Website collection
Four products that each needed one clear story before a visitor scrolled.
Websites
Collection
Role
Website Designer
Timeline
Varied
team
Individual Contributor
platform
Web
The Real Problem
Four different products, the same failure. Each one knew exactly what it was for. Its website did not say it.
uman is an AI platform that covers the whole B2B sales cycle, which reads as a feature list the moment you describe it. table is an AI native personal CRM for people who would never use the phrase. night knight is a single sleep product sold to sceptical people at midnight on a phone. Finite is software supply chain security with four audiences who each want a different thing from the same page.
Breadth, a category with the wrong associations, scepticism, and competing audiences. Four problems, one homepage each to solve them.
Finding the Fix
The method was the same each time, whatever the product.
Find the spine. One sentence the whole site hangs off. Connect your data, understand your context, generate intelligence, take action. The first personal CRM backed with an AI assistant. Better sleep without the grogginess. A site that feels simpler than the technology behind it.
Order the page as a narrative. Problem, product, proof, ask. Each section answers the question the previous one raises, and the call to action sits where the decision happens rather than on every screen.
Show the real product. Actual interfaces and real interactions did more to make an AI or a security platform believable than any illustration.
Restrain the palette. One accent colour per brand (Electric Blue for uman, Electric Purple for table, violet on navy for night knight) so the product stays the focus.
Build the system, not the page. Reusable sections so role pages, industry pages, resources and future promotions cost a fraction of the homepage.
What Actually Happened
uman got a homepage built as a data to intelligence to action story, a four colour system and a modular page kit for product, role, industry, enterprise and resource pages.
table got a one sentence hero, a waitlist form with three proof points, and an AI section made of three real chat exchanges instead of a capability list.
night knight got a slow scroll in the colour of night: benefit, ingredients, how it works, reviews, FAQ and content, with trust placed next to every purchase point.
Finite got a smaller, sharper hero, product information broken into visual stories, and navigation rebuilt around the five questions a buyer actually asks.
Each has its own case study on this site.
What Changed
Across the four sites:
One product narrative per site that a visitor can follow in a single scroll
Clear entry points for each audience without splitting the site
Real product interfaces used as the hero rather than as decoration
A component system on every project that the client team can keep extending
Responsive layouts designed alongside desktop, not after
What I Had to Work With
Every one of these was a real product with real constraints, so nothing could be designed in a vacuum.
Live products with paying customers (uman, Finite), where the site could not be rebuilt in isolation
A pre launch product (table) where the only ask was a waitlist and the app was still changing
A one product store (night knight) with no catalogue to hide behind and supplement claims that had to stay careful
Multiple audiences on one site, with no persona switcher and no separate sites
AI and security messaging in years when every competitor made the same claims
Mobile first reading in every case, and enterprise credibility in two of them
What I'd Do Differently
I would test each homepage narrative with two or three real users before building the wider page system. The stories are logical; I would like to know sooner whether they match how buyers think.
I would also push further on:
Interactive product slices inside the sites rather than static screens
Role specific proof, so each audience sees outcomes for people like them
Behaviour analytics on the long pages to see where readers stop
What I Learned
Complex products need simple websites. The more the product does, the more disciplined the site has to be about what comes first.
Show the product. Real screens beat illustrations every time.
The system is the deliverable. The homepage launches once. The components get used every week after that.
