Agree the words before the claims: how we build messaging frameworks for life-sciences companies

Four people at one life-sciences company gave four first arguments for the same product. This is how we turn them into one messaging framework: separate stakeholder interviews, a workshop spent only on disagreements, a terminology table before the framework, and claims your scientists sign.

We asked four people at one life-sciences company the same question: why do customers buy from you?

The CEO led with quality. Sales led with price. The applications scientist also said price, then told us the headline benefit rested on one customer's story. Operations named a performance figure where they beat every competitor. Sales named a different figure as the first objection it hears.

Same company. Same product. Four first arguments.

Nobody was lying. Each answer came from real customer conversations in a different corner of the business. That is where most life-sciences messaging projects start. The problem is rarely a missing tagline. It is four true answers that don't add up yet, and claims that won't survive a competitor's scientist.

So we start with the words, and the claims come after. The rest of this article follows that order, from the first interview to the page your buyers read.

What is a messaging framework?

A messaging framework is the signed-off record of what your company says, to whom, and with which proof. We build it in two parts. The product messaging framework turns each feature into a capability and a benefit. The value framework sets up to three ICPs, three pillars, proof points and keywords. Positioning picks where you compete. Messaging picks the words.

We don't arrive knowing your market

Before the method, an admission that saves time later. Nobody knows a new company's market on day one, however many years they have spent in the field. We say so before you have to.

What we do bring is knowledge of the field, which is a different thing. We know how a validated lab switches supplier, which words a buyer misreads, and which question a sceptical lab director asks first. We also know how search and AI engines name a category. What we can't bring is knowledge of your company: which customer is about to switch, what the data really shows, what the product is called on the invoice. Only your team holds that.

The line between the two matters, because expertise has a blind spot of its own. In Pamela Hinds's 1999 study on the curse of expertise, experts misjudged how hard a task was for novices, and the bias resisted correction. So we arrive with questions, not answers, and we ask for teaching when we need it. On one project, before we wrote a line, we asked the client's science lead for a ten-minute tutorial on their field.

The deal is simple. You'll teach us your company. You won't have to teach us your field. Here is where that line runs.

Stakeholder interviews: the people who sell it, ship it, test it and buy it

Crossing that line starts with four or five stakeholder interviews: the CEO, sales, an applications or technical lead, and someone in operations or marketing. Each gets an hour, alone. That last part is deliberate, because groups lose what only one person knows. In Diehl and Stroebe's 1987 brainstorming experiments, people working alone produced more ideas than the same number talking together. The main cause was simple: only one person can speak at a time. Research since Stasser and Titus (1985) adds a second problem: groups spend their time on what everyone already knows.

So each person gets the questions in advance, and no two people get the same list. The blocks stay fixed: vision, the problem you solve, what changes in five years, what failure looks like, then category and positioning. The questions inside them change with the chair. The CEO gets the five-year question in full. Sales gets the first objection it hears and the first three questions at a booth. The applications scientist gets the data behind each claim. Operations gets the five questions customers send most.

The call opens with an off-limits check: customers not to name, products not yet announced, history to keep quiet. Then comes one request: give us your view, not the company line. We want the contradictions on record, because the workshop will need them. Every call ends with the same two questions, whoever is in the chair: which question should we have asked, and what did you forget to tell us? On one project, those two questions surfaced the most delicate facts of the hour.

Tailoring the list is also why the list of interviewees runs wider than most teams expect, and why the person who processes orders earns a place on it. At one company, operations met a legacy brand name every day as an invoicing problem, while the CEO saw it as history. Its five customer questions told a similar story. On one project the estimate was that 80% concern logistics, and the website had no FAQ at all.

None of this requires us to pretend we know your market. Fluency shows in the questions instead. Here are eight we ask on most projects, each with the reason it matters in life sciences:

1. "Why would the specialist who runs this process today argue for a machine that does part of their job?" Champions and blockers sit in the same lab. The answer separates defensive specialists from scientists open to change.
2. "What does switching cost your customer before, during and after implementation?" In a validated workflow, unit price is often not the decisive cost. A clinical lab verifies every new reagent lot under ISO 15189:2022, and a GMP site runs a material change through change control under EU GMP Annex 15.
3. "Is that benefit backed by data, or by one customer's story?" The answer decides whether a claim goes on the homepage or stays in a sales conversation.
4. "Does this feature still matter once the customer's process is locked?" A feature that sells to process developers can be irrelevant in routine or clinical use.
5. "Does your customer judge you on your step, or on the last step of their workflow?" Upstream suppliers get blamed for downstream failures. The message has to reach the step the customer measures.
6. "Who writes the SOP, who uses it, and who can kill the deal?" What we typically see: the technician follows the protocol, the lab manager champions, a manager above vetoes and procurement negotiates. Each needs a different sentence.
7. "What are the first three questions you hear at a booth, in order?" The order shows what buyers doubt first. On one project: quality, then sample types, then how to test it.
8. "What does a buyer type when they look for this?" Nobody searches for an "end-to-end" research partner. They search for the specific service.

