Stop Making Bots Download the Brochure

We still talk about websites as if the main visitor is a person with a coffee, a thumb and a little patience.

Increasingly, that is not true.

AI search, shopping agents and web crawlers are becoming a material audience for brand, product and sustainability information. Google has already pushed AI Mode towards shopping and agentic checkout. Cloudflare has gone the other way, building tools so site owners can monitor, block or charge AI crawlers, after reporting extreme crawl-to-referral ratios for some AI companies.

That should change how digital sustainability is understood. Not as a vague argument for fewer pixels, but as a very practical question: why are we making machines download the brochure?

The new visitor is not looking at your page

A human may need the video, the campaign photography, the interactive chart, the executive quote and the emotional proof that a brand has substance.

A machine usually does not. It wants a price, a claim, a certification, a location, a repair instruction, a return policy, a source date or a product attribute. It may never see the visual design at all.

Check it. 👇

The reveal is that a growing share of this traffic is not a better customer journey. It is automated extraction, comparison and summarisation hitting the same heavy assets built for persuasion.

That matters because the web is not exactly getting lighter. The 2025 HTTP Archive Web Almanac found the median desktop page had reached 2.9 MB, up from the previous year. One page is not the problem. Repeated machine access to millions of overweight pages is a different conversation.

Page weight is now part of the AI stack

Most AI sustainability debate focuses on data centres, and for good reason. The IEA’s latest update says data centre electricity use grew sharply in 2025 and projects consumption roughly doubling by 2030, from 485 TWh to around 950 TWh.

But there is a quieter layer below that debate: the everyday digital estate that agents crawl, fetch and interpret.

If a brand’s sustainability information is trapped inside heavy pages, PDFs, image-led explainers and campaign microsites, machines will either work harder to extract it or ignore it. Neither outcome is especially useful.

This is not an argument for killing design. It is an argument for knowing what each layer is for.

The next sustainable website will not be the one with the prettiest green page. It will be the one that knows when to serve a human experience and when to serve a machine-readable answer with almost nothing attached.

The human layer can still carry narrative, context and emotion. The machine layer should carry clean facts: product data, claim boundaries, provenance, availability, certifications, methodology, contact points and update dates.

Build for agents without bloating the estate

The practical move is simple but uncomfortable. Audit the pages that answer engines, shopping agents and crawlers are most likely to hit, then ask whether they need to be pages at all.

Some should become lighter HTML. Some should become structured product feeds. Some should become API-accessible facts. Some should remain rich human experiences, but with the machine-readable answer separated from the theatre.

Marketing teams should also stop treating bot traffic as an SEO curiosity. It is becoming a cost line, a performance issue and a sustainability variable.

The strategic implication is clear: digital sustainability is moving from “make the website greener” to “serve the right format to the right kind of reader”.

If the reader is a person, design for attention. If the reader is a machine, do not make it download your brand film just to learn whether the jacket is repairable.