When someone asks an AI a hard question about God, the answer gets assembled from the pages that were willing to answer it. Most church websites never wrote those pages. They tell you where to park and what time the service starts… and they go quiet on the questions people actually type at 1am. So the answer gets built from someone else’s words.
I’ve written about this edge twice already. The first post was about being quotable: put the question in the heading, the answer in the first paragraph, and stop publishing the same thing three times. The second was about what happens when a real person asks you directly, in a comment or a DM. This one sits in between. It’s about the question nobody asks you, because they asked a machine instead, and your church had nothing on the record.
What do people actually ask a machine about faith?
Not “churches near me.” Well, some of them. But the searches that matter for evangelism sound more like this…
Will God forgive what I did? Can I be a Christian and still doubt most of it? Why would a loving God let my dad die like that? Is it hypocritical to go back to church after everything? What does the Bible actually say about anxiety? Do I have to believe all of it?
Those are the shapes of real questions. I’m not quoting anyone, and you don’t need me to. You’ve heard every one of them in a hallway or a hospital room. The difference now is that most people ask them somewhere else first.
Think about the week. Your church gets one hour of it on a good Sunday. The search box gets the other 167.
Why doesn’t my church have a page for that?
I’ll go first.
It is far easier for me to write about strategy than to write a page that answers whether God hears a prayer you can’t finish. Strategy has no wrong answer that hurts anybody. A page about grief or shame can be wrong in a way that lands on a real person, and that fear has kept me writing the safer post more often than I’d like to admit.
I suspect your website has the same fear baked into it.
Church websites get built by committee, and committees optimise for not getting it wrong. So the site says what everyone already agrees on. A beliefs page written for people who already believe it. Staff bios. A ministries grid. An events calendar that’s out of date by Tuesday. Nobody decided to avoid the hard questions. Nobody was assigned to answer them either, so the site drifted toward the things nobody could argue with.
And a machine reading your site can only quote what is there.
Pastor, if someone asked an AI tonight what your church says about doubt, what would it find? Would it find anything at all? Or would it find a statement of faith that never once admits faith and doubt usually live in the same house?
Isn’t this just the old content-marketing advice?
Fair objection. “Write blog posts that answer questions” has been on every church marketing checklist for fifteen years… so isn’t this the same thing with an AI coat of paint?
Partly. The craft is the same. What changed is the cost of skipping it.
When search was a list of links, a church with no page on doubt simply didn’t show up, and the person scrolled to something else. Now there’s no scrolling. The person gets one assembled answer, in a confident tone, and it reads like settled truth. If your voice wasn’t in the pile it was built from, it isn’t in the answer. The person never knows you were missing. They just walk away with whatever the pile said.
Some of that pile is good. Some of it is written by people who would happily tell a hurting person that God is done with them, or that church is the last place they should go. A lot of it is written by nobody in particular, for clicks, and reads like it.
Church, we used to be the place people brought the hard questions. If we aren’t on the record when they ask, we’ve quietly handed that role to whoever was.
What makes a page like this worth quoting?
It actually answers.
The first paragraph says the thing. If someone asks whether God will forgive what they did, your first paragraph speaks to that shame directly and tells the truth as your church holds it, with compassion, before it says anything else. After that comes the nuance, and Scripture if it helps… one line that opens a door, not a lecture that closes it. And somewhere on the page, a real way to talk to a person. I don’t mean a contact form.
Write it in the voice you’d use in the hospital room, not the one you’d use in a membership class. A machine can tell the difference between a page that is answering and a page that is defending itself. So can the person reading it.
Keep it to one page per question. One. The temptation is to write a series, a sermon recap and a resource page that all half-answer the same thing… and then nothing on your site is the clear answer to anything. I learned that one the hard way, which is the whole story of that first post.
And put a name on it. A pastor’s name, a real byline. An answer about grief from “Admin” reads like a policy. An answer from a person reads like a person.
Who should write these pages?
Not the communications person alone. They know how to make a page work. They usually don’t know what was said when a family sat with your pastor after a funeral.
The best version I can picture is a pastor talking and someone else typing. Sit down with whoever does your web work, take one question, and answer it out loud the way you would if a stranger asked you over coffee. Record it. Let them shape it into a page. AI can help with that shaping, tidying a transcript into readable paragraphs or suggesting a heading phrased the way people actually search. It can prepare the page. It cannot be the pastor on it. The words that matter have to be yours.
That’s the digital evangelism work hiding inside all the talk about AI and search. Being findable comes down to being willing to say something true, in public, before anyone asked you to your face.
If you serve people who have walked away from church, this matters double. The de-churched are the people least likely to ask a pastor and most likely to ask a machine. There’s more on that in reaching the de-churched.
What should I do this week?
Pick the one question you most hope nobody asks you. You already know which one it is.
Ask it in whatever AI tool you use, the way a hurting person would type it, with no church words. Read what comes back. Notice whose voice built that answer, and whether yours is anywhere in it.
Then write the page. One question, answered in the first paragraph, with your name on it.
What would it mean if the next person who typed that question at 1am found your church already there, with an honest answer waiting? Write that one page this week… then sit down with your team and ask which question comes next.
This is part of the Digital Evangelism edge. The full guide is Digital evangelism for churches.
