A surprising number of growth problems are really translation problems.

The product may not be broken. The value may exist. The team may have shipped quite a lot.

But a user lands on the site, sees the ad, opens the page, and thinks:

  • what exactly do you do?
  • why is this relevant to me?
  • why should I click this CTA now?
  • this sounds polished, but where is the proof?

Plenty of growth experiments fail right there. Not because the product cannot deliver, but because the promise is unclear, aimed at the wrong audience, or not carried through properly on the landing page.

So this piece tries to separate four things that teams often bundle into one foggy blob:

  • positioning
  • value proposition
  • messaging
  • landing page

They are related, but they are not interchangeable. If you treat them as one layer, experimentation gets muddy fast.

A common misdiagnosis

When conversion is weak, the first reaction is often something like this:

  • the hero copy is not strong enough
  • the CTA needs work
  • the page design needs polishing
  • perhaps the homepage should look more premium

Those things can matter, but many teams have not answered the more fundamental question:

what value are we promising, to whom, and does that promise match the user’s current intent?

If that is still vague, all the later work becomes cosmetic. The page may look more sophisticated, but conversion does not necessarily improve because the issue was never sentence quality alone. The issue was that the promise was out of focus.

NN/g’s advice on homepage design is refreshingly blunt. The homepage and hero space are not there to impress first. They are there to help people grasp who you are, what you do, and why it matters to them. If users have to infer all of that, many will simply leave. 1

Positioning, value proposition, messaging, and landing are not the same layer

This is the model I find most practical for PM work. It is not the only way to structure it, but it is useful because it clarifies responsibility.

1. Positioning: what are you, really?

Positioning is about context and contrast.

What category are you in? What are you actually replacing? For whom do you matter most? What kind of value do you win on?

April Dunford makes the distinction clearly: positioning is not messaging. Positioning establishes the frame in which the product should be understood. Messaging then translates that frame for different buyers and channels. 2

That matters because if the positioning is still fuzzy, every new headline is usually just a new version of the same blur.

2. Value proposition: why should this person care?

A value proposition is not a slogan and it is not a string of polished adjectives.

Its real job is to answer:

  • why this product deserves attention
  • what problem it helps solve
  • why it is preferable to obvious alternatives

NN/g puts it neatly: a value proposition should answer the question, “Why should I choose this site or company over others?” If that remains unclear, more traffic simply magnifies the ambiguity. 3

3. Messaging: how do you say it to this audience?

Messaging is not just rewording. It is translating the same underlying value into language a specific audience will recognise, care about, and want to act on.

The same product can require very different messages for different segments.

Take a booking product:

  • for families, the promise may be about finding places that are suitable, clearly described, and less likely to create unpleasant surprises
  • for business travellers, the promise may be speed, convenience, and fast confirmation
  • for long-stay users, the promise may be monthly availability, liveable amenities, and less hassle over repeated moves

The product is unchanged. The message is not.

If you send all three groups to one broad, generic page and hope conversion will sort itself out, you are mostly relying on luck.

4. Landing page: how you cash the promise into a next step

The job of a landing page is not to present every possible piece of information about the company. Its job is to receive a specific promise and carry the user into a sensible next action.

That is why I do not like collapsing the homepage and the landing page into one discussion. A homepage can be broader. A landing page usually needs to be narrower, sharper, and more tightly matched to the user’s source intent.

Google Ads’ guidance is explicit here: your landing page should match your ad and keywords closely, so people can quickly find what they expected when they clicked. That is not only a UX issue. It also affects ad relevance and landing page experience. 4

A useful way to remember the stack is this:

positioning decides the battlefield;
value proposition decides the value you lead with;
messaging decides how you say it to this segment;
landing decides whether the promise survives contact with the page.

Many conversion problems are really promise problems

This is how I tend to diagnose the situation.

Case one: users do not understand what you do

This is usually a positioning or value proposition problem.

The hero copy says things like:

  • reimagine your growth workflow
  • the future of intelligent travel experiences
  • smarter stays for modern life

The words are not difficult, but the information density is poor. Users are left to infer what the product actually is, what it replaces, and why it is worth their time.

NN/g’s stance is sensible: do not make people guess. If your brand name does not already explain the business, your hero needs a concise, explicit line that does. Generic welcomes and vague slogans are a bad use of the most valuable space on the page. 1

Case two: users understand the category, but not why this is for them

This is usually a segmentation and messaging problem.

The product may have value, but the page speaks in a flattening, average voice. Average messaging often sounds perfectly acceptable while failing to feel specific to anybody.

That is why I rarely recommend beginning with one universal landing page for everyone. In many cases it is better to pick one audience, one scenario, and one primary promise, then write the page around that.

Case three: the upstream message attracts the click, but the landing page breaks the spell

This is a message-match problem.

Google Ads explains it in practical terms: if the ad promises something, the landing page should help users find that same thing quickly, and the CTA should continue the same action. You cannot advertise a “free tour” and then open on a generic page with no obvious tour flow. You cannot bid on “monthly apartments” and then drop users onto a broad accommodation homepage. 4

That kind of mismatch weakens trust immediately and makes diagnosis harder. Was the traffic wrong? Was the promise wrong? Or did the page simply fail to honour it?

When to test the message before building more product

Here is the practical judgement call I come back to most often.

If the product can already solve the core job, but you are seeing patterns like these:

  • clickthrough is acceptable, but the top of funnel leaks badly
  • conversion varies wildly by traffic source
  • users often say “I thought you did X”
  • awareness exists, but intent does not carry through

then I would usually suspect the message and the landing page before I rush into a new feature build.

