Research-Driven CRO: How we analyze Shopify shops in 7 steps

A low conversion rate is rarely the root cause – it's a symptom. Anyone wanting to systematically optimize a Shopify store first needs a clear understanding of the funnel: Where do users leave the store, and how does their behavior differ between mobile and desktop, and between new and returning customers? Only once these questions are answered can you meaningfully ask: Why is this happening – and what can be done about it?

At tante-e, conversion rate optimization is based on two building blocks: a quantitative analysis (funnel, traffic sources, landing page performance, PLP, PDP) and a qualitative analysis (heatmaps, scroll maps, on-site search, user surveys). The synthesis of these two results in 20–50 concrete, prioritized measures per shop – divided into quick wins, A/B test candidates, and strategic implementations.

This article provides an in-depth introduction to the topic of data-driven shop analysis and is based on the practical knowledge of our CRO expert Leonard.

Our approach at tante-e: Research Driven CRO

Research-Driven CRO means: We don't optimize based on gut feeling or best practices. We analyze systematically before making a single change.

The core of the approach can be summarized in one sentence: We do not analyze isolated pages. We analyze movement patterns.

That makes all the difference. Those who only look at one page see symptoms. Those who track the user's movement through the entire shop find the cause.

Two perspectives, one recommendation

Our analysis framework combines two levels:

level	Ask	method
Quantitatively	Where is the shop losing customers?	GA4, Shopify Analytics, funnel data
Qualitative	Why do users behave this way?	Heatmaps, session recordings, interviews

Only when both perspectives are combined can a reliable recommendation be made.

Overview of the analysis steps

We systematically work through six areas – from the first touchpoint to the purchase decision:

• Funnel – Where are we losing momentum?
• Traffic – Who is coming and with what behavior?
• Landing pages – Where do users enter the shop?
• PLP – Do you find suitable products?
• PDP – Can we safely activate them for purchase?
• Qualitative analysis – Why do they behave this way?
• Prioritization – based on the PXL framework

This path follows the actual customer journey and ensures that we identify the right causes before we start optimizing.

Step 1: The Funnel – where does the shop lose momentum?

Before we consider concrete optimization measures, we need a clear picture of where users drop off in the shop. The starting point for every analysis by our CRO expert Leonard is therefore the conversion funnel: a structured sequence of steps a visitor goes through before making a purchase.

The five funnel stages at a glance:

Level	What it measures
Session Start	All users who enter the shop
View Product	Users who visit a product page
Add to Cart	Users who add a product to their shopping cart
Begin Checkout	Users who start the checkout
Purchase	Users who actually buy

There is a transition between each of these stages – and at each transition, a shop loses some of its users. The crucial question is: Where is the loss disproportionately high?

How we evaluate the funnel

We primarily use GA4 and Shopify Analytics for this – both have different strengths:

• Shopify Analytics is well suited for AOV (Average Order Value) and direct conversion rate.
• GA4 is stronger at segmentation: Which channel performs how well? How does mobile compare to desktop? Which entry pages convert?

A concrete example from practice: The same landing page that achieves a view-product rate of 47% on desktop reaches 70% on mobile – a clear indication that mobile users arrive and navigate differently. Without this breakdown, this difference would remain invisible.

The funnel is the starting point. What we find there determines where we look next.

Step 2: Traffic analysis – not every low conversion rate is a UX problem

When a shop's conversion rate is low, the first impulse is often: We need to improve the pages. New design, better product images, clearer text. That can be right, but it doesn't have to be.

Because the conversion rate doesn't just depend on how good a shop looks. It also depends on who actually ends up in the shop.

What does that mean in concrete terms?

Here's the situation: You run an online shop. At the same time, you send a newsletter to your existing customers – and simultaneously, you're running a paid ad on Google. Both channels bring people to the same product page. But these people are completely different:

• Newsletter recipients are already familiar with the brand, may have ordered before, and click with a specific intention to buy.
• Google Ads users are seeing the brand for the first time and are still in the orientation phase.

The results from practical experience show how big this difference can be:

channel	Conversion Rate	Add-to-cart rate
Email (existing customers)	3.3%	22.38%
Paid Search (CPC)	1.12%	7.99%

The add-to-cart rate – the percentage of visitors who add a product to their shopping cart – is almost three times higher for email marketing than for paid traffic. This isn't because the site is better for email users, but simply because these users are already more ready to buy.

What does this mean for the analysis?

If you were to simply look at the overall conversion rate of the shop and say "it's too low, we need to optimize the shop", you would be starting in the wrong place.

