Conversion Rate Optimization

Google retired its own free testing tool in 2023, and most sites have not run a controlled test since. Ivan Sp runs conversion work built on audits, observed behavior, and a custom A/B testing plugin that has actually produced a measured lift in demo bookings. Many consultants sell CRO and deliver a UX audit with no testing behind it. That gap is exactly why the testing tool exists.

Four Areas This Practice Covers

Why the A/B Testing Plugin Exists

Most CRO practices stop at the audit stage, since building or licensing a reliable testing setup is real technical work most consultancies are not equipped to do. The custom plugin closes that gap here, with the mechanics and the measured result described in full above.

Where the Standard CRO Advice Breaks Down

The usual advice: Start A/B testing. Nothing else in conversion work produces evidence you can actually trust.
What holds up in practice: A test compares two ideas and never supplies one. On most sites a trust audit and five usability sessions surface more in a fortnight than a quarter of testing button colours.

Imagine a team that commits to a testing programme and runs six tests in five months. Four are inconclusive, one wins by a small margin, and one loses.

In the same period nobody watched a session recording. Four of five users could not tell whether the demo was free, and no test on the page ever asked that question.

I run the cheap evidence first and let it write the hypothesis. Testing is how a good idea gets confirmed, and it is a poor way to find one.

A testing programme with no research feeding it consumes a quarter and produces inconclusive results. Justifying a second quarter on that record is difficult.

Four Things Commonly Assumed About Conversion Work

When did anyone last watch a real visitor try to use your most important page?

I meet these four assumptions on most first conversion calls.

Commonly assumed	What actually holds
A/B testing is where conversion work starts.	A test compares two ideas and never supplies one. Audits and usability sessions are what produce a hypothesis worth testing.
More traffic fixes a low conversion rate.	More traffic through the same page produces proportionally the same result. The rate is a property of the page.
Proven patterns can be copied from other sites.	A pattern that works elsewhere was tested against a different audience and offer. Copying it is a hypothesis rather than a fix.
Behaviour tools are expensive.	Microsoft Clarity provides heatmaps and session recordings at no cost, which removes the budget objection entirely.
Conversion work is about the button.	Clarity of the offer usually outranks any single element. A visitor who cannot tell what is being offered will not click anything.

Two weeks of free heatmap data will usually settle an argument your team has been having for a year. The only cost is finding out you were wrong.

What Each Method Answers

Trust audit: Whether a page gives a stranger enough reason to act. Qualitative, and fast.
Heatmaps and recordings: What visitors actually do at scale, without explaining why.
Usability testing: Why a visitor hesitated, gathered by watching and asking a small number of people.
A/B testing: Which of two specific versions converts better, measured rather than argued.
Why a custom plugin exists: Google Optimize stopped serving experiments on 30 September 2023.
What a test needs: One changed element, a stable traffic split, and enough conversions to separate signal from noise.
The measurement surface: GA4 key events, which is what Google renamed conversions to in 2024.

A Worked Example: Four Methods on One Page

Consider a demo booking page converting below expectation. The sequence below is illustrative, written to show how the four methods stack. No client engagement is described here.

Stage	Method	What it produced
Week 1	Trust audit	Three missing signals: no named integrations, no logos, no security statement
Week 2 to 3	Heatmaps and scroll maps	62 percent of visitors never reached the form
Week 3	Usability sessions with 5 people	Four could not tell whether the demo was free
Week 4 to 7	A/B test on the headline	Version B, naming the price as free, measured higher

Ivan Sp runs the cheap methods first. The audit and the heatmap cost days, and they tell the test what to change.

Testing without the earlier stages is guessing with better arithmetic. A test proves which of two guesses is less wrong, and it never suggests the idea.

The usability row is the one most often skipped. Five people who cannot tell whether something is free will explain a scroll map that otherwise looks like a design problem.

