Will this increase our conversion rate?

Nobody can promise that honestly, and a studio that does is guessing. What we can say is that it identifies the specific points where people are visibly abandoning your funnel, with the recording that proves it, and gives you a ranked plan. Whether a fix moves the number is then a measurement question, and we tell you how to measure it.

What access do you need?

Analytics, so we can see the funnel and where traffic comes from, and permission to add a lightweight session-recording script if one is not already installed. If you cannot grant access we can still audit heuristically from the outside, but you should know that halves the value — the recordings are the part that turns opinion into evidence.

How long before you can start analysing?

If recording is already running, immediately. If it has to be installed, useful heatmaps need roughly three days of traffic before they mean anything — a heatmap built on a handful of sessions is noise that looks authoritative. Low-traffic sites need longer, and we will say so rather than reading tea leaves.

Do you fix the problems as well?

The audit produces the diagnosis and a specific prescription for each issue — not improve this, but change this to that. Implementation is separate work, whether it is done by your team or by us. Keeping them separate is deliberate, because it means the recommendations are not shaped by what would be convenient to build.

We only have one landing page. Is this overkill?

No, it is the cheapest end of this service and often the highest return, because everything is concentrated in one place. A single page audit covers the same ground — recordings, scroll and click behaviour, the form, mobile, speed — without the funnel mapping that a multi-step checkout demands.

Why not just A and B test everything?

Because a valid result costs a fixed amount of traffic, which is what limits a site to a couple of tests a quarter, and because obvious defects should be fixed rather than tested. An audit tells you which things are broken and which are genuinely uncertain. The uncertain ones are what a testing programme is for.

UI/UX & Design

Finding where people leave, and what to fix first

Session recordings, heatmaps and a real funnel, turned into a ranked list — what to change next week, what to test rather than assume, and what to leave alone.

From: $500–$4,000
Typical time: 1–3 weeks
Usually bought by: E-commerce, SaaS

What is actually being bought here

The traffic arrives. The sales do not. Advertising costs keep rising, so every visitor is worth more than the last one, and the owner has a strong sense that something on the site is losing them without knowing where. That is the situation that leads to this purchase, and it is worth naming the arithmetic behind it — improving the share of visitors who convert makes every pound already being spent on acquisition work harder, without buying a single extra visitor. That is why an audit is frequently a better first spend than a redesign, and always a cheaper one.

The second thing being bought is a decision. Listing ten things you dislike about your own site takes a minute. What is hard is ranking them, separating the ones costing money from the ones costing nothing, and telling a real defect from a personal preference. A list of problems is easy to produce and close to worthless. The value is in the order.

The method is deliberately unglamorous — watch what people actually do, compare it against known principles, and separate what is demonstrated from what is assumed. Most of the findings will be ordinary. The hour spent watching real people fail to complete your form is nearly always the hour that changes someone's mind.

How the audit runs

Set up measurement and let it collect. Analytics reviewed for where traffic arrives, which pages carry it, and where it leaves. Session recording and heatmapping installed if it is not already there — Microsoft Clarity is free without a session limit, which means there is no reason for a site not to be running something. Then it needs about three days of real traffic before the data is worth reading. Half a day of work, then three days of waiting.
Build the actual funnel. Every step from arrival to completion, with the count surviving each one. This produces the single most useful picture in the report, because it converts a vague sense that sales are low into a specific step where most of the loss occurs, and it is where the rest of the effort gets aimed. Half a day to a day.
Watch recordings and read the heatmaps. Ten to fifteen sessions concentrated around the worst step, plus click maps, scroll depth and rage clicks — the signature of someone repeatedly hitting something that is not responding, which is one of the fastest routes to a real defect there is. Then the same journey walked personally, on a phone, on a deliberately poor connection. One to two days.
Run the heuristic pass and the form autopsy. The journey checked against established principles — clarity of the next action, cost and delivery visibility, trust signals where the anxiety actually occurs, error handling. Then every form field counted and each one questioned on whether it must be there at all, because a required field nobody wants to answer is one of the most reliable conversion killers in existence. Page speed is measured as part of this, since a slow page loses people before any design decision matters. One to three days.
Write the report as a plan, not an inventory. One page at the front with the three biggest opportunities. The funnel picture. Then each issue with its evidence — a screenshot or recording reference — what principle it breaks, the specific change to make, the effort involved and the likely gain. Sorted so low effort and high gain sits at the top. Finally, a separate section for the things that should be tested rather than simply changed. Two to four days.

