Our new site asks who you are before it says a word

As you land on our new home page, there's a moment right at the start of the journey where nothing else happens until you make a choice. Two audiences, side by side - marketing teams and tech businesses - and the site politely declines to continue until it knows which of them you are. We debated that moment more than any other part of the build, because every instinct in web design says reduce friction, never gate content, let people wander. We added the friction anyway, we asked the question, and this piece explores our why.

The short version is that we stopped believing in the generalist website. The long version involves what we've watched happen to attention, to trust and to conversion over the last few years - on our clients' platforms first, and eventually, honestly, on our own.

The polite lie of the general audience

Most B2B websites are written for a reader who does not exist: a composite of every possible visitor, addressed in language broad enough to offend nobody and specific enough to persuade nobody. The services page lists everything, the case studies hedge toward the middle, and the copy performs a kind of averaging - which feels safe, and reads like fog.

The uncomfortable part is that we knew better, because segmentation is the first thing we bring to client work. We'd spent years building personalised journeys for retailers and property platforms, watching relevance move numbers that design polish alone never touched, and then we'd come home to our own site - one message, one journey, everyone welcome, nobody addressed. The cobbler's children needed shoes.

A marketing director and a CTO arrive at our site with different vocabularies, different anxieties and different definitions of a good outcome. One is carrying a campaign calendar and a website that can't keep up with it; the other is carrying a platform that's outgrown its architecture and a board asking about AI. Serving them the same paragraph means translating for neither, and readers can feel when they're being averaged.

What the site actually does

The answer we shipped is structural, not cosmetic. When you choose an audience, the site commits: the services menu reorganises around what that audience actually buys from us, the case studies lead with the proof that speaks to their world, and the whole visual temperature shifts - warm pinks for marketing teams, cool cyans for tech businesses - so even peripheral vision knows which conversation you're in.

The choice also persists. Land on a service page from a search result next week and the site remembers who you are, because the audience lives in the URL structure itself (/digital for marketing teams, /tech for tech businesses) rather than in a fragile personalisation layer. Deep links arrive already dressed for the right reader, which matters more than it sounds - most of a site's visitors never see the front door.

Personalisation isn't a widget you add. It's a structure you commit to.

Underneath, there's one catalogue of everything we do, one taxonomy connecting services to the case studies that prove them, and one design system rendering it all - so the two site personalities can't drift apart, and adding proof in one place surfaces it everywhere it's relevant. The engineering principle is the same one we push on client platforms: personalisation should be a projection of one source of truth, never a second copy of the content.

The friction question, answered with numbers

About that gate. The objection to asking visitors anything is that some of them will leave, and the objection is correct - some do. The question is what happens to everyone who stays, and there the trade has been lopsided: a reader who self-selects spends the entire visit inside copy written specifically for them, and relevance compounds page after page in a way that a broad site never gets to.

We've watched the same trade pay off in client work for years. Journeys that ask an honest early question (what are you here to do, who are you doing it for) consistently outconvert journeys that guess, because the cost of the question is paid once and the benefit is collected on every subsequent screen. Choice is only friction when it doesn't buy the chooser anything.

Generic is the expensive option now

There's a bigger argument underneath the build, and it's the one worth taking away even if you never visit either of our audiences' journeys. The web has quietly inverted: producing generic content used to be the cheap option and tailoring was the luxury, and now (with tooling, with structured content, with AI drafting at the edges) the marginal cost of speaking specifically has collapsed, while the cost of speaking generally keeps climbing - paid for in bounce rates, in shortened attention, and in the deep scepticism readers have developed toward anything that smells averaged.

Every audience you decline to address specifically is being addressed specifically by someone else. That's the whole calculus, and it applies to a studio's own site as surely as it applies to the platforms we build for clients - which is why we finally applied it to ourselves.

The new site is our argument made in public, and like any argument worth making, it's measurable: we'll be watching how the two journeys convert against the old general one, and we'll publish what we learn. If you haven't yet, go answer the question - and notice how differently the site treats you afterwards.