Conversion rate optimisation: a practical guide for established UK businesses

Most marketing budgets are spent on getting people to the website. Search, advertising, content, the trade press, the email list — all of it exists to bring a visitor to a page. Then the visitor lands, and a remarkable number of them leave without doing the one thing the page was built for. Conversion rate optimisation is the work of closing that gap. It is unglamorous, it compounds, and it is almost always the cheapest growth available to an established firm — because it acts on traffic you have already bought and paid for.

This guide is what we tell our own clients when they ask what CRO is and whether it is worth doing. It covers what a conversion is, the small amount of maths you need, how to tell a real result from a lucky one, the experiments worth starting with, and the process that holds it all together. It assumes you run a real business with a sales team, not an app with a growth squad.

What conversion rate optimisation is

Conversion rate optimisation — CRO — is the process of increasing the percentage of your website visitors who complete a specific goal. Not getting more visitors; getting more of the visitors you already have to act. If a thousand people reach your contact page this month and twenty enquire, your conversion rate is two per cent. CRO is the discipline of moving that number, deliberately and measurably, by changing what is on the page and proving the change worked.

It helps to think of the website as the last stretch of a journey you have already invested in heavily. You have done the marketing, earned the visit, won the click. CRO is what makes sure that effort pays off at the point of action rather than leaking away on a confusing page or a form nobody answers. Two firms can spend identically on getting traffic and see completely different returns, purely because one of them has done the work at the moment of conversion and the other has not.

There is a second benefit, quieter but real. To run CRO well you have to study how people actually use your site — where they hesitate, what they ignore, what they click by mistake. That investigation tends to teach you more about your buyers than any amount of demographic data, and the lessons feed back into everything else you do.

Why it matters more than more traffic

The instinct, when a website underperforms, is to want more traffic. More visitors, more enquiries — it feels like the obvious lever. But traffic is the expensive lever, and often the one with the least headroom. Doubling your qualified traffic might mean doubling your advertising spend or waiting two years for an organic search programme to mature. Doubling your conversion rate, by contrast, costs the price of a few experiments and reaches every visitor you already have, including the ones who cost you the most to acquire.

This is the argument we make on our conversion rate optimisation service page, and it holds up across nearly every client we audit: the conversion rate is the most neglected number in the business. Marketing teams obsess over the top of the funnel because that is where the dashboards point. Meanwhile the page that actually turns interest into enquiry has not been touched in three years, and nobody can tell you what its conversion rate even is.

There is an honest caveat. CRO needs traffic to work — you cannot optimise a page that almost nobody visits, because there is nothing to measure. If the problem is that the right people never arrive, then conversion work is premature and the answer sits upstream, in visibility. We wrote about how to tell those two problems apart in why your website isn't generating leads, and it is worth reading before you decide CRO is your priority. For most established firms with a steady flow of visitors, though, the conversion gap is real money left on the table.

What counts as a conversion

A conversion is simply a visitor completing a goal that matters to the business. The goal varies by what you do, but it is always something with commercial value attached:

• Submitting an enquiry or contact form
• Requesting a quote, specification or capability document
• Booking a call or a demo
• Completing a purchase, for those who sell online
• Signing up to a newsletter or downloading a guide
• Reaching a particular page that signals real intent

The mistake we see most often is treating every action as equally valuable, or — worse — not defining the goal at all and hoping the website "does its job." For a considered B2B purchase, the conversion that matters is usually a pipeline-eligible enquiry: a named person, from a real company, with a problem you can solve. Newsletter sign-ups and downloads are softer signals, useful as steps along the way, but you should never confuse a busy top-of-funnel metric with the thing that pays the bills. Decide which conversion actually feeds the sales pipeline, and make that the number you are trying to move. Getting this measurement right depends on having your analytics set up properly in the first place — a default install will not reliably tell you which page produced which enquiry.

The maths, and why sample size decides everything

The calculation itself is trivial. Conversion rate is conversions divided by visitors, expressed as a percentage. If 5,000 people reach your basket and 250 buy, that is a five per cent conversion rate. If 40 people read a capability page and two enquire, that is five per cent as well — but the two numbers are not remotely as trustworthy as each other, and that difference is the single most important idea in this whole guide.