Those last two sections are the report. Twenty ranked issues with the cheap wins first gets acted on; twenty unranked issues gets filed. And the test column is what keeps the document honest — anything we believe but cannot demonstrate belongs there, explicitly labelled as a hypothesis.

What changes the price

How much journey there is. A single landing page with one form is a contained piece of work. A full commerce site — category pages, product pages, basket, a multi-step checkout, an account creation gate — is many funnels, each with its own drop-off, and the worst offender is usually somewhere nobody was looking.

Whether people are involved. An audit built on analytics, recordings and heuristics is one thing. Adding moderated sessions where real users attempt tasks while talking through their reasoning is significantly more work and answers the question the data never can, which is why they are leaving rather than where.

Traffic volume, which sets what is knowable. A site with substantial traffic yields clear patterns in days. A low-traffic site needs a longer collection window and leans more on heuristic evaluation and qualitative sessions. We would rather extend the window than present conclusions drawn from thirty visits as though they were established.

How audits fail

A long unranked list gets delivered. Twenty problems, no priority, no effort estimate, no indication of which matters. You read it, feel flattened, and do nothing — so a piece of work that was entirely correct produces no change whatsoever. Finding problems on a website is the easy half of this job. Saying what to do on Monday is the half being paid for.

Best practice gets presented as diagnosis. Make the button red, move everything above the fold, shorten the page. There is no general law that says any of these improve results, and asserted without evidence from your specific site they are the auditor's taste dressed as expertise. Anything not supported by something observable belongs in the test column, phrased as a hypothesis, not in the recommendations.

Conclusions are drawn from too little, or from the wrong device. A heatmap built on forty sessions read as if it were a survey. A week of data that happened to include a promotion or a holiday. Or the entire audit conducted on a desktop over fast broadband while most of the traffic is on phones over mobile networks — where the layout, the form and the load time behave differently. Segmenting mobile from desktop before drawing any conclusion is not optional; frequently it is the finding.

See where the site stands before you brief anyone.

If the site is already live, start from what it scores rather than from a guess. The audit reports what it checked, what it found, and — the part most tools leave out — what it could not check, and why.

See the website audit Talk about this work

Questions people ask before they commit.

Will this increase our conversion rate?: Nobody can promise that honestly, and a studio that does is guessing. What we can say is that it identifies the specific points where people are visibly abandoning your funnel, with the recording that proves it, and gives you a ranked plan. Whether a fix moves the number is then a measurement question, and we tell you how to measure it.
What access do you need?: Analytics, so we can see the funnel and where traffic comes from, and permission to add a lightweight session-recording script if one is not already installed. If you cannot grant access we can still audit heuristically from the outside, but you should know that halves the value — the recordings are the part that turns opinion into evidence.
How long before you can start analysing?: If recording is already running, immediately. If it has to be installed, useful heatmaps need roughly three days of traffic before they mean anything — a heatmap built on a handful of sessions is noise that looks authoritative. Low-traffic sites need longer, and we will say so rather than reading tea leaves.
Do you fix the problems as well?: The audit produces the diagnosis and a specific prescription for each issue — not improve this, but change this to that. Implementation is separate work, whether it is done by your team or by us. Keeping them separate is deliberate, because it means the recommendations are not shaped by what would be convenient to build.
We only have one landing page. Is this overkill?: No, it is the cheapest end of this service and often the highest return, because everything is concentrated in one place. A single page audit covers the same ground — recordings, scroll and click behaviour, the form, mobile, speed — without the funnel mapping that a multi-step checkout demands.
Why not just A and B test everything?: Because a valid result costs a fixed amount of traffic, which is what limits a site to a couple of tests a quarter, and because obvious defects should be fixed rather than tested. An audit tells you which things are broken and which are genuinely uncertain. The uncertain ones are what a testing programme is for.

Often taken together.

Start with the brief, not the invoice.

Send what you are trying to achieve. You get a written scope, a price and a date — or an honest “this is not the right service for that”.

Talk about Conversion rate optimisation audit All UI/UX & Design services