Those answers give us your team's view of the buyer. The buyers' own view comes later. Customer interviews run after the workshop and before the first draft, once your team has agreed which personas to test. The order matters, because the two views drift apart. In Dentsu's 2024 B2B Superpowers Index, 68% of buyers said many B2B brands sound the same.

Preparing, running and analysing the stakeholder interviews takes ten working days. On one recent project, four interviews ran in five days. By then we hold four or five honest versions of your company. The workshop's job is to turn them into one.

The positioning workshop: spend it only on what you disagree about

Even honest versions overlap, and the overlap needs no meeting. So before the workshop we sort every interview answer into agreed and disputed. Agreed points get written down without debate, and the room spends its time only on the disputes. The founders see those slides in advance, so nobody meets a surprise in front of their own team.

The agenda runs in a fixed order. The agree/disagree pass takes about 20 minutes. Next come the buyer tier, the personas and the order of the messaging pillars. Then the company description, agreed word by word. A tone-of-voice exercise closes the session, with the team choosing between pairs of draft posts.

Each of those decisions gets forced into something concrete. For personas we ask: with a €5,000 LinkedIn budget tomorrow, which job title would you target, and with which sentence? On one project that question surfaced a buyer nobody had named, a risk officer who never joins a sales call. The tier works the same way. One team chose the mid-size labs it can sign within months over the largest ones, which buy in multi-year cycles.

For the pillars, the order comes from customers, not from us. On one project we proposed five. The CEO had met a customer that morning, who ranked quality first and price second. Sales then argued that support belonged next, because competitors' customers complain about service. Five minutes later the team had three primary pillars and two secondary ones.

A team that agrees on most points gets through all of this in about 90 minutes. A team that agrees on little needs a second session, and we book it instead of forcing a decision on the day. Here is where the workshop sits in the whole project.

Agree the words before the claims

The workshop settles positions. What remains is vocabulary, and in our experience it is the harder half. Most stalled approvals we have seen stall on a word, not on a strategy. If R&D, regulatory and the website call one component by three names, reviewers stall, and so do buyers.

Two lists get confused here. The terminology table fixes what each product, component and service is called, while the focus keywords are the words buyers search and AI engines repeat. We build both, and your team signs off the terminology table before the framework. The check runs both ways, too. In one workshop we had read a key quality metric backwards, and the client's team corrected us before it reached a draft.

Category is the first word teams get stuck on, because it does the most work. April Dunford's quickstart guide to positioning treats a market category as a frame of reference. It tells buyers who you compete with, which features to expect and what you roughly cost. Creating a new category is possible, but it is expensive and slow, because you have to teach the market what it is. The category-design school behind Play Bigger argues the prize justifies that cost.

For most scale-ups we pick an existing category plus one precise qualifier, and the argument comes down to search behaviour. We told one software client that refused the LIMS label: buyers who need what a LIMS does search for "LIMS". Our article on positioning for the US medtech market makes the same choice in a different context.

Below the category, some words are right scientifically and wrong commercially. A product can be one kit in the lab and several kits on the price list, and the price list wins. Words also wear out. When competitors copy yours, move one level up. One client's "free of X" claim was being echoed by rivals, so the claim moved to the whole class of hazard that X belongs to.

Keywords follow the same discipline. They come out of the interviews first and only then get checked in search. The competitor check is a spreadsheet, not a report. We open 2 to 5 competitor sites and record each H1, the section under it and the features that recur. The pattern shows what the industry calls the thing you sell. Here are six swaps from recent projects.

Let your scientists break the claims

Once the words hold, the claims built from them have to hold too. Every claim goes past your applications scientist before it enters the framework. The test is one question: would it survive a competitor's scientist reading it with a red pen? Remember the applications scientist in our opening, who flagged that the headline benefit rested on one customer's story. That is this review at work before anyone had scheduled it.

