Your Website Is Now for Robots as Much as Humans

Picture the last time you asked the internet a question. Maybe it was “how much water should I actually drink” or “is my houseplant dying or just dramatic.” You typed it into Google, and before you could reach for a single blue link, a tidy little AI summary appeared at the top and told you the answer. You nodded. You closed the tab. You never clicked a thing.

Congratulations. You just did to some website exactly what strangers are now doing to yours.

For about fifteen years, the job of a nonprofit website was gloriously simple. Get found in search. Earn the click. Convert the visitor. We built entire strategies around feeding the top of that funnel, and the prize at the top was traffic. More arrivals, more chances to turn a curious stranger into a devoted supporter. It worked. Then the search engine learned to answer questions itself, and the whole tidy machine started coming apart while most organizations were looking the other way.

The great divergence, or: why the numbers stopped making sense

Here’s the pattern that should make you sit up. In 2025, the fundraising firm M+R watched a group of nonprofits live through the rollout of Google’s AI Overviews, those summaries that now perch at the top of your results and answer the question before you can click anything. By August, those organizations were seeing roughly a 13 percent year-over-year drop in visitors from organic search. Meanwhile, searches for those same organizations’ names rose about 19 percent.

Sit with that for a second, because it’s genuinely weird. More people were looking for these nonprofits by name. Fewer people were actually landing on their websites. The audience grew and the room emptied out at the same moment, which is the kind of sentence that makes a development director quietly close their laptop and go for a walk.

The culprit is that the search engine now does the answering. Pew Research Center watched real people browse in 2025 and found that when Google served up an AI summary, folks clicked a regular search result only 8 percent of the time. No summary? They clicked about 15 percent of the time. Roughly a quarter of the searches with an AI summary ended right there, with no click at all. The answer was good enough. The visit simply never happened.

Try this now

Open a new tab, search your organization’s name, and see whether an AI summary shows up before any of your actual pages. Notice what it says about you. We’ll come back to that summary in a minute, and it’s going to be important.

What this actually means for your website

The easy reaction is panic, and the internet has an ample supply. “SEO is dead.” “You need an AI strategy immediately.” Most of that noise ends with someone trying to sell you something with a subscription fee. You can ignore it.

Here’s the calmer read. Your website’s purpose is shifting, not vanishing. It used to be a net you cast wide to catch strangers. Now it’s doing two more specific jobs, and honestly, both of them matter more than the old one.

First, it’s where already-motivated people decide whether to trust you. The person who searched your name, read the AI summary, and clicked through anyway is not a tire-kicker. They came looking for you on purpose. That’s a warmer visitor than the old funnel ever coughed up, and your site should be built to help that person act, not to flatter an algorithm.

Second, and this is the part that keeps us up at night in the good way, your website is now the source AI tools quote when they introduce you to a stranger. When someone asks ChatGPT or Google’s AI what your organization does or whether it’s worth supporting, the answer gets assembled from whatever those systems can read on your site. If your “About” page is thin, your impact numbers are from an administration ago, or your programs are entombed in a PDF, that’s the version of you the machine cheerfully repeats to a potential major donor. You don’t get a chance to review it first. There is no approval workflow. The robot just talks.

The part almost nobody connects

At ArcStone, we’ve spent 29 years building websites for mission-driven organizations, and the whole time our first commitment has been accessibility. Clean, semantic structure. Real headings instead of text that’s just big and bold and hoping. Content a screen reader can move through in a sensible order. Descriptions on images. Plain language that says what it means.

And here’s what stopped us in our tracks over the last year. The exact things that make a website work for someone using a screen reader are the exact things that make it legible to an AI. A machine trying to summarize your mission and a blind donor trying to reach your donation page are both leaning on the same foundation: is this content structured, labeled, and written clearly? Or is it a beautiful visual soup that only makes sense if you can see it and squint and guess?

That overlap is not a coincidence. Accessibility and AI-readability are the same discipline wearing two different hats. The organization that did accessibility properly is, without meaning to, already ahead on being understood and quoted accurately by machines. The organization that faked it with a widget and a line of code is now exposed on both fronts at once, which is a lot of exposure for something that promised to be effortless.

We find that deeply satisfying. It means the honest, unglamorous work we’ve always argued for just quietly started paying a second dividend.

What to do about it this quarter

You don’t need to blow up your website. You need to shift what you measure and where you point your attention.

Stop treating raw traffic as the headline number. A 13 percent dip in organic visitors isn’t a catastrophe if the people still showing up are likelier to give, volunteer, or sign up. Watch what visitors do once they arrive, not just how many of them there are. Quality of arrival is the new game.

Remember that AI summary from a few minutes ago? Now go further. Ask ChatGPT or Google’s AI the questions a real supporter would ask, and grade the answers honestly. Most people discover the machine is quoting a stale page, an old annual report, or a program they sunset two years ago. Whatever it’s pulling from, that page is now your introduction to strangers. Fix that page first.

Make your most important pages your clearest ones. Your mission, your impact, and your ways to give should be easy for a person to read and easy for a machine to parse. Real headings. Current numbers. Short paragraphs. A FAQ where it earns its keep. This is the same checklist we’d hand you for accessibility, which is very much the point.

Invest in the moment of decision. The visitor who pushed past the AI summary to reach you is ready to act. Don’t make them hunt for the donate button, translate your program names, or wrestle a form that fights back. That one conversion path deserves more of your budget than another round of chasing clicks into a headwind.

The honest bottom line

The web is changing in a way that rewards the organizations that did the fundamentals well and gently punishes the ones that cut corners. That’s a good outcome for the sector, even when the transition feels like someone rearranged your furniture in the dark.

Your website was never really for the algorithm. It was always for the person who cares about your work. The algorithm just did a decent job of connecting the two for a while, and now that it’s stepping back, the quality of what you actually built is what shows through. To a person and to a machine alike.

If you’re not sure what the machines currently see when they look at your site, that’s worth finding out. It’s usually a faster fix than people expect, and it’s almost always the same fix that makes your site better for everyone.