Growth
CRO and Analytics: Conversion Rate Optimisation
Every effort you put into ads and content is either won or lost on your website. We get the measurement right first, then find where and why people drop off, and improve it through controlled tests.
Short answer
CRO and analytics means improving the rate at which a website turns visitors into customers, using hypotheses drawn from measurement data and controlled tests, and building the tagging and conversion tracking this depends on. UNIT İstanbul runs this work as standard across every performance account: measure accurately, change, then verify the impact with data.
What is CRO, and why can't it be separated from analytics?
Conversion rate optimisation (CRO) is the systematic work of improving pages, journeys and offers so that a larger share of the visitors who reach your site fill in a form, make an enquiry or buy. It rests not on guesswork but on hypotheses drawn from how users actually behave.
Analytics is the infrastructure that makes that behaviour visible. If the measurement is wrong, the test results will be wrong too: a sale counted twice, a form submission that goes unrecorded, or users who decline consent vanishing from view altogether can turn even the best-intentioned decision in the wrong direction. That is why we never start a CRO project with design suggestions; we start with a measurement audit.
For us this is not a separate line item. Unless the bidding algorithms in Google Ads and Meta Ads accounts are fed the right conversion signal, budget cannot be spent efficiently. Our advertising, SEO and content teams all work from the same measurement definitions.
How is a conversion defined, and why does it differ for every business?
A conversion is a measurable user action that carries value for your business. Its definition changes with the sector and the sales model: for an online shop it is a completed order, for a clinic a qualified appointment request, and for a B2B firm an enquiry the sales team considers worth a conversation.
In the measurement plan we write down the primary conversions and the intermediate steps that lead to them (micro-conversions) separately. Primary conversions feed advertising optimisation and business reporting; micro-conversions help us find the blockages in the funnel.
| Business model | Primary conversion | Examples of micro-conversions |
|---|---|---|
| E-commerce | An order that is completed and not returned | Add to basket, moving to checkout, using a voucher code |
| Healthcare and clinics | A qualified appointment or quote request | WhatsApp click, pricing enquiry form, viewing a doctor's profile |
| Real estate | A request for a meeting or site visit | Downloading a floor plan, long time on a project page, click to call |
| B2B services | A lead approved by the sales team | Downloading a catalogue, starting a demo form, visiting the pricing page |
A form submission on its own is not an adequate definition of a conversion. Wherever possible we connect CRM or sales data to the measurement, so you can see which ad and which page produced the enquiries that genuinely became customers.
The core principles of a GA4 and Google Tag Manager setup
A correct setup is one in which every important user action is recorded exactly once, under a consistent name and with meaningful parameters. GA4 is Google's event-based analytics tool; Google Tag Manager (GTM) is the tag management system that controls those events and your advertising tags without touching the site's code every time.
- Data layer: A shared structure that reliably passes information such as order value, product or form type from the site's code to the tags. We prefer this to fragile tags that depend on click selectors.
- Event naming dictionary: The name of every event, the moment it fires and the parameters it carries are written down in a single document, so the same action is never counted under different names in different channels.
- E-commerce events: Product view, add to basket, begin checkout and purchase events are set up so that duplicate order numbers are filtered out.
- Clean-up settings: Internal traffic and developer visits are filtered out, returns from the payment provider are prevented from overwriting the traffic source, and cross-domain tracking is configured where it is needed.
- Verification: Every tag is tested in preview and debug mode, and every change that goes live is recorded with a version note.
When is server-side tagging needed?
Server-side tagging is a set-up in which data is not sent straight from the browser to advertising and analytics platforms, but first to a server under your own control. It can recover some of the signal lost to browser restrictions and ad blockers, reduces the amount of third-party code loaded on the page, and lets you control centrally which data goes to which platform.
It is not the first step for every site, as it brings an extra server cost and maintenance burden. We recommend it for businesses with significant advertising spend, those sending conversions to more than one platform, and those that need to keep tight control over data sharing. The user's consent choices must be applied in exactly the same way within this set-up too.
How can you tell whether your GA4 setup is correct?
If the ad platform, GA4 and your sales records do not match, something is wrong with measurement. Signs you can check yourself:
- Do conversions match real actions? The number of forms, purchases or calls in GA4 should roughly match the records that reached your inbox or CRM in the same period.
- Is the same action counted twice? Conversions inflate when the thank-you page is reloaded or a tag is installed twice.
- Is your own traffic filtered out? Office and agency visits and test transactions should be excluded from reports.
- Do sources look right? Paid ad visits showing up as "direct" or "referral" point to broken link tags or a payment page redirect.