"Better" is the most expensive word in life-sciences marketing. In the EU, a comparison with a competitor must be objective and verifiable under Article 4 of the directive on misleading and comparative advertising. Article 7 of the MDR and the IVDR bans device advertising that misleads on performance. The commercial logic runs the same way: if your advantage holds for one application, "better" on the homepage hands a competitor an easy rebuttal. "Same quality" plus a second, provable benefit is harder to attack.

The next split is between what you can quantify today and what rests on a model. A physical measurement with third-party data goes on the page. An economics model built on assumptions stays in the sales conversation until a customer confirms it. Even a good number is only half a claim until it carries its consequence. One customer's pass rate rose by several points. For a service lab every failed test costs money, so each point is margin. That last sentence is the claim.

"Validated" deserves the same care, because it is a technical word before it is a marketing one. It needs three answers: what was validated, by whom, and against what. ISO 15189 separates verification, where a lab confirms an approved method performs as specified, from validation of a new or modified method. So if your own bar is three repeats and you have one, say what you have: external data and early user feedback.

Some of your strongest data won't reach the page for a simpler reason: you can't publish it. There are two ways round that. Reproduce a published paper's analysis on your own platform, or turn a customer's conference slide into a news post, with their permission.

Through all of this, we bring the questions and your scientists bring the corrections. Here are four claims before and after that review.

One feature, two messages

A claim can be true and still land on the wrong buyer, because the same feature carries a different value for each one. Only a technical question finds the split.

One client built a system in which each unit runs its own conditions, so we asked who uses that, and at which stage. For process development, it means more data and more control per run. In routine clinical use the process is locked, so the same feature matters as capacity instead of control. The client saw it at once: the message changes with the persona.

A second client sold one robotic arm into three assays, and there the split ran by application. For one assay the value was running the instrument day and night. For another the pain was scheduling and incubation timing, not moving plates. We wrote one message per application, and the client's engineers held the content.

That is why we slice the framework by application before persona, with one sentence per application. Here is the first case side by side:

Facts first, story last: a product messaging framework, then a value framework

All of that material, the words, the claims and the splits, now has to land in one document your team can approve. The order inside it decides how the review goes. A first draft that opens with a tagline invites an argument about spin. One that opens with features invites corrections. We want the corrections first, so the framework comes in two parts.

Part one is the product messaging framework. Every feature gets three lines: what it is, what it lets the customer do, and what changes for them. We list everything first, then cut. The feature, capability and benefit ladder is a practitioner convention, used by Myk Pono and by Fletch PMM, not a tested law. Fletch PMM's case for leading with capabilities holds that buyers believe a benefit once they see how the product works.

Part two is the value framework, built on those facts. It sets up to three ICPs: one segment that pays the bills, one small segment with high margins, and one strategic growth bet. Each ICP gets up to three value pillars, with proof points and keywords. To choose what reaches the homepage, we score each value on relevance to each ICP, that ICP's share of revenue, and uniqueness. The top scores win.

Three is our working rule, not an industry standard. Geoffrey Moore's Crossing the Chasm argues for something narrower still: win one beachhead segment before you move to the next. We stop at three because beyond that the map stays correct but stops being usable.

Only once both parts are signed off do we write the 30-second pitch, the company description and the boilerplate. By then the story has little left to argue about.

How the framework reaches you matters as much as what is in it. We deliver in slices: ICPs first, approved, then the messaging per ICP. A 20-page document landing in one go reads as too fast to be true. Where a word stays contested, we don't settle it on the call. Instead we send the options, the pros and cons of each, and our recommendation, and your team decides internally. V2 then highlights every change, so nobody re-reads 20 pages.

The framework itself arrives as a Word document with tracked changes, so every stakeholder comments in the same file.

Six ways a messaging project stalls (and how to avoid them)

Most of the method above exists because of what goes wrong without it. These are the six patterns we watch for from the first call:

1. Debating the tagline before the facts are agreed. The room argues about how the company sounds instead of what the product does. Fix: sign off features and proof first, and write the story last.
2. Putting every audience in one map. Five ICPs on three levels in one document reads as correct and confusing. Fix: agree the ICP count before work starts, cap it at three, and slice each one by application.
3. Copying claims from your own website. Old pages carry applications and claims the science no longer backs. Fix: your applications scientist checks every claim before it enters the framework.
4. Leaving terminology open. A team can't approve a document while it still disagrees on what a component is called. Fix: sign off a terminology table before anyone reviews the framework, and name one person who signs off technical wording.
5. Fixing positioning in page reviews. When the framework isn't signed off, every wireframe review reopens it, one headline at a time. Fix: framework first, brand second, website third.
6. Designing the commercial offer last. A campaign built on personas falls apart when the offer doesn't fit them. Fix: decide the offer in the same session as the persona it serves.

