How to measure what communication design returns for a deep tech company
The fee for a website or deck is on an invoice; its return is not. Six measures, a before-and-after baseline, and what not to credit to design.
- Set a 4 to 8 week baseline before launch, then track the same measures for 4 to 8 weeks after.
- Track inbound share of pipeline, qualified inbound conversations, calls before a buyer understands the product, and how fast a stranger can explain the company back.
- Add investor follow-up rate, search rank and citations in AI answers, and read each change beside what buyers and investors say.
- Read the results as direction, not a single ROI figure. A grant, partnership or raise is not a return on design unless someone documents the link.
The fee for a website, a deck or a data room is visible on an invoice. The value of shorter explanations and better qualified conversations is rarely tracked, which is why it keeps losing budget arguments to things that are. We made that argument in communication design is becoming a growth function. This is how to measure it.
What should a deep tech company measure after a redesign?
Six measures, each defined before launch: inbound share of pipeline, qualified inbound conversations, calls until a buyer understands the product, time for a stranger to explain the company back, investor follow-up after a first meeting, and search rank alongside AI-answer citations.
They tell you more than page views. A visitor can leave without understanding the technology; a shorter explanation or a qualified conversation is a signal that it landed. Comprehension also has to reach the whole buying group, not just the first contact. Forrester's State of Business Buying, 2026 puts a typical purchase at 13 internal stakeholders and 9 external influencers, most of whom will read your materials without you in the room.
How do you capture each measure, and what does a change suggest?
Capture each the same way before and after launch, and treat a change as something to investigate, not proof.
| Measure | How to capture it | What a change suggests |
|---|---|---|
| Inbound share of pipeline or revenue | Record pipeline value and revenue by account, with the tracked source and an open-text "How did you hear about us?" answer | More buyers arriving without being chased |
| Qualified inbound conversations per month | Count enquiries meeting the same buyer, use-case and readiness criteria each month | The materials help the right buyers recognise a fit |
| Calls before the buyer understands | After each call, note whether the buyer can describe the product and its use accurately | Clearer explanation, allowing for the buyer's prior knowledge |
| Time for a stranger to explain the company | Show the same page to 3 people unfamiliar with it and time how long until each describes it accurately | The materials are easier to understand |
| Investor follow-up rate | First meetings leading to a substantive follow-up, divided by all first meetings | Stronger interest, though fit and evidence weigh more |
| Search rank and AI-answer citations | Check fixed buyer queries and prompts on a schedule; record rank and whether answers cite you | More chances to enter buyer research, not more sales |
The stranger test is the cheapest and the most telling. We describe it, and why the window is 10 to 20 seconds, in why deep tech should be understood in twenty seconds.
How long should you measure before and after launch?
4 to 8 weeks on each side. That is long enough to smooth out an unusually quiet week and short enough to keep the comparison tied to the launch. Fix the dates and the definition of a qualified enquiry before you start.
It will not capture the whole effect on deals that take months to close. One contact can understand the product before colleagues have read anything, so enquiries usually move before qualified conversations do. Hold pricing, outbound effort and sales hiring as steady as you can across both windows, and log any change: a rise in inbound means little if outbound doubled in the same month.
Why can't you attribute pipeline growth to design alone?
Because no attribution model sees the whole B2B buying journey. First-touch credits the earliest recorded interaction, last-touch credits the final one, and data-driven models spread credit across what was recorded. None of them sees research that was never tracked.
A buyer can find you through an AI answer and later type your address straight in. Analytics records a direct visit; it does not record the answer, or whether a clearer explanation helped. An open-text "How did you hear about us?" field, stored separately from the tracked source, catches some of what analytics misses. When it agrees with movement in qualified conversations, treat that as supporting evidence, not proof.
What should you not credit to design?
A grant, partnership or raise that lands while new materials are in use. For each outcome, record which version of the website, deck or data room was live, when the other party saw it, and what decision followed. Credit the materials only when a buyer, investor or reviewer independently names them as a factor, and record exactly what they said.
Deep tech decisions often turn on evidence the materials cannot supply. A reviewer awards a grant because the results meet the programme's criteria. A clearer deck helps them assess those results; it does not validate the science.
Where do search rank and AI citations fit?
Beside the pipeline measures, because buyers increasingly meet a company in an answer before they meet it on a call. In Forrester's buyer research, 87% of B2B buyers selected generative AI conversational search as a meaningful interaction in their purchase, the top-ranked source.
Track the questions a buyer would actually ask about your technology, its evidence and its use. Record where your pages rank and whether AI answers cite them, with the same queries before and after launch. Rank and citations show buyers can find your explanation. They do not show buyers trust it, so read them against qualified conversations and what buyers say they understood before the first call.
What about trust, when the numbers are ambiguous?
Record what buyers and investors say without asking them to judge the design. Note when they describe the application correctly, ask fewer basic questions, or point to a specific diagram when they follow up. Those notes, beside the comprehension times and follow-up rates, are often the clearest evidence you will get.
Trust in the communication is still separate from trust in the technology. Investors will examine test results and commercial readiness however good the deck looks. A credible claim about design's return describes the change you observed and the limits of what you can attribute, and stops there.
Written by Mejo Kuriachan. More in the blog, the glossary and the FAQ.