Your Shopify Meta ads add to cart but don't buy. The leak is the store, not iOS-14.

Meta says people are clicking and adding to cart. Your bank says they're not buying.

The click-through rate looks fine. CPMs are tolerable. Add-to-Cart events are firing in Events Manager. And Purchases sit near zero while ROAS glows red. So the theory forms fast: iOS-14 broke attribution, Meta is under-reporting sales that actually happened, the numbers are lying. That theory is comfortable because it means the store is fine and the platform is at fault. It is also, almost always, wrong. The gap between "adds to cart" and "buys" on Meta traffic is rarely an attribution artifact. It is the store failing to close the coldest visitor it will ever get.

Skip the manual diagnostic. The StoreMend audit checks your live Meta landing page across 12 conversion categories in 60 seconds and tells you which leak is losing the sale. $39 one-time. 30-day no-questions refund. Run an audit at storemend.com.

First, kill the attribution theory. It takes two minutes.

Rule out gross tracking breakage once, then stop thinking about it. Fire one test purchase on your own store and watch Events Manager: if a single Purchase event lands within a few minutes, the pixel works and the funnel numbers are directionally real. iOS attribution changes which ad gets credit for a sale. It does not manufacture add-to-cart events that never converted, and it does not hide purchases that actually happened from your own order feed in Shopify Admin. If Shopify shows the same near-zero order count that Meta shows, there is nothing left to blame on attribution. The adds to cart are real. The purchases are real. The leak is between them, and it lives on the store. Everything below is about finding it.

Why cold Meta traffic dies on a store that organic tolerates

Meta traffic is the coldest, most impulse-driven, most mobile traffic a Shopify store gets. Every one of those three properties makes it less forgiving of a store-side leak, and they stack.

Coldest. A Google searcher typed a need. A returning visitor typed your brand name. A Meta scroller was mid-feed, watching something else, when a product interrupted them. They tapped on impulse and landed on a store they had never heard of. Zero prior context. Zero relationship. The trust an organic or branded visitor brought with them, the cold visitor needs the store to build from scratch, in seconds, before the impulse fades.

Most impulse-driven.Intent-based traffic waits a beat. A searcher who wanted "waterproof hiking boots size 11" will tolerate a slow page because the page holds the thing they came for. Impulse traffic does not wait. The desire that produced the tap is seconds old and evaporating. Any friction between the tap and the buy button competes with a feed the visitor can flick back to instantly.

Most mobile. Meta traffic is overwhelmingly phone traffic. Mobile share of Shopify sessions runs around 79% platform-wide, and paid social skews higher still. So the cold impulse visitor is also on a small screen, on a mobile connection, with one thumb. The store is being judged in its least capable rendering context by its least patient visitor.

Put together: cold plus impulse plus mobile is the hardest visitor a Shopify store will ever try to close, and it is exactly the visitor Meta sends. A store that converts organic search at a healthy rate can convert cold Meta traffic at a fraction of it, and the difference is not the ad account. It is that the store was built for people who already trust it. The diagnostic below walks the four leaks that hit cold Meta traffic hardest.

Leak 1: The mobile product page is too slow for an impulse click

Speed is not a tie-breaker for cold Meta traffic. It is the gate. Largest Contentful Paint, the time from tap to the largest visible element finishing paint, is the number that matters, and Google's "good" threshold is 2.5 seconds at the 75th percentile of real mobile visitors. "Poor" is above 4. Plenty of Shopify product pages sit in the 4-to-8-second range on a mid-tier phone once the app stack loads.

An organic searcher will wait through a 5-second paint because the page holds the result they searched for. The cold impulse tapper will not. The desire is already fading, the feed is one flick away, and a white screen is the fastest way to send them back to it. This is the exact leak covered in the mobile page-speed deep-dive: the second-four bounce, where the visitor taps, stares at a blank or half-painted page, and leaves before the hero image arrives. It hits every vertical, and it hits cold paid traffic hardest because the click was already paid for.

To check your own Meta landing page: open the exact URL your ad points to in PageSpeed Insights, read the mobile score first, and read the field data (real visitors over 28 days) rather than the lab number. A mobile field LCP above 4 seconds means the impulse click is dying on the doorstep. The usual cause is not the theme, it is the stack of marketing, review, and popup apps each injecting its own JavaScript on every page view.

Leak 2: The store built no trust for a visitor who arrived with none

A cold Meta visitor has never seen the brand before. They did not search for it. A friend did not send them. An ad interrupted them, they tapped, and now they are deciding whether to hand a card to a store they cannot vouch for, on a product they saw ninety seconds ago. Trust is the whole game here, and most stores never built the signals a cold visitor needs because their organic and returning traffic never needed them.

