When a campaign underperforms, the first reflex is almost always aimed at the campaign: bids, keywords, copy. Yet an enormous share of the outcome is decided in the twenty seconds after the click, on a page nobody has touched in months and which, often enough, was never designed for paid traffic at all.
What Google actually assesses
Landing page experience is one of the three components of Quality Score, alongside expected CTR and ad relevance. In Google's own definition it covers the usefulness and relevance of the information provided, ease of navigation, the number of links on the page, and the expectations users formed from the ad they clicked.
That last part carries the most weight and is the part technical tools do not measure: consistency between what the ad promises and what the page shows on the first screen.
The five-second test. Open the page on a phone, look at it for five seconds, close it. Can you say what it sells, to whom, and what action is being asked for? If the answer is uncertain, no amount of bid work will recover what is lost in those five seconds.
Speed: the thresholds that matter
Perceived speed is measured by Core Web Vitals, and the reference thresholds are public:
| Metric | What it measures | Threshold |
|---|---|---|
| LCP — Largest Contentful Paint | When the main element of the page appears | within 2.5 seconds |
| INP — Interaction to Next Paint | Responsiveness to user interaction | within 200 milliseconds |
| CLS — Cumulative Layout Shift | Visual stability while loading | below 0.1 |
Two practical caveats. First: measure in the field, not only in the lab — real user data tells a different story from a test on a fast connection. Second: CLS is the metric teams most underrate, and it is the one that makes people mis-tap form fields, which is exactly where conversions are lost.
The three causes of slowness we find most often
- Unoptimised images served at full resolution and resized by the browser.
- Third-party scripts — chat widgets, heatmaps, pixels accumulated over years — blocking rendering. Inventory them: there is nearly always at least one nobody needs any more.
- Badly implemented consent banners, delaying the main content and stacking a speed problem on top of a measurement problem, as we describe in Consent Mode v2 and tracking.
The arithmetic of conversion
Worth writing out, because it makes an often-inverted priority obvious. At 1,000 clicks a month and a £4 CPC:
| Conversion rate | Leads per month | Cost per lead |
|---|---|---|
| 2% | 20 | £200 |
| 3% | 30 | £133 |
| 4% | 40 | £100 |
| 5% | 50 | £80 |
Going from 3% to 4% cuts cost per lead by 25% at unchanged spend. Achieving the same through bidding takes months and is not guaranteed. Which is why, before raising a budget, we always look at the page — the reasoning we set out in how much budget you actually need.
Fixes in order of return
- Match the message. The page headline has to carry the ad's promise, in the same words. Cheapest intervention, fastest effect.
- Cut form fields. Every non-essential field costs conversions. The test: is it needed to qualify right now, or can it be asked on the call?
- Make the action singular. A paid-traffic page with a full menu, a blog link, eight nav items and three different calls to action is asking the visitor to choose instead of to act.
- Put proof above the fold. Numbers, names, certifications, checkable reviews — next to the form, not at the bottom of the page.
- Design for the thumb. Most traffic is mobile: reachable buttons, fields that open the right keyboard, no pinch-zoom required.
- Then technical speed. It matters, but it is rarely the first bottleneck while the five points above are unaddressed.
One page per source, not one page for everything
A visitor who searched "solar panel installation quote" has already decided they want something: the page's job is to get them there in as few steps as possible. A visitor arriving from a YouTube Short was not looking for anything: they have to be convinced before they can be converted.
Using one page for both means failing one of them. It is one reason Demand Gen campaigns fail more often than they should: upper-funnel creative landing on a lower-funnel page.
Measuring without fooling yourself
- One test at a time, with enough volume to separate signal from noise. Below roughly a hundred conversions per variant, almost no test is conclusive.
- Watch quality, not only quantity. A shorter form produces more leads and sometimes worse ones: verify in the CRM, not in the form counter.
- Segment by device. An average improvement can hide a mobile regression offset by a desktop gain.
The landing page is the one part of the chain you fully control: it does not depend on an algorithm and it does not change because Google shipped something. Which is also why work done well here stays done.
If you want to know what your current conversion rate is costing you, the free audit puts a number on it against the traffic you are already buying. You can also see how we work and which services cover this ground.