Those are often signs that the product exists, but the understanding does not.

That is when it is worth testing:

  • which segment framing resonates most
  • which promise moves people into the next step
  • which proof reduces scepticism
  • which landing structure best carries the promise forward

How I would run a message test: protect the single main variable

One of the most useful principles in your earlier notes was the idea of a single-variable test.

That matters because teams often say they are “testing messaging” when they are actually changing:

  • the audience
  • the hero line
  • the image
  • the CTA
  • the page structure
  • the social proof
  • the form length

If a version wins, nobody really knows why.

So I prefer to think about message testing in layers.

Layer 1: test the core promise only

Keep the page broadly the same and change only the main promise.

For a booking product, that might look like:

  • less re-searching: find relevant places faster
  • less uncertainty: clearer rules, fewer nasty surprises
  • more speed: compare and book viable options in minutes

This layer is asking: which promise is worth developing further?

Layer 2: test the audience frame only

Keep the product the same, but build the page for one audience at a time.

For example:

  • a family travel landing page
  • a business travel landing page
  • a long-stay or monthly-rental landing page

This layer is asking: which segment’s pain aligns most tightly with the product?

Layer 3: test the proof and fulfilment mechanism

Keep the promise, but change how you support it.

For instance:

  • feature explanation
  • outcome-led proof
  • sample inventory
  • user evidence
  • rules and pricing transparency
  • demo, trial, or template

This layer is asking: what kind of proof makes the same promise believable?

Once you keep the main variable under control, the learning gets much cleaner.

That is also why, in many low-to-medium traffic contexts, I would start with A/B or A/B/n rather than immediately reaching for multivariate testing. Optimizely is fairly explicit on this point: A/B tests are simpler and quicker when the number of changing variables is small, whereas multivariate testing needs far more traffic and carries heavier interpretation costs. 5

A landing page is not a general information page. It is a promise-fulfilment page.

I want to put this more sharply.

Many weak landing pages are not failing because the design is ugly. They are failing because they are not really landing pages at all. They are storage units for information.

A useful landing page should do at least four things.

1. Confirm immediately that the user has arrived in the right place

That is message match.
What the ad promised, what the link implied, what the search intent suggested: the first screen should honour that.

2. State the promise clearly in the user’s language

NN/g stresses this too. A value proposition should be expressed in language users understand and care about, not in internal jargon or feature-heavy brand phrasing. 3

3. Offer enough proof to support the claim

If you say you are the less risky option, what makes that true?
Transparent rules? Better reviews? Faster confirmation? Fewer hidden costs?
Short promises are fine. Empty promises are not.

4. Make the next step feel coherent

The CTA needs to fit the promise that came before it.
If the pitch is “take a free tour”, the CTA should not suddenly become “contact sales”.
If the pitch is “see available examples first”, the page should not demand a long form before showing anything useful.

When not to start with message tests

This needs a counterexample section, otherwise the article becomes a sermon about copy.

There are several cases where message testing is not the immediate priority.

1. The product cannot reliably deliver the promise

If supply is weak, error rates are high, the first screens are empty, or the core flow keeps failing, better messaging just sends more people into a bad experience.

2. The real issue is activation, not acquisition

If people are arriving and trying the product, but getting stuck before the first value moment, the better question is how to improve the activation path. That was the point of the previous article.

3. Traffic is thin and the test scope is too broad

That is where teams end up with several weak variants and very little learning. Better to narrow to one battle and one main promise than to run a vague page derby.

4. The audience has not been segmented at all

If everyone is being sent to the same page, with the same framing, and the same CTA, the issue may not be that the value proposition is weak. It may be that you are still speaking to a statistical average instead of a real segment.

What a PM should actually produce here

If this is going to become real PM work rather than a copy debate, I would want at least four outputs.

1. A message-stack map

Something that separates:

  • positioning
  • value proposition
  • message
  • landing promise
  • CTA

2. Two or three priority segments

Not a broad market description. A small set of audiences worth testing properly.

3. One core promise per segment

Not a feature list, but a single result-oriented promise each audience would actually care about.

4. A message-test plan

At minimum:

  • what the main variable is
  • what will stay fixed
  • the primary metric
  • the guardrails
  • what the team will do next if the test works

Once you have those, conversion work stops being endless headline debate and starts becoming structured learning.

Closing thought: the product may have value. The page still has to carry it.

I used to blur value proposition and messaging together as well. The more I separated acquisition, landing, and activation in real work, the less convincing that became.

Positioning decides what battle you are fighting.
Value proposition decides the value you lead with.
Messaging decides how you say it to a particular audience.
Landing decides whether the promise survives the handoff into action.

A lot of growth experiments fail because the promise is vague, too broad, or badly matched to user intent. In those situations, testing the message can be far cheaper and far more informative than piling on more product work.

The next article moves one step further downstream, into lifecycle. Not as a machine for sending more reminders, but as a way of designing reasons to come back.

References

Footnotes

  1. Nielsen Norman Group, “Homepage Design: 5 Fundamental Principles”. ↩ ↩2

  2. April Dunford, “A Product Positioning Exercise”. ↩

  3. Nielsen Norman Group, “Homepage Design: 5 Fundamental Principles”. ↩ ↩2

  4. Google Ads Help, “Optimize your ads and landing pages”. ↩ ↩2

  5. Optimizely, “How is multivariate testing different from A/B testing?”. ↩