Every fix on that list moves a decision earlier, to the point where it costs one meeting instead of one redesign. The last one, framework before website, leads to the part most teams care about: what happens once the words go live.

A message is a habit, not a document

Sign-off is where the work starts showing in public. Homepage and technology page copy start while your team reviews V2, so the framework meets a real page within weeks. Product and application pages follow later; the framework feeds them, it doesn't replace them. On the homepage the brand message leads, and product detail lives one click deeper. One test settles most hero-line debates: put a competitor's name in front of your headline. If it still reads true, rewrite it.

After launch, the framework keeps moving with the market. We review it every six months against what sales heard in the field. That matches the rhythm April Dunford describes in her 2026 interview on positioning. As an in-house marketing lead she checked positioning every six months, and positioning under a year or two old rarely needed changing. So we change words when sales hears something new, not because the calendar says so. The keywords and customer questions from the interviews feed the search and AI-search work in the same way, once the site exists.

Which brings us back to the four people in our opening. Their four first arguments become one, in words their scientists will defend and their buyers will search for.

Frequently asked questions

How do you communicate complex science simply to buyers?

We don't simplify the science. We move it. Each feature gets three lines: what it is, what it lets the customer do, and what changes for them. The buyer reads the third line first. Jargon makes readers resist a message, as Bullock and colleagues showed in 2019. So the mechanism moves to the technology page, where the scientists who check your claims will look for it.

How long does building a messaging framework take?

About two months from kickoff to a signed-off framework, plus one month for customer interviews. Stakeholder interviews take ten working days to prepare, run and analyse. The workshop takes about 90 minutes. Then one comments call, a second version with every change tracked, and a final version.

What happens in a positioning workshop?

We write down everything your team already agrees on, without discussion. The room then spends its time on the points where answers differed. Expect to decide which buyers you speak to first, the order of your messaging pillars and your company description, word by word. A short tone-of-voice exercise closes the session.

Who should we put forward for interviews?

Four or five people, interviewed one at a time: the CEO, someone in sales, an applications or technical lead, and someone in operations or marketing. Pick the person who answers customer emails, not only the person who presents to them. They hear objections the CEO never hears.

Do you interview our customers?

Yes. Customer interviews come after the stakeholder interviews and the workshop, and before the first draft. By then your team has agreed which personas to test, so each conversation checks a decision instead of opening a new debate.

How many ICPs should a life-sciences company have?

Our working rule is three. One segment that pays the bills today, one small segment with high margins, and one strategic growth bet. Geoffrey Moore's classic advice is narrower: win one beachhead segment first.

Which scientific claims can we safely use?

The ones your applications scientist will defend in front of a competitor's scientist. Run four tests. Does it hold for every application you list? Is the data yours and repeated, or one customer's story? Does "better" survive a competitor's own comparison? Can you quantify it today? If not, the claim stays in the sales conversation.

Start with the claims you argue about

Send us your current deck and the three claims your team argues about most.

Send your deck

Sources

• April Dunford, A Quickstart Guide to Positioning
• April Dunford, interview in A History of Marketing, 2 April 2026
• Diehl, M. & Stroebe, W. (1987), Productivity loss in brainstorming groups, Journal of Personality and Social Psychology 53(3)
• Stasser, G. & Titus, W. (1985), Pooling of unshared information in group decision making, JPSP 48; meta-analysis: Lu, Yuan & McLeod (2012), Personality and Social Psychology Review
• Bullock, O. M. et al. (2019), Jargon as a barrier to effective science communication, Public Understanding of Science 28(7)
• Hinds, P. J. (1999), The curse of expertise, Journal of Experimental Psychology: Applied 5(2)
• Directive 2006/114/EC on misleading and comparative advertising, Article 4
• Regulation (EU) 2017/745 (MDR) and 2017/746 (IVDR), Article 7
• ISO 15189:2022, Medical laboratories: requirements for quality and competence, clauses 7.3.2 and 7.3.3
• European Commission, EU GMP Annex 15: Qualification and Validation (2015), section 11
• Fletch PMM, Lead with Capabilities over Benefits
• Geoffrey A. Moore, Crossing the Chasm
• Al Ramadan et al., Play Bigger (2016)
• Dentsu, B2B Superpowers Index 2024 [CHECK SOURCE: primary report not found; 68% figure seen via MarketingProfs]