Cold traffic scans for a specific set of trust signals in the first seconds, and they are almost always the ones a store built for warm traffic never had to add. Its returning and organic buyers already trusted the brand, so the store was never forced to prove itself to a stranger. The signals a cold visitor looks for before handing over a card:

Reviews with photos, above the fold on the product page. Five text reviews read as new and risky. Real photo reviews are the single strongest proof a cold visitor can get that the product is real and other people received it.
A returns line near the buy button.Not buried in the footer. "30-day returns, no questions" next to the price removes the "what if this is a scam" hesitation that cold traffic carries by default.
An about or story surface that answers who is behind this. A branded visitor already knew. A cold visitor is asking. A store with no about page tells a cold visitor nothing, and for anything applied to the body or ingested, that silence is disqualifying.
Security cues at the payment step. A visible secure-checkout or padlock cue near the card field, and recognized payment logos. Cheap to add, and they answer the exact fear a first-time visitor has at the exact moment they feel it.

None of these are new to a store that sells to warm traffic. They feel redundant because the store's existing buyers never needed them. Cold Meta traffic needs all of them, and the store that never installed them loses the cold visitor at the trust check, not the price. The conversion diagnostic covers the trust-thin pattern in depth, including the recurring shape where a brand converts organic well and paid poorly for exactly this reason: the traffic changed, the store did not.

Leak 3: Checkout surprises an impulse buyer at the worst moment

A cold impulse visitor who reaches checkout is the closest thing to a won sale that Meta traffic produces. Losing them at the final step is the most expensive leak of all, because the store already paid for the click, already survived the speed gate, and already earned enough trust to get a card halfway out. Two things kill it here, and both hit impulse traffic harder than any other kind.

Surprise cost at checkout. The Baymard Institute puts average cart abandonment at 70.22% across 50 studies, and the single most common reason people give for abandoning is extra costs (shipping, taxes, fees) being too high, cited by 39% of abandoners, almost double the second-place reason. A warm buyer might forgive a shipping charge that appears at the last step. An impulse buyer whose desire is already cooling treats the surprise as the excuse to close the tab. The cost the visitor sees in the cart has to match the cost at checkout. Free shipping above a threshold is the cleanest fix.

No express-checkout path. A cold visitor on a phone will not type a full address and card into a multi-field form for a store they met two minutes ago. Shop Pay, Apple Pay, Google Pay, and PayPal Express let them skip the forms entirely and pay with a thumbprint. For impulse mobile traffic, the presence of an express button in the cart is often the difference between a sale and an abandon.

Checkout integrity is the one leak Meta traffic shares with paid search, so this article does not re-run the full checkout teardown. The mechanics of the checkout surface, field count, guest checkout, cost transparency, and where standard Shopify limits your control, are covered in the Google Ads conversion diagnostic, which shares this exact checkout-integrity core. The difference is only in how forgiving the traffic is: paid search buyers arrive with intent and tolerate a step or two, cold Meta buyers do not.

The 3-second above-the-fold trust test

Here is the fastest way to feel the gap between a store built for warm traffic and a store that can close a cold Meta visitor. Open the exact URL your Meta ad points to on your phone, in incognito, so no login or history colors it. Let it load. Look for three seconds, the length of a cold scroller's patience, then look away and answer three questions out loud:

What is this, and does it match the ad I just tapped?
Why should I trust this store enough to buy?
What do I tap next?

A page built for people who already trust you fails this test quietly. It opens with a lifestyle hero and a brand-aspiration headline that means something to existing customers and nothing to a stranger. It puts reviews and returns policy below three scrolls. It assumes the visitor already knows the brand, because every visitor it was designed for did. The cold Meta visitor answers "I don't know" to all three questions and taps back to the feed.

A page built to close cold traffic answers all three in the first viewport: the product and its match to the ad are obvious, at least one trust signal is visible without scrolling, and there is exactly one clear thing to tap. If your Meta landing page fails the 3-second test, no amount of ad-account tuning will fix it, because the leak is happening after the click.

Not sure which of the four leaks is losing your Meta traffic? Run the free checker on your Meta landing page at storemend.com/free and see the store-side score a cold visitor experiences.

What 1,091 audited stores show about the store side

The reframe is not a hunch. It comes out of the corpus. Across 1,091 Shopify storefronts audited in Q2 2026, the store-side surface that cold traffic depends on was broken far more often than most operators assume, and each break maps to one of the leaks above.

Product and Organization schema was absent on 733 of the 1,077 classified stores, 68.1% of the classified cohort. Schema is the structured layer that tells search and AI systems what the product is, what it costs, and whether it is in stock. Its absence is a symptom of the same neglect that leaves cold visitors without visible reviews and returns cues: the store was built to look right to a human who already trusts it, not to prove itself to a stranger or a machine.

The same neglect shows on the human-trust side. The reviews, returns, and security cues a cold Meta visitor scans for are exactly what a warm-traffic store skips, because its returning and organic buyers never needed to be convinced. Combined with the mobile-speed findings covered in the page-speed deep-dive, the result is a fleet of stores that work fine for the traffic they already have and fail the traffic they are paying Meta to send.

