There are two ways to double your revenue. You can double your traffic — more ad spend, more content, more channels, more agencies. Or you can double the proportion of visitors who convert.
The first is what most companies try. It's also the one that gets more expensive every year, as acquisition costs rise and the same channels get more crowded. The second costs a fraction as much, compounds with everything you do afterwards, and most organisations have never seriously attempted it.
Conversion rate optimization is the systematic version of the second option.
What is conversion rate optimization?
Conversion rate optimization (CRO) is the practice of increasing the percentage of visitors who complete a desired action — purchase, signup, enquiry, subscription — through research, targeted changes, and measured validation.
The arithmetic is what makes it compelling. If 10,000 monthly visitors convert at 2%, you have 200 customers. Lifting conversion to 2.5% produces 250 customers from the same traffic, the same ad spend, and the same team. Achieving that through acquisition instead would require 2,500 additional visitors every month, indefinitely.
The improvement is also permanent in a way that traffic isn't. Stop paying for ads and traffic stops. A fixed checkout flow keeps converting.
CRO is not a single tactic. It is a repeating cycle: find where conversions are lost, understand why, change something specific, and verify the change worked.
What CRO is not
The field has a reputation problem, largely earned by a decade of bad practice.
It is not button colours. The green-versus-red button test is the most-cited example in CRO and one of the least representative. Real conversion problems are structural — an unnecessary step, an unexplained requirement, a form asking for information users aren't ready to give, a trust signal missing at the moment doubt appears.
It is not a checklist of tactics. Urgency banners, exit popups, and countdown timers are tactics, not strategy. Some work in some contexts. Applied without diagnosis, they add friction to products that were already converting fine and annoy users who would have bought anyway.
It is not dark patterns. Manufactured scarcity, hidden costs revealed at the last step, and deliberately confusing opt-outs can lift short-term conversion. They also increase refunds, churn, chargebacks, and complaints, and they're increasingly a regulatory exposure in the EU and UK. A conversion you obtained by misleading someone is a liability recorded as revenue.
It is not guessing with better vocabulary. If a change ships without a stated hypothesis and without measurement, running it under the heading "CRO" doesn't make it evidence-based.
Why acquisition is the expensive answer
Most organisations spend disproportionately on getting people to the site and comparatively little on what happens after they arrive. The imbalance is usually historical rather than deliberate — acquisition has clear owners, established budgets, and vendors actively selling it.
Three arguments for rebalancing:
- Acquisition costs rise; conversion improvements don't decay. Paid channels get more expensive as competition increases. A structural fix to your signup flow keeps working at the same cost forever.
- Conversion improvements multiply your existing spend. A 25% lift in conversion makes every acquisition channel 25% more efficient simultaneously. It doesn't compete with your marketing budget — it increases the return on all of it.
- You already paid for the traffic that's leaving. Every visitor who abandons mid-funnel was acquired at full cost and produced nothing. That is the most expensive traffic you have, and it's already on your site.
Where conversions actually leak
Losses are rarely concentrated in one place. Mapping the full path usually reveals several moderate leaks rather than a single dramatic one.
Landing and first impression. Whether the page confirms, within seconds, that the visitor is in the right place. Mismatch between ad promise and page content is one of the most common and most expensive leaks in paid acquisition.
Navigation and findability. Users who can't locate what they came for don't complain — they leave. This registers as a traffic quality problem when it's actually a structural one.
Product and pricing clarity. Unanswered questions become abandonment. Unclear pricing, ambiguous terms, and missing specifications each add hesitation at exactly the moment intent is highest.
Forms. Consistently the highest-yield area in CRO. Field count, validation timing, error message wording, required-versus-optional labelling, and input formatting all measurably affect completion, and all are cheap to change.
Checkout and payment. Where the most valuable users are lost — people who decided to buy and still didn't finish. Unexpected costs, forced account creation, and limited payment options are the recurring causes.
Trust and reassurance. Security signals, return policies, contact availability, and social proof, present at the moments doubt actually arises rather than clustered in a footer nobody reads.
Performance. Load time has a direct, well-documented relationship with abandonment. It is frequently excluded from CRO work because it looks like an engineering problem, and it is frequently the largest single lever available.
Mobile specifically. Mobile conversion rates typically run well below desktop, and the gap is usually addressable rather than inherent. Tap targets, form input types, and viewport behaviour are ordinary problems with ordinary fixes.
The CRO process
1. Establish the baseline
Before changing anything, quantify where value is being lost. Funnel analysis showing drop-off at each step, segmented by device, traffic source, and returning versus new. The output is a ranked list of leaks with a value attached to each.
2. Diagnose the cause
Analytics show *where*, not *why*. A 60% drop at checkout is a fact; the cause could be a surprise shipping cost, a payment method that isn't offered, a form that fails silently on mobile, or a trust problem. Diagnosis uses session recordings, heatmaps, form analytics, usability heuristics, support tickets, and — where the question justifies it — direct user testing.
This step is what separates CRO that works from CRO that runs tests at random.
3. Prioritise
Not every leak is worth fixing, and order matters. Ranking accounts for potential impact, traffic volume through the affected step, implementation cost, and confidence in the diagnosis. Frameworks like ICE or PIE formalise this; the point is that the sequence is defensible rather than driven by whoever asked most recently.
4. Hypothesise
Each change is stated as a prediction: because users abandon at X for reason Y, doing Z should increase completion by roughly N%. If a proposal can't be written this way, it usually can't be evaluated either.
5. Test or implement
Where traffic supports it, A/B test. Where it doesn't, implement and measure against baseline with proper controls. Both are valid; pretending the second is the first is not.
6. Validate and iterate
Confirm the effect, check for second-order consequences elsewhere in the funnel, and record the learning. Then move to the next item. The compounding comes from repetition, not from any single change.