A "good" conversion rate is not a universal benchmark you can look up. It depends on your industry, the kind of conversion, the quality of your traffic and the price of what you sell. A good conversion rate is, practically, one that is better than yours was before. Chasing some published industry average is a distraction; beating your own baseline is the only target that means anything.

What does matter universally is sample size. Small numbers lie. If only forty people have seen a page, one extra enquiry swings the rate from five per cent to seven and a half — a thirty per cent "improvement" that is pure noise. Low conversion volumes produce wild, unrepresentative figures and tempt you into changes based on nothing. The fix is patience: gather a meaningful number of visitors and conversions, and run any test for weeks rather than days, so that the result reflects how people genuinely behave rather than who happened to visit on Tuesday. This is also why CRO suits firms with steady traffic and frustrates those without it.

Statistical significance without the statistics degree

Once you start comparing two versions of a page — the current one and a new idea — you need a way to tell a real improvement from a fluke. That is what statistical significance gives you. You do not need to study statistics, and you certainly do not need to do the sums by hand; every serious CRO tool calculates this for you. But you should understand what the number is telling you, because it is the difference between a decision based on evidence and a decision based on a hunch wearing a percentage sign.

Statistical significance is just the likelihood that the difference you are seeing was caused by your change rather than by chance. There are two schools of thought on how to measure it, and your testing tool will use one of them.

The Frequentist approach is the conservative one. It starts by assuming there is no real difference between the original and the variant — the "null hypothesis" — and then asks how surprising your results would be if that assumption were true. It produces a confidence level, and the convention is that you need to reach at least 95 per cent confidence before you declare a winner. Below that, you cannot rule out that the difference was luck. It is binary and demanding: you either clear the bar or you do not, with little room for interpretation. If a test will not reach significance, the usual answer is not to abandon it but to give it more traffic and more time.

The Bayesian approach is more intuitive and is what many modern tools default to. Instead of a pass-or-fail verdict, it expresses a probability: "the variant has an 80 per cent chance of beating the original." It can reach a usable conclusion on smaller samples and faster, which suits a business that needs to make decisions at commercial speed. A purist statistician would call a Bayesian result an educated guess rather than proof — but for most business decisions, an educated guess backed by real visitor data is exactly what you want, and a great deal better than the alternative, which is the loudest person in the room deciding by taste.

You do not have to take sides between the two methods. What matters is that you never call a winner on a handful of visitors and a flattering-looking number. Significance is simply the discipline of waiting until the evidence is real.

The practical rule that falls out of all this: pick the tool, trust its significance calculation, run every test for at least a couple of weeks, and resist the powerful urge to stop early the moment your favourite version is ahead. Early leads reverse all the time. The discipline of waiting is most of the skill.

Seven experiments worth starting with

The hardest part of CRO is not the tooling; it is knowing what to test. Even the best platform will not tell you where your page loses people. Here are seven experiments that reliably earn their keep, and which are a sensible place to begin while you build the habit.

1. Your calls to action. The button that asks the visitor to act is the most leveraged element on the page. Test its wording, colour, size and position. A button that contrasts sharply with everything around it gets noticed; a vague label like "Submit" converts worse than a specific one like "Book a call" or "Get a quote." And place it where the visitor is actually ready to act — sometimes that is high on the page, sometimes it is at the end of the argument.

2. Social proof. People take their cues from other people. Customer reviews, named testimonials, the number of firms you have served, recognisable client logos — all of it lowers the hesitation a visitor feels before committing. The most persuasive proof is the kind your particular buyer recognises as being like them. We keep ours on the clients page, because for a considered purchase a credible example does more than any adjective.

3. Trust signals. Especially near the point of action — a checkout, a contact form — small marks of reassurance matter. Security badges, guarantees, accepted payment methods, accreditations and third-party endorsements all reduce the quiet anxiety that makes someone close the tab. For B2B firms this often means industry certifications and the names and faces of the people who would actually do the work.

