For years the goal was a rank. You wanted to be the third blue link for “church near me,” and if you got there, people showed up.
That is not how most people find anything now. They ask a question in full sentences, and something hands them an answer with two or three sources under it. Nobody scrolls a list. The question is no longer whether you rank. It is whether you are one of the sources the answer got built from.
That distinction changed how I write everything, and it started with an uncomfortable audit of my own site.
What “findable” means when a machine answers first
An AI answer is assembled, not retrieved. Something reads a pile of pages, finds the sentences that plainly state a thing, and stitches them into a response. Which means the unit of value is no longer the page. It is the quotable sentence.
That has a blunt implication for how most of us write. The classic ministry blog post opens by setting the scene, tells a story, builds to a point somewhere around paragraph nine, and lands the actual answer in the conclusion. It is a fine essay. It is nearly useless as a source, because there is nothing in the first screen that says anything.
So I inverted it:
- The heading is the question, phrased the way a person types it. Not “Rethinking Digital Community” but “How do I keep an online small group from dying after a month?”
- The first paragraph answers it. Completely, in two or three sentences, before any story. If someone reads only that, they got what they came for.
- Then the story, the nuance, the caveats. Those are what make the page worth staying on. They are not what makes it findable.
It feels like giving away the ending. It is. That is the point.
The duplicate problem nobody warns churches about
Here is the part I got badly wrong, and it cost me more than any keyword decision.
I was publishing the same content in three places without realising it. One message became a sermon page, a blog recap, and a podcast episode page. Three URLs, three titles, one piece of actual content. Multiply that by a few years of faithful publishing and you get what I found when I finally measured it: 61.9% of my pages read as near-duplicates of something else on my own site.
To a crawler that is not thoroughness, it is ambiguity. Three pages saying one thing means something has to choose which one represents you, and it will not ask your opinion. Often it chose the thinnest of the three.
The fix was unglamorous. I picked one canonical home for each thing — the episode page for episodes, the guide for topics — and pointed everything else at it with rel=canonical instead of letting them compete. Near-duplicates went from 61.9% to 3.4%.
If you do one technical thing after reading this, do that one. Not because duplicate content carries some penalty, but because you cannot be quoted clearly if you have not decided what you actually say and where you say it.
The four-minute version
If you want to test your own site this week:
- Pick your five most important questions. Not topics — questions, in a real person’s words. “Do I have to be baptised again?” beats “Baptism.”
- Search for each one the way someone would ask it, in whatever AI tool you use. See whether anything of yours comes back.
- Find the page that should have answered it. Read its first paragraph. Does it answer, or does it warm up?
- Check for twins. Search your own site for the same topic. If three pages cover it, decide which one is home.
- Rewrite one first paragraph. Just one. Put the answer in it.
That is a morning’s work, and it is worth more than a year of keyword research — because the questions people ask a machine at 1am are the ones they would never ask your greeter on Sunday. Those questions are the actual front door now, and the answers are either yours or somebody else’s.
Questions people ask
How do I get my church website to show up in AI answers? Answer one question per page, in plain words, in the first paragraph — then keep that answer in exactly one place. AI answers are assembled from pages that state something clearly enough to quote. A page that circles a topic for 800 words before saying anything has nothing quotable in it.
Is SEO still worth it for churches if people use AI instead of Google? Yes, but the goal moved. You used to want a rank; now you want to be the source an answer is built from. The work is the same in kind and different in shape: clear answers, one canonical home per topic, and questions phrased the way people actually type them.
What is the most common mistake churches make trying to be found online? Publishing the same thing twice. A sermon page, a blog recap and a podcast page all covering one message look like three competing pages to a crawler, so it picks one and often not the one you wanted. Pick a canonical home for each thing and point the others at it.
This is part of the Digital Evangelism edge. The full guide is Digital evangelism for churches.