The testing reality nobody mentions
A/B testing is the most rigorous validation method available, and it is unavailable to a large share of the businesses being sold it.
Detecting a realistic improvement — say, 2% conversion rising to 2.4% — requires thousands of conversions per variant to reach statistical significance. For a site converting 200 customers monthly, that's a test running for many months, by which point seasonality and other changes have contaminated the result.
Underpowered tests are worse than no tests. They produce a number, a winner, and false confidence — and teams then build strategy on noise.
Where volume is insufficient, the honest alternatives are:
- Before/after measurement over a sufficient window, controlling for seasonality and concurrent changes.
- Qualitative validation — moderated testing with five to eight users catches most serious problems and requires no statistical power at all.
- Changes with strong prior evidence, where published research already establishes direction and the remaining uncertainty is small.
- Testing larger changes, since bigger effects are detectable with smaller samples.
Any CRO provider unwilling to tell you your traffic is too low for meaningful testing is selling you tests rather than results.
Metrics that matter
Conversion rate by segment, not in aggregate. A flat overall rate frequently conceals mobile collapsing while desktop improves. Aggregate numbers hide the actionable detail.
Revenue per visitor. Guards against the classic failure where conversion rises and average order value falls further, producing a celebrated test and less money.
Micro-conversions. Add-to-cart, form starts, page depth. These reach significance faster than final conversion and give early signal on whether a change is working.
Post-conversion quality. Refunds, churn, support contacts. A change that increases signups while increasing cancellations proportionally has moved a number, not the business.
The full funnel, always. Improving one step while degrading the next is common and only visible if you look at the whole path.
Common mistakes
Testing without diagnosis. Running experiments on hunches produces mostly null results and burns months. Research first, then test — the hit rate difference is substantial.
Copying competitors. Their checkout was built for their users, their price point, and their traffic mix. It may also be underperforming; you have no way to know.
Calling tests early. Results fluctuate wildly in the first days. Stopping when the number looks good is how noise becomes strategy.
Ignoring qualitative signals. Analytics can't explain intent. Optimisation driven purely by numbers eventually starts degrading the experience in ways the metrics don't yet reflect.
Optimising a broken product. If the value proposition doesn't land or the product doesn't fit the market, CRO improves the rate at which people discover the mismatch. No funnel change fixes that.
One-off engagements. A single round of improvements produces a single lift. The returns come from an ongoing cycle where each round informs the next.
How we approach CRO
The distinguishing choice: we diagnose before we test.
Most CRO work is tactics-first — a backlog of experiments drawn from a general playbook, run in sequence, hoping something lands. It produces a lot of inconclusive tests and a slow, expensive learning rate.
We work from usability research instead. Before proposing a single change, we establish where users are actually failing and why, using behavioural data, session review, heuristic evaluation against established frameworks, and direct testing where the question warrants it. Changes proposed after diagnosis have a materially higher success rate than changes drawn from a playbook, because they address causes rather than symptoms.
We won't sell you tests you can't run. If your traffic doesn't support valid A/B testing, we'll say so and use methods appropriate to your actual volume. False rigour is worse than acknowledged uncertainty.
We don't use dark patterns. Not on principle alone — they measurably increase refunds, churn, and complaints, and they carry growing regulatory exposure in the EU and UK. Conversions obtained by misleading people are borrowed against future revenue.
We measure the whole funnel. Including the parts a change wasn't aimed at, and including what happens after conversion.
Engagements run as focused projects working through a prioritised set of improvements, or as ongoing collaboration where optimisation becomes continuous. Products that compound are almost always on the second model.
The approach comes out of thirteen years in UX, more than 50 in-depth audits and 300+ rapid design reviews since 2017 across insurtech, fintech, health, education, enterprise, and transportation, and certification from the Nielsen Norman Group — the institution that shaped modern UX practice — in information architecture, journey mapping, usability testing, UX leadership, and analytics.
Every engagement is run personally. No account layer between you and the person doing the work.
Frequently asked questions
What's a good conversion rate?
There isn't a universal benchmark worth using. Rates vary enormously by industry, price point, traffic source, and purchase complexity — an enterprise software trial and an impulse e-commerce purchase are not comparable. The meaningful benchmark is your own rate over time. Competitor comparisons are usually a distraction.
How long before we see results?
Quick wins — form fields, error messaging, unexpected costs surfaced earlier — often produce measurable effect within weeks. Structural changes take longer to implement and longer to read reliably. A reasonable expectation is early signal within a month and a clear picture across a quarter.
How is CRO different from a UX audit?
An audit is diagnosis: it identifies problems and recommends fixes, once, as a document. CRO is the ongoing cycle of implementing changes, measuring effect, and iterating. An audit is often the most efficient way to start CRO, because it produces the prioritised problem list the optimisation work then executes against.
Do we need a lot of traffic?
For A/B testing, yes — typically thousands of conversions per variant. For CRO more broadly, no. Research-led improvements, qualitative testing, and before/after measurement all work at lower volumes. Lower-traffic sites often see larger gains, because they've usually never had the funnel examined.
Can you work with our existing team?
Yes, and it's common. The external contribution is usually the diagnostic rigour and prioritisation framework; your team knows the product and implements faster than anyone external could.
What if conversion doesn't improve?
Then you've learned something specific about your users that you didn't know, and the next hypothesis is better informed. Not every change works — anyone claiming otherwise isn't measuring. What matters is the rate at which the cycle produces gains over time, not the outcome of any single test.
Where to start
The diagnostic question: do you know what percentage of visitors abandon at each step of your funnel, and what each of those steps is worth?
If you don't, that's the first thing to establish. Most organisations are surprised by the answer — and the surprise is usually where the money is.