When it pays off
and when it doesn't

When it's slow, when it loses people on mobile, or when it no longer says what you sell today. Not because it's “old.”

Google commissioned a study from Deloitte that measured mobile sites in retail, travel and luxury for four weeks. A speed-up of one tenth of a second raised retail conversions by 8.4% and average order value by 9.2%. In travel, conversions rose by 10.1%.

A tenth of a second, not three seconds. If your site loads slowly, you aren't losing traffic. You're losing orders you have already paid for.

It does, but not because you like how it looks. It decides whether the visitor believes you.

Stanford's web credibility research looked at what people go by when they judge whether a site can be trusted. 46.1% of all assessments relied on visual impression, meaning layout, typography and colors. For financial services it was 54.6%. The accuracy of the information itself came up in only 14.3% of assessments.

Design is not decoration stuck onto content. It's the first filter the content has to pass to reach the reader at all.

Less often than an agency will tell you. But in small steps, more often than you'd like.

Jakob Nielsen measured that the average homepage changes by about 40% a year, which adds up to a complete redesign once every three years. He also points out something you won't find in proposals. A new design tends to perform worse than the old one at first, because it breaks the habits users have had time to build.

A complete redesign every year is both wasteful and risky. Ongoing changes based on what the data shows are cheaper and safer.

Longer than is comfortable. And most pages never get there at all.

Ahrefs tracked a million new URLs. Within a year, 1.74% of them reached the top ten in search. The second number is the more interesting one, though. Of those that made it, 40.8% got there within a single month.

Either it works fast or it doesn't work. If a page is nowhere after six months, time is not the problem. The content is, or the search demand it's aiming at.

SEO alone is starting to fall short. First place on Google isn't what it used to be.

Ahrefs compared 300,000 keywords, half with an AI answer above the results and half without. When an AI answer appears above the results, the click-through rate of the top-ranking page is 34.5% lower on average.

Being first is no longer enough. You also need to be cited. That's a different discipline. Structured data, unambiguous answers, named sources. The page you're reading right now is built that way on purpose.

When it works out in a simple calculation. Not when someone shows it to you on a nice chart.

I don't trust numbers from automation vendors and neither should you. Zapier, for example, claims that marketers save 25 hours a week through automation. That's more than half a working week, and the number comes from a company that sells automation. Better to work it out yourself. Take the hours the task eats up in a month, multiply them by the hourly rate of the person doing it, and compare that with the one-time price of the automation plus its maintenance.

If the investment doesn't pay for itself within 12 months, don't automate it. And only automate what is boring, repetitive and rule-based. Everything else breaks at the first exception.

Source of the number I don't trust: Zapier, survey of 1,500 workers at small and midsize businesses in the U.S.

When you put it on top of a broken process. Then you get a broken process, only faster.

McKinsey surveyed companies around the world. 88% of them use AI in at least one area. But only about 6% see significant value from it, and what those have in common is that they reworked the workflows themselves while rolling AI out. Overall, 39% report a company-wide impact on profit, and most put it at under 5%.

The difference between 88% and 6% is not in which tool the companies bought. It's in whether they were willing to change how they work. That's why I sometimes talk you out of AI, even when I could sell it to you.