Conversion Rate Optimisation

Why this sits alongside SEO, not after it

SEO brings people to a page. Whether they do anything once they arrive is a separate discipline with separate evidence. We put them together because the same page data answers both questions, and because ranking a page that does not convert is an expensive way to be ignored.

If your site takes 2% of visitors to an enquiry, moving that to 3% is worth the same as a 50% traffic increase — and it usually costs a fraction as much to achieve.

Research before testing

Most conversion work fails because it starts with a test. Someone changes a button colour, the number moves, nobody knows why, and the next test contradicts the last one. Testing without a hypothesis is just noise with statistics attached.

We start by establishing where people are leaving and what they were trying to do:

• Funnel analysis — where the drop-offs actually are, by device and by traffic source, because those behave very differently
• Session review — watching real journeys through the pages that matter, not a sample of the homepage
• Form analytics — which field people abandon on, which is usually not the one anyone guesses
• Search query review — what people expected the page to be about when they clicked
• Speed and stability — measured on real devices, since layout shift costs more conversions than most copy decisions

Then structured testing

Each test starts as a written hypothesis: what we think is happening, what we expect to change, and what result would prove us wrong. That last part matters — a test that cannot fail is not a test.

We will also tell you when your traffic is too low to test reliably. Many UK service businesses do not have the volume for statistically meaningful A/B testing, and running one anyway produces confident nonsense. At those volumes the honest approach is qualitative research and considered changes, and we will say so rather than sell you a testing programme.

What we do

• Conversion audit across the journeys that lead to enquiries
• Analytics and event tracking configured to measure the right things
• Funnel and drop-off analysis segmented by device and source
• Form and checkout optimisation
• Landing page structure and messaging review
• Test design, execution and honest analysis
• Reporting against enquiries and revenue, not sessions

Measurement first

A surprising number of sites cannot answer "how many enquiries did we get last month, and where did they come from". Before optimising anything we make sure the measurement is trustworthy: goals that fire once, forms tracked correctly, spam and internal traffic excluded, and calls attributed if the phone is how people contact you.

It is unglamorous and it is the reason later work can be believed.

What we usually find

The specifics differ but the categories repeat across UK service and retail sites.

The form asks for too much

Every additional field costs completions. Phone number on a first enquiry is the most expensive one, because it signals a sales call. Fields that exist because a CRM requires them are worth challenging — the CRM can be updated later by the person who follows up.

The page answers a different question than the visitor asked

A visitor arriving from a search for a specific service lands on a page about the company. The gap between the query and the page is the most common cause of a high bounce rate, and no amount of button testing fixes it.

Mobile and desktop are different products

Aggregate conversion rate hides that mobile often converts at a third of desktop. Segment first: the problem is usually specific and mechanical — a form that will not scroll, a sticky header eating the viewport, a tap target too small.

Nobody knows what happens after the form

A site can be converting well while enquiries go to an unmonitored inbox. We have found this more than once, and it makes every conversion metric meaningless. Following one enquiry all the way through is part of the audit.

How a test actually runs

A worked example of the shape, not a promise of the result. Funnel data shows mobile quote requests dropping at the second step. Session review shows people reaching a date picker that is unusable on a small screen. The hypothesis is written down: replacing the picker with three preset options will increase completions, and if it does not, the problem is intent rather than interface.

The variant runs until it reaches the sample size agreed in advance — not until it looks good, which is how most tests produce results that fail to replicate. Then it is reported with the number, the confidence interval and a plain statement of whether the hypothesis held.

What we need from you

Analytics access, and a conversation with whoever handles enquiries. That second part is the one people skip and the one that produces the useful insight — the person answering the phone knows exactly which questions the website failed to answer.