Native Advertising Agency · Cologne
6H Agency
Native advertising buys cold traffic. Someone is reading an article, sees a promoted tile, and clicks with no idea who your client is. The creative earns the click; the landing page has to earn everything after it. My job was the page, and the system that let a media buyer get a new one live fast enough to test.
See LiveThe problem
6H is a performance agency in Cologne running native campaigns on Taboola and Outbrain for brands across food, health and education. Native traffic behaves differently from search traffic in one decisive way: nobody arrives looking for the product. There's no intent to meet, no brand recognition to lean on, and the visitor is usually on a phone, mid-scroll, one tap from leaving.
That puts the entire burden on the landing page. Media buyers can optimise bids and placements all day, but if the page loads slowly or fails to make its case in the first screen, the money is already gone. Every page is measured against one number: what it costs to acquire a customer.
Design here isn't judged on taste. It's judged on whether it moved the cost per acquisition.
Design approach
I joined as a web designer and developer, which in practice meant designing the page and building it, then handing the media buyers something they could put spend behind the same week.
Design for a reader who has never heard of the brand
A landing page for cold native traffic can't open the way a brand site does. There's no logo recognition doing the work, so the page has to state the offer, prove it, and remove the obvious objection before asking for anything. On the projects I worked on, including web work for HelloFresh, that meant the first screen carried the proposition and the proof, not the brand statement.
Speed is a design constraint, not an engineering afterthought
Native traffic is overwhelmingly mobile, often on ordinary connections. A heavy hero image or a slow-loading font is not a craft detail, it's a measurable share of the traffic gone before the page paints. Building the pages myself rather than handing off a file meant the performance cost of a design choice was visible while I was still making it.
Velocity beats perfection when the page is a test
In performance marketing a landing page is a hypothesis. Most will be replaced. Designing each one as a bespoke artefact would have starved the buyers of variants to test, so the work optimised for how quickly a credible new page could exist.
A component system, not pages
The most useful thing I built there wasn't a page, it was the set of reusable Webflow components the pages were assembled from: hero variants, proof blocks, offer panels, comparison tables, form and CTA patterns.
The point was throughput. Once the pieces existed and were known to hold up, a new campaign page became an assembly job rather than a build from scratch, and the parts that had already proven themselves got reused instead of re-argued. It's the same instinct behind the design systems I work on now, arrived at from the opposite direction: not "we should have a system" but "we cannot ship enough variants without one".
The other side of the account
Not all of it was acquisition. I also designed dashboards and internal analytics tools for the agency's marketing clients, the screens where the campaign results actually get read, and owned the product requirements and prioritisation for those client-facing dashboards.
Working both ends of the same account was the useful part. I was designing the page that spent the budget and the dashboard that reported on it, which makes it much harder to treat a metric as an abstraction. A number on a dashboard was one I had watched a landing page produce.
The unglamorous half
I also supported the media buyers on the creation of the advertisements themselves, and I owned the agency's advertising spreadsheet, the record of which ads were live and which had been switched off across campaigns.
It sounds like admin, and it partly is. It also turned out to be the clearest window into the work I've had. Sitting with the tracking meant seeing which creative and page combinations were still running and which had been killed, which is a far more honest feedback loop than a design review. It's where I learned to expect my work to be measured.
What they said
“Mr. Akinbode was responsible for creating landing pages, developing reusable components in Webflow, and supporting our media buyers with the creation of advertisements. He also worked on web projects for major clients, including HelloFresh, where he showed a good sense of design and technical execution.”
Quoted from a written reference dated September 2025. The full letter is available on request.
What it taught me
This was six months early in my career, and it's on this site for one reason: it's where design stopped being a matter of opinion for me. Every page had a number attached, and the number didn't care how the page looked.
The habit stuck. It's the same reasoning behind the search architecture on NorthStar Surgery, where the question was never "does this look right" but "does this get found and does the next step get taken", and behind the payout table on Klærus, where showing the losing outcomes was a conversion decision as much as an ethical one.
An honest note on numbers. 6H publishes strong agency-wide results, including an average CPA reduction across its campaigns. Those are the agency's numbers over years of work by its media buyers, not mine, and I've deliberately left them off this page. My contribution was the pages and the component system behind them during six months in 2022.