The ad is the easy part. Copy, budget, audience, live. The question that comes next gets answered in passing in most small accounts: where does the click go? Usually to the homepage, because it is already there and because it shows everything the business can do. That is precisely the problem. Your ad made one single promise, and the homepage is the one page on your site that knows nothing about it.
What happens in the first seconds after the click?
Very little, and that is half the answer already. Someone arriving from an ad does not read, they check. And they check at the top. For its analysis published on 15 April 2018, the Nielsen Norman Group examined more than 130,000 eye fixations from 120 participants: 57 per cent of viewing time falls on the first screenful, 74 per cent on the first two, 81 per cent on the first three. The top fifth of the page alone accounts for more than 42 per cent of the time.
That does not mean nobody scrolls. It means the decision is taken at the top and the rest of the page only confirms or contradicts it. If the ad promises "payroll for businesses with up to 20 staff" and the first screenful of the destination says "Welcome to Muster Ltd, your partner since 1998", the visitor already has their answer. They will not scroll to hunt for your list of services. They go back and click the next result, and you paid for that first click.
Why is the homepage so rarely the right address?
Because it has a different job. The homepage receives existing customers, applicants, suppliers, the merely curious and prospects arriving for five different reasons. It is a junction, and a good junction distributes: navigation, six services, case studies, team, news. For the visitor who came from your ad, every one of those routes is an exit from the route you just paid for.
Google is remarkably direct about this in its own documentation. According to its explanation of Ad Rank, the quality assessment includes "the expectations users have about your landing page based on the clicked ad creative, ease of landing page navigation". So: the expectation your ad copy created, and how easy the destination is to use. Quality Score breaks that into three parts, expected click-through rate, ad relevance and landing page experience, and rates each of them against every other advertiser that appeared for the same searches over the past 90 days.
What follows from that is one sentence in Google's own text: "Higher quality ads can often lead to lower CPCs." So the wrong destination costs you twice. Once because fewer visitors enquire. And once because the same click gets more expensive than it is for the competitor whose page answers the question.
One exception matters: for searches on your company name, the homepage is the right answer. There the promise of the ad really is the company itself, and any specialist page would narrow it.
What does the wrong page cost?
Two data sets give the order of magnitude. For its Conversion Benchmark Report, Unbounce analysed more than 41,000 landing pages with 464 million visitors and 57 million conversions and arrived at a median of 6.6 per cent across all industries. Between the weakest and the strongest industry sits 3.8 against 12.3 per cent, more than a threefold gap.
The second set is WordStream's Google Ads benchmarks for 2026: 13,474 US search campaigns running between 1 April 2025 and 31 March 2026, median 5.42 dollars per click, 8.18 per cent conversion rate, 66.69 dollars per lead. Those are US figures from a different competitive environment and no forecast for your account. What transfers is the structure of the sum, not the amount.
And the structure is uncomfortably simple: your budget buys clicks, your destination page decides how many of them become enquiries. 200 clicks a month on a page converting at 6 per cent is twelve enquiries. The same 200 clicks on a page converting at 3 per cent is six. The budget is identical, the price per enquiry has doubled. The lever here is not the bid, it sits on the page. That worked example is judgement from our editorial work, not a measurement.
Homepage, service page or a dedicated landing page?
In practice this is not a choice between two options but between three. The third is the one most often overlooked: the service page you already own.
The service page you already have is often the most sensible address to start with. It is already narrow in subject, it may well rank organically, and it gets maintained anyway. What it usually lacks is three things: a headline that picks up the wording of the ad, one main button instead of three equally weighted offers, and a form where the decision is taken rather than at the very bottom.
A dedicated landing page earns its keep when three conditions meet: the offer is narrow enough for a page of its own, meaning one audience, one occasion, one next step. The campaign runs for longer than a few weeks. And enough clicks arrive to see anything at all. If one of the three is missing, you are building a second object that has to be maintained, kept current and eventually switched off, and you still measure nothing.
How do you see whether the page keeps the promise?
First through the one number that counts: how many visitors from this campaign take the next step? Split by source, not as a site-wide average. That takes UTM parameters on the ad links, otherwise paid visits mix with organic ones and the conversion rate describes nothing any more.
The second look goes at the page itself. A heatmap shows what actually gets clicked. The most interesting part is clicks on things that are not links at all: those are expectations your page raises and does not meet. Scroll depth shows how far the page is read, and therefore whether your form sits in a zone only a fifth of visitors ever reach.
Which leaves the awkward question: can you test properly at all? An A/B test needs more data than small accounts produce. At a baseline of 6.6 per cent, the Unbounce median, and an improvement of half to 9.9 per cent, you need roughly 1,100 visitors per variant to tell the difference from chance. For the more realistic improvement of a quarter it is roughly 4,000 per variant. At 100 clicks a week that test finishes in over a year, and by then the campaign has changed three times.
This is not an argument against measuring, it is an argument against the A/B test as a first step. Below that volume you decide on the argument, meaning the wording of the ad, one goal, a short form, and you measure the one page against its own past rather than two variants against each other.
Three levers for the next fortnight
1. Click your own ad and stop after five seconds. On a phone, not on a large screen. Is the answer to the promise of the ad inside what you can see right now? If you have to scroll for it, you already have your answer.
2. Put the wording of the ad into the headline of the destination. Not the sense of it, the words. It is the cheapest change in the whole account, it takes ten minutes, and it hits exactly the expectation Google names in its quality assessment.
3. Count the exits on your destination page. Every link that does not lead to the next step is one. Past three, the question is no longer whether the page looks good but whether it has a job at all.
If you would like to walk through your campaigns and the pages behind them with us, from the ad through the headline to the form, we are happy to do that. The biggest lever is usually not in the budget but in the first three lines of the page. 🎯