Therefore, we always look at the traffic sources individually first before we think about page optimization. The key question is:

Do we have a problem with the site – or a problem with the traffic?

Specifically, we are examining:

• Which channels bring in the most visitors, and which bring in the most buyers?
• How do conversion rate and add-to-cart rate differ depending on the traffic source?
• Are there any channels that deliver a lot of traffic but perform far below the shop average?

Only when these questions are answered will we know whether an optimization problem lies on the page itself.

Step 3: Landing pages – not every entry page has the same task

When someone clicks on a link to the shop – whether in an email, an ad, or a Google search result – they land on a landing page. This is the first page a visitor sees. In principle, this can be any page in the shop – the homepage, a category page, a product page, or even a custom-created campaign page. The crucial question is: Which page does the visit originate from?

The problem arises when visitors are sent to a page that doesn't match the expectations they've built up from the link. For example, someone sees an ad for "vegan sneakers" and lands on the general sneaker category, which also includes non-vegan products – this creates confusion, and the visitor leaves.

What is the purpose of a landing page?

A landing page has precisely one task: to confirm the user's expectations and entice them to scroll further or make a purchase directly. Therefore, it's less about design and more about relevance.

Specifically, our data analyst Leonard asks the following questions about each entry page:

• Does the content match the source? Anyone who arrives via a Google ad for a specific product shouldn't have to navigate through a category first.
• Is the page optimized for the channel? A page for cold paid traffic needs different elements than a page for newsletter subscribers.
• Does the page answer the user's implicit question? The landing page should immediately fulfill the user's expectations.

What happens if that's not true?

The bounce rate is increasing – meaning more visitors are leaving the shop without even viewing another page. This is expensive because paid traffic is going nowhere.

The good news: Landing page problems are often easy to identify. Google Analytics 4 (GA4) shows precisely which entry pages have the highest bounce rate and which channels these visitors came from. This combination directly reveals where action is needed.

Step 4: Collection Pages (PLP) – Orientation determines relevance

A collection page – also called a PLP (Product Listing Page) – is the page where users browse products before clicking on an individual one. It may sound unassuming, but it's a critical point in the sales funnel: If users don't find what they're looking for here, they won't click any further.

What task must a collection page fulfill?

The main task of the PLP (Product Placement Platform) is to provide orientation. Users who land on a collection don't yet have a clear intention to buy a specific product – they are simply getting their bearings. The page must give them the feeling within seconds: I'm in the right place, I'll find what I'm looking for here.

That sounds simple, but in practice it's more complicated. Orientation isn't just a matter of design, but also of the relevance of the displayed products.

The key metric: Itemlist CTR

How well a collection page provides orientation can be measured directly. In Google Analytics 4, there is the Itemlist CTR metric for this: it measures how many users who viewed a collection clicked on a product.

Target value: ≥ 80%

If the CTR is below this level, a significant portion of users have left the page without viewing a product: a clear signal that the collection is not delivering what was expected.

Practical example: A shop that carries vegan products has a general category called "Body Care." Users who arrive at this page via a vegan search term initially see no products marked as vegan. The item list click-through rate (CTR) drops—not because the product range is incorrect, but because the page doesn't confirm the user's expectation quickly enough.

What are the most common problems?

Low item list CTR is usually caused by one of these three patterns:

problem	Description
Incorrect sorting	Products that would be suitable for beginners are too far down.
Missing filters	Users cannot quickly narrow down their search to their actual goal.
Unclear product description	Images or titles do not convey enough information at first glance.

How we proceed

In his analysis, our expert Leonard first examines which sources users use to access which collections. A collection primarily driven by paid ads has different expectations than one found organically via a long-tail search term. Sorting, filters, and image selection are then specifically tailored to this entry point.

Step 5: Product Pages (PDP) – Activation and Risk Reduction

The product detail page (PDP) is the moment when it's decided whether a user buys or abandons the purchase. Anyone landing here has already shown interest. The page now has a clear task: to convert that interest into a decision.

What happens on the PDP

User behavior changes on the PDP. Unlike on the collection page, where users are still browsing, here they are focused on a specific product. Two things happen simultaneously:

• Activation – the user wants to understand whether the product is right for them.
• Risk reduction – the user looks for reasons to justify their decision (size chart, return policy, reviews, material).

If either of these two elements is missing, the likelihood of a purchase decreases, even if the product itself is good.