- Does cookie consent work properly? If consent mode is missing or faulty, measurement is incomplete and may also breach regulations.
If any of these signs appears, measurement should be fixed before budget decisions are made; otherwise every optimisation amplifies the wrong data.
How do you build measurement that complies with KVKK and cookie consent?
Compliant measurement respects the user's cookie choice, works consistently with your privacy notice and never collects personal data it does not need. Analytics and advertising cookies should not run until the user has given explicit consent, which is why the link between the consent management tool and the tags is the most critical part of the setup. KVKK is Türkiye's personal data protection law.
- The consent management platform is integrated with Google's Consent Mode; how the tags behave when consent is refused is tested and documented.
- Information from form fields, such as names, phone numbers and email addresses, is never sent to analytics tools in raw form.
- In sectors that handle special categories of personal data, such as healthcare, event and parameter design is reviewed separately so that no sensitive information leaks out, even through a page name or URL.
- The content of the privacy notice and cookie policy is the responsibility of your legal adviser; we make sure the technical setup works consistently with those texts.
How does funnel analysis show where users give up?
Funnel analysis follows, in order, the steps a user takes towards a conversion and shows the step at which there is a clear drop-off. Where the drop-off happens usually tells you where to look for the problem: is trust lacking on the product page, does a surprise cost appear in the basket, or is there an unnecessary field in the form?
We also break the funnel down by device, traffic source and new versus returning visitors. A checkout step that works perfectly on desktop can lose users on mobile, and an average across all traffic hides that difference.
What do heatmaps and session recordings show, and what don't they?
A heatmap visualises where users click on a page, how far they scroll and where they pause. Session recordings let you watch anonymised replays of individual visits. Both are powerful tools for answering the question of "why" behind a problem the funnel analysis has uncovered.
These tools generate hypotheses, however, not proof. To avoid mistaking behaviour seen in a handful of recordings for a general conclusion, we compare the findings against quantitative data, and we use the tools with form fields and personal information masked from the recordings.
How is an A/B test run properly?
An A/B test is a controlled experiment in which traffic is split at random into two groups: one sees the current page, the other a modified version, and the difference is assessed statistically. Set up badly, it is all too easy to declare the wrong winner.
- Every test begins with a written hypothesis: what data did we see, what will we change, and which behaviour do we expect to change?
- A single primary success metric is chosen before the test starts; we do not go looking afterwards for "whichever metric came out well".
- The required sample size and test duration are calculated in advance, and the test runs long enough to cover differences in behaviour across the days of the week.
- Results are not checked constantly and the test is not stopped early; that habit is the most common way of mistaking random fluctuation for success.
- Winners and losers are both recorded; a losing test is still something learned about your users.
| Test type | When it is appropriate | What to watch out for |
|---|---|---|
| A/B test | Measuring the effect of a single significant change | The change should be meaningful; small colour tweaks rarely produce a result |
| Multivariate test | Seeing the combined effect of several elements | Requires very high traffic; unsuitable for most sites |
| Split URL test | Trying out an entirely different page or journey | Canonical and redirect settings must not harm search visibility |
On sites without enough traffic for testing, we do not set up artificial tests. Instead we implement improvements based on usability findings and established best practice, and interpret the before-and-after data with care.
Reporting dashboards: the right view for each reader
A good dashboard answers, at a glance, the question its reader needs to make a decision on. A managing director and a campaign specialist look at the same data at different depths, so rather than one crowded report we prepare separate views for each audience.
- Management view: Total enquiries or sales, customer acquisition cost and the contribution of each channel.
- Channel view: Conversions and cost by campaign, creative and landing page.
- Site view: Funnel steps, device breakdown, speed indicators and the status of tests in progress.
Dashboards are connected to live data using tools such as Looker Studio, and the definition of every metric is written inside the dashboard itself. We have explained the logic of reading return on investment channel by channel in our article on ROI-focused performance marketing.
What is the difference between CRO and a website rebuild?
CRO improves the existing site step by step, guided by data; a rebuild sets up the infrastructure and design from scratch. If the problems are concentrated in particular journeys, CRO is the faster and lower-risk route. Where the infrastructure genuinely stands in the way on speed, security or measurement, we recommend a rebuild together with our web and e-commerce team, and in that case the new site's measurement plan is prepared before the design.
Why is GA4 setup the first step of conversion optimisation?
Conversion optimisation only works with correctly measured data. A GA4 setup covers defining events and conversions, organising tags with Google Tag Manager, consent settings and the link to ad platforms. If the setup is incomplete, A/B test results are misleading too. That is why our conversion rate optimisation projects start with a measurement audit; how this feeds paid media is described on our Google Ads page.
