Most UX redesigns get judged on taste. A stakeholder prefers the old button color, someone else misses the hero image, and the debate turns into opinion versus opinion. The work looks expensive because nobody agreed in advance what it was supposed to move. That changes the moment you fix the numbers in the brief and instrument them in the build. This is the discipline behind effective ux design: task completion, form and checkout abandonment, support contacts, revenue per session. These are not vanity metrics. They are the outcomes a redesign either earns or fails to earn.
This guide is for teams that want to judge a redesign on evidence rather than applause. You will learn how to choose the outcome metrics that matter, record a baseline with a target and a measurement window, instrument events during the build rather than after launch, and judge results against the plan instead of the loudest voice in the room. It also includes a measurement plan table you can paste directly into your brief and the traps that make a good redesign look like a failure.
At aarland.dk, a Danish design studio and Webflow partner working across UX/UI, webshops and CMS, we treat the measurement plan as part of the brief. Here is how to put proof before design starts.
Why a UX Redesign Gets Judged on Taste Until You Give It a Number
A UX redesign is a hypothesis about user behavior, not a deliverable. If no number is fixed before design starts, the only verdict at the end is aesthetic preference.
That is why UX design work reads as expensive. The cost appears on the invoice; the return stays invisible until someone writes down the task completion, abandonment, support-contact and revenue-per-session figures the work is meant to move. Usability is defined by whether specified users achieve specified goals effectively, not by how clean the interface looks. And "it feels cleaner and more modern" cannot survive a change of stakeholder, a budget review or a new marketing lead, because it is an opinion, not evidence. It rarely survives the brief stage either, as the brief that wastes your budget before work even starts shows.
Much vendor content on this topic prioritises conversion over method, which is why your team has to supply the measurement discipline itself.
The brief that fixes those numbers before the work begins is the document that makes the verdict possible.
What to Have in Place Before the Brief Is Signed
With the argument made, the preparation is unglamorous but decisive. Before the brief is signed, lock five things.
Trusted analytics access. Confirm analytics and ecommerce tracking are live, unfiltered, and reconciled against orders or finance. Do not baseline numbers you have not checked.
One named measurement owner. Give one person authority to sign off the metric list, targets, and window, so the plan survives handoff from business owner to development team.
A data baseline, not memory. Pull enough historical funnel data, typically several months at minimum, to cover seasonal variation, and annotate any sale, price change, or campaign inside that period.
A dated measurement window. Attach start and end dates, long enough to cover a full commercial cycle, not launch week.
Written brief inclusion. State the metric list in the same document that scopes the design work. A kickoff slide is not an agreement.
Pressure-test the setup with questions to ask any agency before you sign a UX brief. If brand work shares the funnel, apply the same rule when commissioning brand development services. With those five items confirmed, Step 1 can begin.
Step 1: Choose the Outcome Metrics the Redesign Is Meant to Move
With analytics reconciled and a named owner in place, the next task is to fix a short metric list in the brief. Focus on two or three money flows, finding a product, configuring an option, submitting an enquiry, and track completion as a rate, not clicks. Treat this as recommended instrumentation, since no neutral benchmark exists for your flows.
Define form and checkout abandonment before measuring it, a common formula divides incomplete transactions by transactions started, and apply the same logic to key forms. Abandonment usually signals unanswered questions at commitment, not missing intent. Add add-to-cart rate and product-detail-page conversion rate to locate the exact step that moved.
Include support contacts per 100 orders or sessions, read with reason codes. Repeated questions about one step are a design signal. Finally, track revenue per session, a commercially readable number that connects design decisions to business outcomes. This is design with a goal, not presence, as Formål kontra tilstedeværelse: designet med et mål explains. Cut every metric that cannot change a decision.
Step 2: Record the Baseline, Then Set a Target and a Window
You have your four to six metrics. The next job is to pin each one to a number before anyone opens a design tool.
Take the baseline from the instrument you will measure with. Pull it from the same analytics source, event definitions, and filters you will use after launch.
Record the date and the sample size. Write the baseline down with the day it was captured and how many sessions or orders sit behind it. A 62% completion rate drawn from 40 sessions is not a baseline, it is a coincidence. Treat that example as illustrative, not as a benchmark.
Set a target in business terms. Instead of "improve abandonment by 15%," write "reduce checkout abandonment by 4 points, worth roughly 60 additional orders per month at current traffic." Derive the order figure from your own data, never from a blog post.
Choose a window that covers a full commercial cycle. Annotate any sale, price change or marketing push inside it so those days can be excluded rather than explained away later.
Define failure in advance. State what result would count as a failure and what would trigger a second iteration. A plan with no failure condition will always be read as a success, which defeats the purpose of writing one.
Each of those targets then needs an event behind it. That is the build stage.
