Impressions with no conversions is a symptom, not a diagnosis. Five quite different problems produce it, they live at different layers of the funnel, and the order you check them in decides whether this takes an afternoon or three weeks.
The order below is deliberate: cheapest and most common first. Most of the accounts I see with this complaint are stuck on one of the first two, and both can be ruled out before lunch.
Layer 0: is it zero, or zero measured?
Start here every time, because it is the single most common cause and it costs nothing to check.
Open your backend — orders, CRM, lead inbox — and count what actually arrived in the same window. Then compare against the platform. You are looking for one of two situations:
- Backend has sales, platform shows none. It is a tracking problem. Everything else on this list is irrelevant until it is fixed.
- Backend is also empty. The conversions genuinely are not happening. Move down the list.
That comparison takes ten minutes and determines everything you do next. Skipping it is how teams end up rewriting perfectly good ad copy for a fortnight.
If it is tracking, the usual suspects, in the order they go wrong:
The conversion event is not the one the platform accepts. This bites hardest on newer platforms. ChatGPT Ads only recognises standard conversion events — a custom event will fire perfectly and still never count as a conversion.
Consent mode is blocking the tag. Conversions fire only for users who accepted, which in some European markets means you are measuring a third of reality. The data is not wrong, it is partial, and the gap is not evenly distributed.
The tag fires on a page nobody reaches. A thank-you page behind a redirect, or a single-page checkout where the conversion step never produces a page view.
Attribution windows are hiding the conversions. They happened, outside the window the platform is looking through. Longer cycles need revenue connected back through server-side tracking rather than a longer window.
Layer 1: do you have enough clicks to expect anything?
This is the layer people skip, and it produces the most wasted work. “Zero conversions” from a small number of clicks is not a finding. It is the expected outcome.
At a 2% conversion rate, the probability of seeing zero conversions:
| Clicks so far | Chance of zero conversions by luck alone |
|---|---|
| 30 | 55% |
| 100 | 13% |
| 300 | 0.2% |
Thirty clicks and nothing happened is a coin flip. It is not evidence of anything, and any conclusion you draw from it is noise dressed as insight. A hundred clicks starts to mean something. Three hundred clicks with zero conversions is real evidence that something is broken.
Note that the table is in clicks, not impressions. If you have 5,000 impressions and a 0.6% CTR, you have thirty clicks, and you are in the top row no matter how large the impression number looks. Impression counts are the reason this problem feels more urgent than it usually is — and on impression-billed platforms, they accumulate whether anyone engages or not.
Layer 2: impressions but barely any clicks
If CTR is far below the benchmark for the placement, the problem is upstream of your landing page and no amount of CRO will reach it. Three causes:
Intent mismatch. The ad is answering a question the audience did not ask. On search this is usually broad match pulling in queries adjacent to yours but not yours — the lead-quality problem in its earliest form, covered in lead quality signals for Google Ads.
Placement expansion you did not choose. Display expansion, audience expansion, and Performance Max placements that put a considered B2B offer in a mobile game’s reward slot. The impressions are real; the attention is not. When Performance Max works and when it does not is mostly a question about this.
Creative fatigue. If CTR started healthy and decayed, nothing is broken — the audience has simply seen it. That is a refresh cadence question, not a diagnosis.
Layer 3: clicks arrive and nothing happens
Now the landing page is genuinely in scope. In rough order of how often it is the culprit:
The page does not continue the ad. The ad promised a price, a product or an outcome; the page opens on a generic homepage. Every click costs you the same whether or not the page keeps the promise, and a message mismatch is the most expensive cheap problem in paid media.
The page is slow on the device the traffic arrives on. Check mobile specifically, on a real connection. Paid traffic is disproportionately mobile and disproportionately impatient.
The form or checkout is losing people. This is where a heatmap and a session recording earn their keep: they show you the field people abandon, which is almost never the field you would have guessed.
The conversion asks for too much too early. A demo request with eleven fields converts a fraction of what four fields convert. Whether that trade is worth it depends on whether you need volume or qualification, and the answer differs sharply between ecommerce and SaaS.
Layer 4: the audience cannot buy
Impressions served to people who were never candidates. This shows up as a decent CTR with catastrophic conversion rates, which is the combination that confuses people most.
Common versions: geographic targeting including markets you do not ship to; a B2C audience on a B2B offer; retargeting pools padded with people who already bought; and on some placements, a meaningful share of invalid traffic. Check the geographic and placement breakdowns before concluding that your funnel is broken, because this looks identical to a funnel problem from the summary view.
Layer 5: the offer is the problem
Last, and the one nobody wants to reach. If tracking is clean, volume is sufficient, CTR is normal, the page is fast and relevant, and the audience is right — then the conversions are not happening because the offer is not competitive at that price against the alternative one click away.
No amount of bid strategy fixes this. It is genuinely useful to establish, because it moves the problem to the people who can actually solve it, and continuing to optimise media against it just spends money proving the same point more slowly.
The sequence, compressed
- Backend vs platform. Are conversions happening and not being recorded? Ten minutes.
- Click count. Do you have 300+ clicks? Below that, you are reading randomness.
- CTR vs benchmark. Far below means the problem is targeting or creative, not the page.
- Landing page on mobile. Does it continue the ad, and does it load?
- Geography and placement breakdown. Are these people candidates at all?
- The offer. Only after the five above come back clean.
Work it in that order and most cases resolve at step one or two. Work it backwards — which is the instinct, because the ad and the page are the parts you control most directly — and you can spend a month rewriting creative for an account whose conversion tag was never firing.
If you have been through this and the numbers still do not add up, send me the account and the raw figures — the tracking layer is usually where it comes apart.
