01
02
03
04
05Sections
A company approves a website redesign because the current site feels old, the business has changed or investors have criticised the presentation. Six months after launch, the team likes the new site. Whether the money was well spent depends on what the website was meant to change.
A site built to bring sample requests from buyers needs a different test from one built for recruitment, investor diligence or a market launch. “Modern and professional” is a fair design requirement, and it leaves the commercial objective unstated.
Name the decision before you choose the measures
Name the reader, the action you want from them and the evidence they need to take it. For an ingredient company, that might be a formulation lead who requests a technical sample after confirming the application and the specification. For an industrial technology company, it might be an operations team that agrees to a qualification conversation. For a company raising money, it might be a meeting with an investor whose mandate fits, which is a narrower thing than every visit to the team page.
If you filled in the deep tech website brief before the project, its first line and its conversion events already hold this. Measure against them.
Record the baseline before anything moves
Take a snapshot of the current site before the redesign starts. Keep it in one document with a date on it.
The baseline snapshot
- The key pages, the search queries that reach them and the visits each receives.
- Every conversion point, such as forms, sample requests and booking links, with the number each received.
- The inbound enquiries over a recent period, each marked for fit with your market.
- The questions the sales team still answers by hand in most first meetings.
- The misunderstandings buyers and investors repeat.
- The planned launch date, and any campaigns, events or press running near it.
Write down the limits of the data as well. Missing consent, inconsistent tags, direct visits and introductions made offline all leave gaps in digital attribution. For the search side of the snapshot, the biotech website migration checklist lists what to export and keep.
Judge four kinds of outcome
Four layers of outcome
- Understanding
Can the intended visitor say what the company does and what is available now? Test it with short interviews. Time on page measures something else.
- Evidence access
Can a technical buyer or investor find the data and its limits without another meeting? Review the routes through the site and count the questions sales still answers by hand.
- Qualified action
Are the right people asking for the next sensible step, such as a technical discussion, a sample, a proposal or the deck? Count actions by fit as well as in total.
- Commercial progress
Do those actions become opportunities, agreements or projects over time? Price, product and the sales process shape this layer as much as the site does.
Choose two or three measures in each layer. Give each measure a decision it will inform, such as which page to change next or which route to drop. A dashboard of numbers with no decision attached tells the team very little.
Give each layer its own time horizon
When each layer can be read
- Understanding and evidence accessWithin weeks of launch, through interviews and a review of the routes.
- Qualified actionOver a few months, set against the same period in the baseline.
- Search visibilityOver a longer period, once crawling and indexing have caught up with the new structure.
- Commercial progressOver the length of your own sales cycle.
Hold judgement until each layer has had its horizon. A weak first fortnight is too early to read, and one large referral in launch week is an anecdote.
When the numbers are small
Where enquiry numbers are small, a percentage change rests on a tiny denominator. Review five or ten relevant opportunities one by one. For each, record how the person found the company, which pages they read, what they asked first and what happened next. That record shows whether the site did its job for the people who matter, which a rate built on a handful of enquiries cannot show.
Treat traffic as a diagnostic
A broad educational article can bring thousands of curious readers who will never buy. A page on one specific industrial process can bring far fewer, and some of them may hold budgets.
Traffic still helps you diagnose whether the right people can find the site. Track three links in the route, from the search query to the page, from the page to an action and from the action to a qualified conversation. Where one link is unclear, fix the measurement before you decide which content is valuable. Our guide to technical websites that draw visitors and few enquiries covers the case where the route breaks between the page and the action.
Record what happened, and what it proves
A result after launch shows timing. Causes need more. Our own case pages follow that rule. The Fronterra case records that the company's public investment round grew from roughly €45,000 to €300,000 before closing, in the period after the new brand, website and investor materials went live. The case records the timing and makes no claim about the cause, because a round also turns on the investors, the terms and the market.
Write your own results the same way. Note what changed, when it changed and what else was happening at the time, and keep the cause as an open question until the evidence closes it.
A one-page redesign scorecard
Fill this in before launch, then review it at each horizon above, alongside notes from the sales team.
| Field | What to write |
|---|---|
| Business problem | The reason the redesign was approved, in one sentence. |
| Reader and action | Who the site is for first, and the action that counts as a result. |
| Baseline | The dated snapshot, with the limits of the data. |
| What changed | The pages, routes, words and evidence that are new or moved. |
| Understanding | What interviewed visitors could say before and after. |
| Evidence access | The questions sales still answers by hand, before and after. |
| Qualified actions | Actions by fit and in total, against the baseline period. |
| Commercial progress | Opportunities that followed, each with its source and status. |
| Uncertainties | What the data cannot show, and other events in the period. |
| Next decision | The change the scorecard points to. |
If the scorecard shows that visitors still cannot say what the company sells, the offer needs work before the next round of measurement. The free Brand Strategy System starts with those choices.
When the reader, the action and the baseline are written down and you need the site built, our website work for science and climate companies designs and builds against them.
Sources and further reading
Brighter Future worked with FronterraRead the case study
Studio serviceWebsite design for science and climate tech companies
Tell us what needs to move.
Bring the brief if it is clear. If it is unclear, tell us where the work is stuck.
Bring us the problem