The key metric: Begin-Checkout Rate (event-based)

The most important metric on the PDP is the event-based add-to-cart rate, i.e.: How many users who accessed the PDP added the product to their shopping cart?

Target value (event-based, PDP): 10–15%

If the rate is significantly lower, there is either a lack of activation or a lack of trust.

A practical example from Leonard: A Shopify store had an add-to-cart rate of only 1–2% for sneakers, while the store average was 4%. The initial assumption was poor image quality or a pricing issue. However, qualitative analysis (heatmap, session recording) revealed that the size chart was not visible on the page. Users looked for it, couldn't find it, and abandoned the purchase. A classic example of a risk mitigation vulnerability.

What we analyze on the PDP

Area	Guiding question
Product information	Is all purchase-relevant information available and easily accessible?
Trust signals	Are there any reviews, USPs, or return information?
Mobile Experience	Is the most important information immediately visible even on small screens?
Add-to-Cart button	Is it prominently located, clearly labeled, and accessible without distractions?

Why mobile is especially important

CRO expert Leonard points out a particularly crucial data point: Mobile users have a view-product rate of 70% – desktop users only 47% – on the same landing page. This means that the majority of relevant traffic arrives via mobile devices. A PDP that works well on desktop but is cluttered on mobile loses the majority of its users before a purchase decision can be made.

Step 6: Qualitative analysis – from the data to the cause

Quantitative analysis shows where a problem lies in the funnel. It doesn't explain why. This is precisely where qualitative analysis comes in.

What qualitative analysis means

Qualitative analysis means: We look at what users actually do on a page. The most important tools for this are:

• Heatmaps – showing what gets clicked and what gets ignored.
• Scroll Maps – show how far users scroll before they abandon the page.
• Session Recordings – actual recordings of user sessions that show how someone navigates through a page

The tools we use for this are Microsoft Clarity and Crazy Egg.

How we analyze qualitatively

We specifically examine the pages that stood out in the quantitative analysis – that is, pages with poor funnel values. Specifically, we ask the following questions:

• Where do users click when they shouldn't? (Note: confusing UI elements)
• How far do users scroll? (This indicates which content is actually noticed.)
• Where do users stop using session recordings? (Note the specific point of frustration)

These observations lead to hypotheses that prompt the appropriate action.

The difference to purely data-driven work

Quantitative data tells us we lose 60% of users on this page. Qualitative data tells us users get lost because they're looking for the delivery date and can't find it. Both perspectives together provide the complete picture and thus the basis for sound prioritization.

Step 7: From analysis to action – prioritization with the PXL Framework

After the analysis, you're usually faced with a long list of potential optimizations. The crucial question is: Where do we start? This is where the PXL framework comes in – a prioritization framework we adopted from CXL, which Leonard uses in his daily CRO work.

What the PXL Framework is

PXL is a structured scoring model that evaluates each hypothesis or measure based on several criteria and assigns it a score. The goal is to replace subjective gut decisions with a transparent, data-driven ranking.

Each measure is evaluated according to these criteria:

criterion	Ask
database	Is the hypothesis based on quantitative or qualitative data?
visibility	Does the change affect an area that many users see?
Traffic impact	How many users are potentially affected?
Expected uplift	How significant is the expected improvement?
Implementation effort	How much technical and time is required?

Each criterion is assigned points. The sum of these points gives the PXL score – and thus the priority.

From hypothesis to action

In practice, the process looks like this:

• Formulate a hypothesis – e.g., “The size chart is not visible on mobile devices, therefore some users abandon the process before adding items to their cart.”
• Calculate PXL score – data available? High traffic? Low effort?
• Derive a measure – prominently place the size chart on mobile devices.
• Prepare the test – depending on the score: direct rollout or controlled A/B test.

Scoring ensures that the team is talking about the same groundwork and that resources are deployed where they will have the greatest impact.

Conclusion: This is how CRO becomes a sustainable process on Shopify

Conversion rate optimization doesn't work as a one-off project. It works as a structured, continuous process.

What this process specifically means:

• Understand the funnel before optimizing. Those who start without a data foundation are solving the wrong problems.
• Each page has a purpose. Landing page, collection page, product page and checkout follow different logics – and must be measured and evaluated accordingly.
• Qualitative data complements what numbers cannot tell us. Session recordings and heatmaps reveal the why behind the what.
• Prioritization trumps short-term actions. The PXL framework ensures that resources are deployed where they will have the greatest impact.
• Learnings are just as valuable as winners. Those who internalize this principle optimize more sustainably than those who simply wait for positive test results.