How we work
Measurement audit
We review your existing GA4, Tag Manager, advertising conversions and cookie consent setup, and document any events that are missing, double-counted or faulty.
Measurement plan and setup
We define primary and micro-conversions, the event dictionary and the data layer, then implement and verify the tags in line with users' consent choices.
Funnel and behaviour analysis
Using funnel breakdowns, heatmaps, session recordings and form analysis, we uncover where users give up and why.
Hypotheses and prioritisation
We turn the findings into testable hypotheses and rank them by expected impact, difficulty of implementation and whether there is enough traffic to test them.
Testing and implementation
We run A/B tests against a success metric and duration agreed in advance, and roll out the winning changes permanently.
Reporting and the next cycle
We keep the dashboards up to date, share each period's test results and lessons, and agree the next list of hypotheses with you.
What we deliver
- Measurement audit report and list of fixes
- Measurement plan covering primary and micro-conversion definitions
- Event naming dictionary and data layer documentation
- GA4 and Google Tag Manager setup with verification
- Cookie consent and Google Consent Mode integration
- Server-side tagging setup where needed
- Funnel, heatmap and session recording analysis report
- Prioritised testing roadmap and test result log
- Live reporting dashboards tailored to each audience
Get a quote
How does the quote process work?
It starts with a message. We prepare the rest, and no work begins until you have seen in writing what you pay for, why, and how much.
Message us
Tell us briefly about your business, your goal and your website, on WhatsApp or by email.
Free initial analysis
We review your search visibility, how AI answers mention you and any ad accounts you run, and prepare a one-page summary.
Strategy call
We go through the summary together and agree on priorities, goals and the metrics we will track.
Written proposal
We send a proposal that spells out scope, deliverables, timeline and fee. We start once you approve it.
Request a quote
Fill in the form; we will review your goal and current position and come back to you with a written proposal.
Service:CRO & Analytics
Frequently Asked Questions
- How much traffic do you need for conversion rate optimisation?
- Analysis and improvement can be done at any scale, but a statistically reliable A/B test needs enough traffic and conversion volume. On sites with limited traffic, instead of testing we implement improvements based on usability findings and interpret the before-and-after data carefully. We tell you plainly which approach suits you once the measurement audit is complete.
- What happened to our Universal Analytics data, and does that mean our move to GA4 is finished?
- Google has shut down the old Universal Analytics properties, and measurement now runs through GA4. On many sites, however, the migration stopped at adding the basic tag: conversions, e-commerce events and Consent Mode were left incomplete. In our audit we list each of these gaps and close them one by one.
- Do users who refuse cookie consent disappear from the measurement?
- Analytics and advertising cookies do not run for users who have not given consent; that is a legal requirement. Google's Consent Mode allows aggregate modelling from cookieless, identifier-free signals when consent is absent. The gap does not disappear entirely, but measurement is preserved as far as possible while respecting the user's choice.
- Does every site need server-side tagging?
- No. Because it adds server costs and a maintenance burden, we recommend it for businesses with significant advertising spend, those sending conversions to more than one platform, or those that need tight control over data sharing. For smaller sites, a well-configured Tag Manager setup is usually sufficient.
- Does A/B testing harm SEO?
- Not when it is set up correctly. Google advises that test variants should not behave as if they were permanent, that split URL tests should use a canonical tag, and that tests should not run longer than necessary. Showing search engines different content from users is prohibited; we design our tests within these rules and end them as soon as they finish.
- Do heatmaps and session recordings collect personal data?
- They can if configured badly. We mask form fields, personal information and sensitive pages from recordings, and run these tools only when the user has given consent. In sectors handling special categories of data, such as healthcare, we decide together with your legal adviser which pages should never be recorded at all.
- What do you commit to in CRO work?
- We commit to accurate measurement, a disciplined testing method and transparent reporting of every result. Every test starts with a hypothesis, a primary metric and a decision criterion written down in advance. Tests that do not show the expected effect are a valuable lesson too; the findings set the priorities for the next tests.
- We use a different agency for advertising. Can you set up our analytics?
- Yes. We document the measurement plan and the tag setup so that every team can use them, and keep all access within your own accounts. That way, whichever agency you work with, the conversion definitions and ownership of the data stay with you.
- How do CRO and SEO affect each other?
- SEO brings visitors with the right intent to the site, and CRO makes it easier for them to take action. Speed, a clear page structure and content that answers the question benefit both. Conflicts usually arise over decisions such as shortening copy, hiding content in tabs or merging pages. We measure every change for its effect on both conversions and organic traffic.
Let us measure
your visibility today.
We map your current search visibility and your standing inside generative engines. Free, one page, real data.
