Ask anyone who has sat through a post-launch review what happened, and you will usually hear the same story. The new site looks better, but nobody can agree on whether it actually works. Redesigns get judged on taste because nobody agreed in advance what was supposed to change. The fix is simpler than it sounds: baseline a small set of behaviours before a single page is designed.
In this tutorial, we will walk through a practical process for making your next website design Melbourne project provable rather than debatable. You will learn why redesigns get argued about after launch, why a baseline beats an opinion when a board asks, and how to pick two or three behaviours the new site must move, such as enquiry quality or mobile task completion. We will then cover defining each behaviour precisely, recording today’s numbers, setting targets and review points, and comparing results after launch. We will also look at what to do if you have already launched without a baseline, how to choose a partner who expects to be measured, and what to do before a single page is designed.
Why Website Redesigns Get Argued About After Launch
After launch, the conversation drifts to taste: colours, layout, whether the site “feels right”. Taste arguments have no referee, so the loudest voice in the room usually wins. It is a familiar trap, and why SMB brand visibility fails even when the logo looks right describes the same gap from a brand perspective.
That is how the same project can be a clear success to the design team and a disappointment to the board.
The missing piece is rarely measurement skill. It is timing. Success criteria get set after the design is finished, by which point recording a before number is already too late.
Most organisations hold the raw material already: enquiry volume in a CRM or shared inbox, page traffic in analytics, form submissions in a spreadsheet. Nobody captured them at a fixed point before work began.
A redesign changes behaviour, not just appearance. If you cannot say which behaviour should move and roughly how far, the project has no test to pass.
If you are commissioning website design in Melbourne, this is the first question worth asking: what will we measure, and what does it look like today?
Why a Baseline Beats an Opinion When a Board Asks
That is the conversation a baseline prevents, and it arises precisely because approvals rest on promises while reviews rest on opinion. Boards, funding partners and partner groups approve budgets, so they need a sentence they can repeat: we said this would improve X, and X moved from A to B. A baseline writes that sentence before you need it.
UK government digital teams publish this discipline openly. The GOV.UK Service Manual has a dedicated “Measuring success” section covering performance metrics, usability benchmarking, completion rate and cost per transaction. Treat it as useful precedent for method, not a standard Australian organisations must meet.
GDS reported in 2016 that it built a performance framework to check the success of the changes it was making, alongside research and iteration. That supports measuring as part of the redesign, not as an audit months later. It sits naturally beside UX research that comes before the brief is written.
A baseline also protects the designer. When someone questions a decision, you point to the behaviour the change was meant to move instead of defending a preference. It makes iteration possible too: without a before number, “let us try a different version” has no way to prove the different version is better.
The next step is deciding which two or three behaviours to track.
Step 1: Pick Two or Three Behaviours the New Site Must Move
Choose your behaviours deliberately. Pick behaviours, not aesthetics: a behaviour is something a visitor does, something your organisation already cares about, and something a website change can plausibly influence.
Most shortlists come down to three candidates:
- Enquiry quality: not how many forms were submitted, but how many were the right kind of enquiry.
- Service findability: whether someone reaches the right service page quickly.
- Mobile task completion: whether a phone visitor finishes the job they came to do.
Balance the list. One behaviour should be mission or revenue critical (enquiries, bookings, donations). Another should cover people who never contact you, where findability and task completion sit.
Test each candidate: is it already tracked, can it be counted the same way before and after, and would a change alter a decision you would make?
Look for different users, not just the average. People engage with guidance differently depending on their experience, so one figure can hide a group the redesign helps or hurts.
Write the list down and get it approved alongside the budget. A NFP website brief checklist can help shape that one-page sign-off. That page turns the post-launch review into measurement, not debate.
Step 2: Define Each Behaviour Precisely Enough to Count
Step 1 gives you the shortlist. This next step is the one most teams skip, and it decides whether your baseline survives contact with reality. A behaviour you cannot define precisely is a behaviour you cannot count twice.
Define each behaviour in a single sentence
For enquiry quality, decide what makes an enquiry “right”: the service requested, budget range, location, referral source, whether it was in scope. Then count how many of last month’s enquiries met that definition. A manual count of recent enquiries, 30 to 50 is workable, is a practical starting point.
For service findability, pick two or three high-value services and define success plainly: for example, reached the service page in three clicks or fewer, or searched and landed on the right page, though the right threshold is a working definition for your context, not an industry standard. Measure how often that happens today.
For mobile task completion, name the task (submit an enquiry, book an appointment, download a referral form) and count how many mobile visitors finish it compared with how many start it.
Write each definition as one sentence anyone on the team could apply, and note where the number comes from: analytics, CRM, form submissions, or a counted sample. This is the same discipline you would apply when writing a brief that an agency can actually use: vague inputs produce vague outputs.
Avoid measures you cannot repeat after launch. If a number depended on one staff member eyeballing a spreadsheet, it will not survive a busy quarter. Be honest about limits too: with no prior data, anything you record now is a snapshot, and a small sample is a reference point rather than statistical proof.
With definitions settled, you are ready to record today’s numbers.
Step 3: Record Today’s Numbers Before Design Starts
With your definitions in place, the next step is to capture the numbers themselves. Fix a date and a window, the last 90 days as a starting point, though the window should match your sales cycle, and record each behaviour as it stands, labelled with the date it was captured. Note the context alongside: seasonality, a campaign that was running, a referral partner, a form that was broken for a fortnight. A number without context gets misread six months later.