4. Your unique selling points. What genuinely separates you from the competitor the buyer is also considering? Testing different ways of stating that — and how prominently — tells you what your audience actually values, which is frequently not what you assumed. Most established firms undersell their real advantages because they have stopped noticing them.

5. Urgency, used honestly. Genuine scarcity or a real deadline moves people who would otherwise defer. "Last few places," a closing date on an offer, a countdown to a launch. The caveat matters: manufactured, fake urgency erodes the trust you spent years building, and a sceptical buyer spots it instantly. Use it only where it is true.

6. Human faces. People are drawn to other people. A real photograph of your team — not a stock image of models — tends to make a page feel more trustworthy and more human. Eye contact builds connection; a face looking towards your call to action even helps direct attention to it. Authenticity is the whole point, so use your actual people.

7. Navigation. How your menu is structured and labelled shapes whether visitors find what they came for. Test the order of items, the words you use for them, and how much you show at once. A clearer hierarchy often lifts conversion across the entire site at once, because it helps everybody rather than one page.

None of these is exotic, and that is the point. The value is not in any single clever idea; it is in testing them properly, one variable at a time, and letting the data rather than opinion decide. Every test teaches you something about your buyer, even the ones that lose.

The process: how a CRO programme actually runs

A test or two is a tactic. CRO as a discipline is a repeating loop, run month after month, and the loop is what produces compounding results. The shape we use, and the one described on our optimisation service, runs roughly like this: study the data to find where visitors drop off; form a clear hypothesis about why and what would help; build the variant; run the experiment until it reaches significance; read the result honestly; and write up what you learned — win, loss or draw — so the next hypothesis is sharper than the last.

The honesty about losses matters more than it sounds. A test that fails is not a wasted test; it is a paid-for lesson about what your buyers do not respond to, and it stops you rolling out a change that would have quietly cost you money. Teams that only celebrate the wins tend to fool themselves, and tend to stop testing the moment a result flatters them.

The clearest illustration we can point to is the work Cohaesus did with Tate Gallery. With their physical shops closed during the busiest sales season of the year, Tate needed their online shop to carry more of the load, and quickly. A rapid programme of A/B experiments — testing urgency, surfacing the shop's distinctive selling points, smoothing the checkout — produced a more than ten per cent rise in completed orders over the holiday period and a five per cent lift in average order value. Just as importantly, Tate's own team came away trained to keep running the programme themselves. That is what a CRO programme looks like when it works: not one heroic redesign, but a series of small, evidenced changes that add up, plus a team that can keep the loop turning. Nine Stones is part of the same group as Cohaesus, which is how that experimentation discipline reaches the established firms we work with.

A note on where CRO sits. It is not a substitute for the rest of your marketing; it is the thing that makes the rest pay off. Good search brings the right people. Genuinely useful content earns their trust. CRO makes sure that, having arrived and been convinced, they actually act rather than drift away. The three reinforce each other, which is why we tend to run them together rather than in isolation.

Where to start

If you want to find out whether CRO is worth your attention, work through this in order:

1. Confirm you have the traffic. CRO needs a steady flow of visitors to a few key pages. If you do not have it, the priority is visibility first — start with why your website isn't generating leads to work out which problem you actually have.
2. Define the conversion that matters. Pick the one action that feeds real pipeline — usually a quality enquiry — and make it the number you are trying to move. Soft metrics can wait.
3. Get your measurement right. Make sure your analytics can reliably tell you your current conversion rate and which page produced which enquiry. You cannot improve what you cannot see.
4. Find where people leave. Use the data to identify the page or step that loses the most visitors. That is where your first experiment should live.
5. Run one honest test. Change one thing, run it for at least two weeks, and let significance — not impatience — decide. Then write down what you learned and do it again.

Done properly, CRO is the rare piece of marketing that gets cheaper and more effective the longer you do it, because every experiment narrows the next. The firms that win at it are not the ones with the cleverest single idea; they are the ones still testing in month twelve.

If you would like a sober second opinion on what your current conversion rate is, where you are losing people, and whether a structured programme is worth it for your firm, that is a conversation we are always glad to have, with no obligation attached. Thirty minutes, no pitch deck: hello@ninestones.co.uk.