This is also why the "adds to cart but no sales" symptom shows up across so many stores running paid social. The full teardown of that specific funnel shape lives in the add-to-carts-but-no-sales diagnostic. The short version: the add-to-cart is the visitor telling you the product and the price landed. The missing purchase is the store telling you it could not carry that interest across the trust and checkout gaps.

The leak most operators never check: the store is invisible to AI

There is one more leak, and it is the one almost no one running Meta ads thinks to check, because it does not show up in Ads Manager at all.

Cold Meta traffic increasingly does not go straight from the ad to a purchase. A visitor taps the ad, sees the product, and then does what cold buyers do in 2026: they leave to verify. They ask ChatGPT whether the brand is legit. They ask Perplexity for alternatives. They search for reviews. The store that surfaces as a credible, structured answer in that second-opinion moment wins the sale back. The store that is invisible to AI systems loses the visitor to whichever competitor the assistant recommends instead.

The corpus number here is stark. Of the 1,077 classified stores, 778 (72.2%) were AI-invisible: their pages carried too little structured data for an AI shopping assistant to read and cite them. Another 299 (27.8%) were only transitional. Zero were fully ready. That means the overwhelming majority of Shopify stores paying to send cold traffic into the top of the funnel are also invisible at the exact moment that cold traffic goes to verify them. The ad spend creates the doubt, and a competitor's store answers it.

This is the twist that closes the loop on the attribution theory. The operator blames Meta for under-reporting sales. In reality the store did get the interest, the visitor did go to verify, and the store was not there to be found. That is not an attribution problem. It is a store that was never built to be discovered by the systems cold buyers now trust.

Fix the store before you scale spend

The order of operations is the whole point. Pouring more Meta budget into a store that cannot close a cold visitor buys more add-to-cart events that never become revenue. The click is paid for at the moment of the tap. If the landing page is slow, if the trust signals are thin, if checkout surprises the buyer, if the store is invisible when the visitor goes to verify, that spend leaks out the four holes above no matter how good the targeting gets. Fix the store first. Then let the budget compound instead of drain.

FAQ

My Meta ads show adds to cart but almost no purchases. Is it iOS-14 attribution?

Almost never. Rule out gross tracking breakage first: fire a test purchase and confirm one Purchase event lands in Events Manager within a few minutes. If it does, tracking works and the numbers are directionally real. iOS attribution blurs which ad gets credit, it does not invent add-to-cart events that never converted. A large add-to-cart count with near-zero purchases is a store that cannot close a cold visitor, not a reporting artifact.

Why do my Meta ads convert worse than my Google traffic or my organic traffic?

Because Meta traffic is colder. A Google searcher typed intent and a returning visitor typed your brand name. A Meta scroller was mid-feed, tapped an impulse, and landed on a store they had never heard of, on a phone. That visitor has zero prior context and needs trust the store never built. The same store that converts organic at 2% can convert cold Meta traffic at a fraction of that, and the gap is store-side, not ad-side.

Should I pause my Meta ads until the store is fixed?

Fix the store-side leaks first, then scale spend. Every dollar of Meta budget pushed into a store that cannot close a cold visitor is a dollar spent buying add-to-cart events that never become revenue. The click is already paid for by the time the visitor lands. If the landing page loses them in the first four seconds, that spend is gone. Fix the leak, then let the budget compound instead of leak.

How do I know the problem is my landing page and not my creative?

The symptom tells you. If click-through is healthy and add-to-cart events are firing, the creative did its job: it earned the tap and the visitor got far enough to add a product. The drop happens after that, between the cart and the bank. That is the store surface, not the ad. Weak creative shows up as low click-through, not as adds-to-cart that never buy.

Do trust badges fix cold Meta traffic?

A single badge does little. Trust signal density is what moves a cold visitor. A cold Meta scroller arrives with no prior relationship to the brand, so the store has to build trust from zero in seconds: visible reviews with photos, a real returns line near the buy button, a secure-checkout cue at the payment step, and an about or story surface that answers who is behind the product. Five visible signals outperform one badge by a wide margin.

How fast does my mobile product page need to load for Meta traffic?

Largest Contentful Paint under 2.5 seconds at the 75th percentile of real mobile visitors is the good threshold Google publishes. Cold Meta traffic is less forgiving than any other source because the visitor never chose to be on the page. An organic searcher waits a beat for a result they wanted. An impulse tapper does not. A 5-second mobile paint on a cold Meta landing page loses the visitor before the hero image finishes.

Methodology and disclosure

Findings draw from 1,091 Shopify storefronts audited in Q2 2026. Each store was checked across 12 conversion-facing categories using public storefront pages, the same surface a human visitor would see. Findings are aggregated and anonymized; no private data was accessed and no store's ad-account data was used. Cart-abandonment figures are from the Baymard Institute's published cart-abandonment research. Cohort sample refreshes each quarter.

Fix the store before you scale spend. The StoreMend audit checks your live Meta landing page across all 12 conversion categories in 60 seconds and tells you which leak is losing the sale. $39 one-time. 30-day no-questions refund. storemend.com