Put it in writing and share it. If the baseline lives in one person’s head and that person leaves before launch, you are back to opinion. Frame any figures as examples, not benchmarks. For example, suppose 40 enquiries a month come in today and 12 are the kind you want, that is a comparison point for your organisation, not a target for anyone else. It is also worth remembering that measurement sits alongside the investment you have already approved for the build, so understanding what drives website design costs in Australia helps you see the baseline as part of the same project, not an extra.
Set collection up so the same numbers can be pulled the same way after launch: same tool, same definition, same reporting window. If the tracking needs fixing, fix it now. Definitions belong in the same file as the numbers, without them, the before figure is uninterpretable.
Next, you will set the target and the review point.
Step 4: Set the Target and the Review Point
With today’s numbers stored, decide what you are aiming for and when you will check. Set a direction and rough size, not a promise: more of the right enquiries, or more mobile visitors finishing the form. Say which part is a considered target and which is an informed guess. Pick the review point in advance. As a rule of thumb, thirty days confirms tracking works, 90 days gives a first read on behaviour, and seasonal or long-sales-cycle measures need six to twelve months, these are practical starting points, not sourced standards. Agree who reviews it and what happens if numbers move the wrong way; without a decision attached, it becomes a report nobody reads. Clear ownership from the start helps, as our guide to NFP website governance explains. Be realistic about what a website controls. If enquiry quality depends on pricing, response times or availability, say so up front so the site is not blamed. Avoid unsupported percentage promises: “We expect more of the right enquiries and will report at 90 days” is defensible; “We will double leads” is not.
Step 5: Compare After Launch, Then Iterate
Once the review point arrives, compare like with like: same definitions, same window length, same sources. A jump caused by a changed definition is not a result, it is a reporting error.
Report it on one page a board member can read: the number, the context and the caveats, in plain language. Plain English beats a dashboard nobody opens.
Expect a mixed picture. Usually one behaviour improves, one flatlines and one gets worse. The baseline is what lets you tell which is which and act on it, instead of explaining the whole project by its best result.
Feed the findings into the next round of changes, alongside user research. Performance analysis is most useful when it shapes what you test next, not when it sits in a report that closes the project.
Then update the baseline file. The next redesign starts with a before number already in hand, and every review after that gets faster.
Already Launched Without a Baseline? What to Do Now
If the launch has already happened, you are not out of options. You cannot reconstruct a true before number, but you can build a comparable one starting today: record current figures, note the date, and make the next change measurable.
First check what history survives. Check whether your analytics tool has retained pre-launch data, many do, though retention periods vary by platform and configuration, and your enquiry team can describe how the mix of enquiries has shifted since. Add qualitative evidence honestly: even a small sample of recent enquirers and people who looked but never got in touch, say five of each, is a practical starting point, not a statistically representative one. Report their comments as insight, not proof.
Where no before data exists, use a paired comparison instead: set the redesigned site against a section you have not changed, or mobile behaviour against desktop.
The updated Magenta Book (May 2026) represents, in the words of the Civil Service blog, “the most significant shift in government evaluation standards since 2020, moving the needle from ‘measuring what happened’ to ‘learning how to improve in real-time’.” That principle, building evaluation in from the start rather than retrospectively, applies to any organisation running a website project, not just government teams.
Write the baseline for your next round now. If platform limits are part of the problem, our guide to custom web application or off-the-shelf can help you decide before the next rebuild. Treat this project as the reason the next one will be provable, not arguable.
Choosing a Partner Who Expects to Be Measured
Whether you’re planning ahead or recovering from a launch, your choice of partner decides whether measurement happens at all. When comparing a web design agency, ask what they will baseline and how they will report it. Any team serious about UX design should answer clearly. Ask for metric definitions in writing before work starts, and who owns the numbers once the site is live. Look for a partner who will tell you what cannot be measured rather than implying every metric will improve. Keep the baseline file in your own hands; the evidence should belong to your organisation, not the agency that built the site. For help weighing partners, see our guide to strategic partners versus build shops. Slant Digital is a Melbourne digital agency working with not-for-profit, healthcare, automotive, logistics, retail, property, architecture, legal and other professional-service organisations. Our work covers brand strategy and identity, UX research and design, website design and development, ecommerce, custom development, hosting, and ongoing support and maintenance. A baseline can sit alongside research and design work before a page is designed.
What to Do Before a Single Page Is Designed
Whether a partner runs it or you do, the work starts here. For most organisations the shortlist of behaviours covers enquiry quality, service findability and mobile task completion, get it approved alongside the budget, not after it.
Record today’s numbers with the date and the context around them, such as a campaign that was running or a form that was broken.
Without a baseline, every post-launch conversation is opinion against opinion. With one, it becomes a short conversation about evidence you can defend and improve.
Conclusion
A successful website redesign is built on evidence, not opinions. This simple habit of establishing a baseline delivers clarity and confidence for your team and your board.
If you are planning a redesign, the most valuable step is to start this process now. Having a partner like Slant Digital who expects to be measured can make all the difference. They can help you put a baseline in place alongside the research and design work, ensuring your next project is set up for success from the very beginning.