How the Four Pieces Work Together

I sequence the four deliberately. An audit identifies likely conversion problems on a page. Heatmaps and session recordings show how visitors actually behave, confirming or contradicting the audit’s assumptions. Usability testing explains the reasoning behind that behavior. A/B testing then confirms whether a specific fix actually moves the number, rather than just looking like it should.

Skipping straight to testing without the diagnostic work first usually means testing the wrong variable. Skipping testing entirely after a diagnosis means never confirming the fix worked.

How a CRO Engagement Runs

I work from cheapest evidence to most expensive, since each stage narrows the next.

Confirm GA4 key events are firing correctly, since a test measured against broken tracking proves nothing.
Run a trust audit against the page, checking security statements, named integrations, testimonials, and logos.
Install a heatmap and session recording tool, and let it collect at least two weeks of data.
Read the scroll map first, and identify the depth at which most visitors stop.
Watch at least ten session recordings covering both converting and abandoning visits.
Run think-aloud sessions with roughly five participants, which surfaces most usability problems.
Write one hypothesis naming a single element, the change, and the expected direction.
Split traffic between two live published versions, changing one element only.
Run until the conversion count is large enough to separate a real difference from ordinary variance.
Implement the winner, then return to the scroll map to choose the next element.

How CRO Work Is Scoped

I scope conversion work against traffic volume, since traffic decides what can be tested at all.

What arrives: A trust audit, a behaviour report, usability findings, and a prioritised test queue.
Typical duration: Six to ten weeks for the full sequence on one page or template.
What the client supplies: GA4 access, site access for tag installation, and participants or a recruitment budget.
Included: Tracking verification, audit, heatmap setup, usability sessions, hypothesis writing, and test execution.
Not included: Design and development of the winning variant, and any promise of a specific lift.
What changes the estimate: Low traffic, which lengthens every test, and pages behind a login.
How it is quoted: After a short scoping call, once traffic and conversion volume are known.

Tools and Primary Sources

The references below cover the measurement surfaces and the research method. Each opens in a new tab.

Google Optimize sunset notice confirms the date Google stopped serving its own experiments.
GA4 key events documents the event type a conversion test is measured against.
Microsoft Clarity provides heatmaps and session recordings at no cost, which removes the budget objection.
Hotjar combines heatmaps, recordings, and on-page survey prompts in one product.
Nielsen Norman Group on test participants publishes the reasoning behind testing with a small number of participants per round.

Find Your Own Drop-Off Point

I suggest a free check that needs one afternoon and no budget.

Create a free Microsoft Clarity account and install the tag on one important page.
Wait two weeks, so the sample covers a full weekly cycle rather than one busy day.
Open the scroll map, and note the depth where half your visitors have stopped.
Check whether your main call to action sits above or below that depth.

A call to action below the halfway scroll depth is seen by fewer than half your visitors. Moving it is usually cheaper than testing the words inside it.

Common Questions About Conversion Rate Optimization

Do I need a full CRO engagement, or just an audit?

An audit alone often surfaces enough to act on for a smaller site or a single underperforming page. A fuller engagement, including testing, suits a page carrying enough traffic to reach a statistically meaningful result within a reasonable timeframe.

How does CRO work relate to SEO?

The two are separate disciplines that share a page. SEO work earns the visit, and CRO work decides what happens once a visitor arrives. A page ranking well and converting poorly still underperforms its potential.

Can CRO work start before a site has meaningful traffic?

The audit and usability testing pieces can start early, since neither depends on volume. Heatmap data and A/B testing both need enough traffic to produce a reliable read, so those pieces wait until volume supports them.

How much traffic does a page need before A/B testing is worthwhile?

Conversions matter more than visitors. A page producing a handful of conversions a month needs many months to separate a real difference from ordinary variance. Audits and usability research return more at that volume.

Discuss a Conversion Rate Optimization Engagement

CRO work gets scoped around current traffic and what has already been tried. The first call is a free 30-minute consultation with no pitch attached.

Book a free 30-minute consultation