acquiring new customers or just re-engaging existing ones. And you cannot optimise what you cannot see.
Inside ASC specifically, set your existing customer budget cap to no more than 20 to 30 percent. This forces Meta to allocate the majority of your prospecting budget to genuinely new audiences rather than the path of least resistance, which is always re-targeting people who already know you.
Fix 5: Weak or missing nurturing is killing your conversion rate
Not every buyer converts on the first click. For most eCommerce products with an average order value above $50, the purchase decision involves multiple touchpoints over days or weeks. If your account has no nurturing layer, you are paying to generate interest and then abandoning the buyer before they are ready to purchase.
The fix: build a retargeting education layer
This is the manual retargeting layer we described in our Meta Ads retargeting for eCommerce: the full-funnel playbook. Its job is not to push another product ad at someone who already saw one. Its job is to serve content that helps the buyer make a faster, more confident decision.
What works in the education layer:
- Testimonial and review content: Real customer voices addressing specific hesitations
- Product education: How to use the product, what makes it different, why it matters
- Behind-the-scenes and founder content: Builds brand trust and emotional connection
- Before-and-after or results content: Visual proof that the product delivers on its promise
This content will not always get direct attribution credit. A buyer may see your education ad three times, then convert through a retargeting DPA or a direct site visit. Meta gives the credit to the last touch. But the education layer is what made the conversion possible.
Evaluate this layer not on its own ROAS but on the overall account conversion rate and the average time from first touch to purchase. If the education layer is working, both of those numbers should improve.
Fix 6: Your data infrastructure is not supporting your growth
This is the root cause that sits underneath everything else. When your tracking is partial, your metrics are wrong, and your creative is flying blind, the core issue is often that your data infrastructure was set up for a smaller, simpler operation and has not scaled with you.
The fix: implement CAPI, customer segmentation, and first-party attribution as a connected system
The highest-impact sequence of changes we make on inherited accounts is:
- Implement CAPI to increase event match rates. This gives Meta a more complete picture of your conversions and immediately improves audience pool size and targeting accuracy
- Layer customer segmentation on top by passing customer type data (new versus returning) into your conversion events. This lets you separate nCAC from repeat purchase metrics and finally see whether your prospecting is actually working
- Add a first-party attribution tool (we use Blotout) to get an independent read on what Meta is actually driving, separate from Meta’s self-reported numbers
These three changes, done in sequence, give you the data foundation to make real scaling decisions. Without them, you are guessing. With them, you can see exactly where your money is going, who it is reaching, and whether the growth is real.
The diagnostic sequence matters
One of the most common mistakes we see is brands trying to fix the wrong layer first. They invest in new creative when their tracking is broken. Or they increase budget when their metrics are misleading them. Or they restructure their campaigns when the real issue is that they have no nurturing.
The sequence matters. Always start with tracking and data quality, then fix your measurement framework, then address creative and audience structure, then build out nurturing. Each layer depends on the one before it.
If your Meta Ads are underperforming and you are not sure where to start, work through these six fixes in order. The answer is almost always in the first three.
Frequently Asked Questions
Why did my Meta Ads suddenly stop converting? Sudden performance drops are most often caused by creative fatigue (the same ads have been running too long), a tracking disruption (Pixel or CAPI misconfiguration after a site update), or a seasonal shift in buyer behavior. Check your event manager for data gaps first. If events are firing normally, check your creative frequency and hook rate. If both look fine, look at your audience overlap and segmentation.
How long does it take to fix an underperforming Meta account? Tracking and measurement fixes can be implemented in a few days and show impact within 1 to 2 weeks. Creative diversification takes longer because you need to produce and test new assets. Most accounts we take over show measurable improvement within 30 days, with full stabilisation at 60 to 90 days.
Should I pause my Meta Ads while I fix these issues? Generally no. Pausing campaigns resets the learning phase and loses the signal Meta has already accumulated. Reduce budget if you need to limit spend while you fix tracking, but keep campaigns active so the algorithm retains its learnings.
How do I know if my CAPI is working correctly? Check Meta Events Manager for event match quality. A healthy CAPI implementation shows an event match rate above 80 percent. If your match rate is below 50 percent, your CAPI setup needs attention. Also verify that your event deduplication is working by comparing Meta-reported conversions against your Shopify order count for the same period.
What is a good nCAC for eCommerce? It depends entirely on your product margins and customer lifetime value. As a general benchmark, if your nCAC is less than 30 to 50 percent of your first-year customer LTV, your acquisition economics are healthy. If nCAC is approaching your first-order AOV with no repeat purchase revenue, you need to either reduce acquisition costs or increase customer value before scaling.
How many different creative angles should I be testing? At minimum, 3 to 4 distinct angles running simultaneously. Each angle should target a different pain point, benefit, or emotional trigger. Within each angle, you can test variations of format (static, video, UGC), copy, and hook. The goal is to always have creative in testing so you are never caught without a replacement when a winner fatigues.
What to Read Next
- Meta Ads for eCommerce: The Complete Guide (2026) — The full strategic framework for running Meta eCommerce accounts
- Meta Advantage+ Shopping Campaigns: The eCommerce Playbook — How to set up and optimise ASC as your primary driver
- Meta Ads Benchmarks for eCommerce 2026 — CPM, CTR, hook rate, and nCAC benchmarks to compare your account against
- Meta Ads creative for eCommerce: what actually works in 2026 — How to structure creative tests and build a pipeline of winning ads