Step 3: Instrument the Events During the Build, Not After Launch
Convert each metric into an event definition: a name, a trigger, and the parameters it must carry. Hand that specification to the developer in the same sprint as the design work, not as a post-launch request. Treat this handoff with the same rigor you would apply when you interrogate the handoff and training process.
Before release, verify every event in staging. Each key step should fire once, carry the correct values, and appear in the reporting interface. Then check naming: keep event names and definitions consistent with pre-redesign tracking, because changed definitions break the historic comparison and make the baseline useless.
On launch day, annotate the analytics property with a link to the measurement plan. Months later, anyone reading the trend should know what changed and when. Make instrumentation part of the definition of done: a flow is finished when it can be measured, not merely when it works. Data lost in the first weeks after launch cannot be recovered.
Step 4: Judge the Result Against the Plan, Not Against the Room
With events verified in staging, the redesign finally has a scoreboard. Put the review date in the calendar before launch, set to the measurement window's end date, and invite everyone who signed the brief.
Open with the comparison the plan requires: baseline versus target, metric by metric. Then investigate. Report which funnel step moved and which did not, rather than flattening everything into one verdict. A checkout that improved while the enquiry form stayed flat tells you where to look next; "the redesign worked" tells you nothing.
Flat numbers are information, not failure. If conversion held steady while support contacts fell, the redesign still earned its budget. Say so in those terms.
Close the session with two decisions: what the next iteration will test, and which metrics to retire. Any metric that no longer drives a decision should leave the plan, so the document stays short enough that people actually read it.
The Measurement Plan Table You Can Paste Into the Brief
With the review date fixed, give the team the plan in a form they can use.
Metric | Definition or formula | Baseline | Target | Window | Event to instrument | Owner |
|---|---|---|---|---|---|---|
Money flow completion (2-3 flows) | Completed intended outcome / started flow | [value, date] | [target] | [start-end] | flow_start, flow_complete | [name] |
Checkout abandonment | (Carts created - completed purchases) / carts created | cart_create, purchase | ||||
Form abandonment | (Forms started - submitted) / forms started | form_start, form_submit | ||||
Add-to-cart rate | Add-to-cart / product views | product_view, add_to_cart | ||||
Product-detail-page conversion | Purchases / product-detail views | pdp_view, purchase | ||||
Support contacts | Contacts per 100 orders or sessions, with reason codes | support_ticket | ||||
Revenue per session | Revenue / sessions | purchase, session |
Rules: keep it to one page. Store it beside the brief in the same folder. Replace placeholders with real numbers as they arrive so targets are checked without renegotiation.
Traps That Make a Good Redesign Look Like a Failure
With that table in the brief, the review becomes a comparison, but only if you avoid five traps. Seasonality is first: a November launch faces holiday demand, a February launch does not. Compare like periods or state the limitation openly. Second, traffic-mix shifts. A new channel or campaign changes who visits, and a changed visitor mix can move conversion without the design doing anything. Third, moving too many metrics across too many flows at once destroys your ability to say which change moved which number. Keep the plan narrow. Fourth, separate vanity from decision metrics. Sessions, time on page and scroll depth are context, never proof. Fifth, borrowed benchmarks and opinions. Widely cited average cart-abandonment figures attributed to Baymard Institute research should be checked on Baymard's live page before they enter a brief. Claims about what shoppers do or want in any given year are characterisations, not measured data unless you can trace them to a primary source. A number you cannot source is worse than no number, especially when a design-led e-commerce build is non-negotiable.
Where a UX Design Agency Fits, and What to Ask For
Checking the source behind every number is only half the discipline. The other half is choosing a partner who will define those numbers with you before design work begins.
A studio that cannot show you the measurement plan before work begins cannot show you whether the work succeeded. A studio that requires those numbers before starting is describing accountability, whatever its portfolio looks like. The cost is modest: one planning session and a short instrumentation pass, which is small next to a redesign delivered with no verdict attached.
Aarland frames the work that way. As a Danish design studio and certified Webflow partner covering UX/UI design, websites, webshops, branding, prototyping, and wireframing, it states a focus on user-centric digital solutions with measurable impact. That makes the metric list a first-conversation item, not a closing report. For the local selection question, see hvad du bør kigge efter i et lokalt designbureau i Aarhus.
Be skeptical of any claimed improvement, from any studio, that omits the baseline, measurement window, and event definitions.
Conclusion: Put the Numbers in the Brief, Not the Retrospective
A measurement plan belongs in the brief, not in the retrospective. Fix four to six outcome metrics before design starts, baseline, target, window, and confirm events in staging. Then hold the review on the date you set at kickoff.
Your action step: add a one-page measurement plan table to your next brief. If a proposal for UX design work arrives without one, ask for it before you sign.
Use Baymard's cart-abandonment and checkout usability research pages as context and further reading. Verify any benchmark against your own data before quoting it to your board.
