Blog / Shopify & e-commerce /

Choosing a useful baseline before a storefront redesign

The redesign is ready to start and the team wants a clean before-and-after result. One small problem: nobody has written down what “before” actually includes. Meanwhile the store keeps moving, with campaigns, stock changes and ordinary catalogue work all carrying on.

A useful baseline keeps enough of the current experience and measurement to make later comparisons meaningful. It can’t freeze the business. It can stop the review relying on selective screenshots and whichever metric looks encouraging afterwards.

State what the redesign should improve

I’d choose the customer problems first. Perhaps product options are hard to distinguish, navigation hides useful categories or delivery information is difficult to find. Write the intended improvement in terms of a task someone can attempt.

Don’t define success as a new appearance alone. The visual direction can be approved separately, but the evaluation needs to know what should work better. That tells you what evidence to collect.

Include constraints such as mobile usability, accessibility and preservation of important content. A redesigned page should not gain a cleaner layout by quietly removing answers customers need.

Preserve representative journeys

Record the current route for a small set of meaningful tasks. I keep the list short, because a baseline nobody can repeat is worth nothing. Include the starting page, product state, device context and intended outcome. Save appropriate screenshots or notes that show the relevant issue.

Choose examples that expose the variety in the store: products with several options, pages with long content and journeys involving an error or unavailable item. The homepage alone is not a representative baseline for a whole shopping experience.

Keep the cases repeatable. Another person should be able to follow the same route after launch without asking the original reviewer what they meant. That makes the comparison less dependent on personal impressions.

Check measurement before collecting the comparison

Confirm which interactions and outcomes are reliably recorded. Review event definitions, key events and known gaps with the measurement owner. If the redesign introduces new collection, flag that discontinuity before using the resulting trend.

A newly measured interaction can appear to increase after launch simply because it was invisible before. Likewise, a missing event after release can look like a behavioural decline when the event itself stopped firing.

Keep a reference to the configuration used for the baseline. You don’t need every technical setting in the report, but someone should be able to say whether the before and after measures mean the same thing.

Capture the trading conditions

Note important promotions, pricing changes, availability constraints and campaign activity during the baseline period. Record the mix of acquisition channels and devices where it affects the question.

A hypothetical mobile campaign can change the audience before the redesign even ships. If the post-launch period contains a different kind of visitor, an overall rate may move without showing what the new navigation contributed.

Use a comparison period that suits the store’s operating rhythm. A short, unusual week is a weak reference for a normal busy period, and no magic number of days makes every evaluation reliable.

Keep business records and analytics distinct

Use the appropriate source for each question. Analytics can describe recorded visits and interactions, while order or support records may answer a different operational question. Do not assume those systems have identical scope or timing.

Where figures differ, check the definitions instead of picking the one that flatters the redesign. Differences often come from collection conditions, exclusions or how a record is counted.

Keep customer information out of the shared baseline unless you need it. Aggregated findings and anonymous task evidence are usually enough.

Agree how the review will be interpreted

Before launch, decide which evidence would support keeping, adjusting or reversing a change. A broken purchase control needs an immediate response; a modest movement in an unstable metric needs investigation.

Identify competing explanations the review must consider. Seasonality, merchandising changes and acquisition shifts do not become irrelevant because the redesign has a launch date. Include them in the discussion from the start.

Also record what the baseline cannot establish. If the existing setup cannot identify a particular task reliably, use a different source or accept that limitation. I’d rather show a gap than fill it with a guess. A blank beats false precision.

Store the baseline where it will be used

Keep the task cases, configuration reference, trading context and intended outcomes together in the release or evaluation record. Give the review an owner so the baseline does not become an abandoned preparation exercise.

Preserve known weaknesses honestly

A baseline should include inconvenient findings as well as the areas the redesign is intended to improve. If an existing form has an intermittent delivery problem, record it. Otherwise, a post-launch report may wrongly assign the pre-existing issue to the new interface.

Do not let that record become a reason to postpone a necessary correction. Address consequential faults through the normal process and annotate the change. The baseline can acknowledge that the starting state evolved while still preserving enough context for a useful review.

Include accepted limitations in the task evidence. Perhaps a particular route cannot be exercised safely in production, or a measurement gap prevents a meaningful comparison. State what alternative check will be used and what remains unknown.

That makes the evaluation more credible. A baseline is a description of the available evidence, not a promise that the business can be held still while the design team changes one variable.

Use the evidence to make the first redesign decision now. Can the team point to a specific customer problem in the current store and explain how the new version will address it? Start there before choosing the chart that will announce success later.