Invisible on Google and AI Search

The family you never met

Picture a family that moved to your town in June. New jobs, new schools, boxes still half unpacked in the garage. One night after dinner, one of them picks up a phone and types a plain question: what is a good church near me for families with kids.

That question used to lead to a list of church websites. The family clicked two or three, compared service times, and picked one to try on Sunday.

That is not how the question gets answered anymore. Increasingly, it gets answered before any website is opened at all. Google now writes an answer at the top of the page, drawn from sources it trusts. Chatbots answer in conversation. Maps answer with three pins and a set of reviews. If your church is not in the material those answers are built from, your church is not in the answer. The family does not reject you. They never hear your name.

This article makes a simple case. Being findable is part of a church's witness in its town. A congregation that cannot be found by the newcomer who is looking for it has a hospitality problem, and it is worth treating as one.

What changed in search

Pew Research Center tracked the actual browsing of 900 U.S. adults through March 2025, covering 68,879 Google searches. About one in five searches, 18 percent, produced one of Google's AI generated summaries. Nearly six in ten adults, 58 percent, ran at least one search that month that produced one (Pew Research Center, 2025).

The summaries change what people do next. When a summary appeared, people clicked a traditional search result in 8 percent of visits. When no summary appeared, they clicked in 15 percent of visits, nearly twice as often. Clicking a source cited inside the summary itself happened in just 1 percent of visits. And people ended their browsing session after 26 percent of visits to a page with a summary, compared with 16 percent without one. In other words, a large share of searchers read the answer on Google and stop there.

The pattern matters for churches because of which searches trigger a summary. Sixty percent of searches phrased as a question, starting with words like who, what, when, or why, produced one. So did 53 percent of searches of ten words or more. A newcomer does not search in two words. A newcomer asks a question in a full sentence, and that is precisely the kind of search that now gets an AI answer first.

This was already heading in one direction before AI summaries arrived. A 2024 study by SparkToro using Datos clickstream panel data found that 58.5 percent of U.S. Google searches ended with no click to a website at all. For every 1,000 U.S. searches, only about 360 clicks reached a website outside Google's own properties (SparkToro and Datos, 2024). Read that study for what it is, panel research from two companies in the search industry, not an official Google count. The direction it points is the same direction Pew measured a year later with a different method.

The second change is that people now ask AI tools directly. Pew found in early 2025 that 34 percent of U.S. adults say they have used ChatGPT, roughly double the share in the summer of 2023. Among adults under 30, the figure is 58 percent (Pew Research Center, 2025). BrightLocal's Local Consumer Review Survey, published in February 2026, found that use of ChatGPT and other generative AI tools for local business recommendations rose from 6 percent in 2025 to 45 percent in 2026, while Google remained the most used source for local recommendations at 71 percent, down from 83 percent (BrightLocal, 2026). BrightLocal sells services to local businesses, so weigh its survey as a vendor survey of 1,002 U.S. consumers. Even with that caution, a jump from 6 to 45 percent in one year is not noise. Asking an AI for a local recommendation has gone from a curiosity to a habit in twelve months.

Why a church website decides the answer

Here is the part that surprises most pastors. The AI answer about your church is not built from your reputation in town, your history, or the faithfulness of your preaching. It is built from what machines can read about you: your website, your Google Business Profile, your reviews, and the directories that copy your details around the internet. When those sources are thin, stale, or in disagreement with each other, the machine has little to work with, and it works with the churches it can read instead.

Google is plain about the local half of this. In its own guidance, Google says local results rest mainly on relevance, distance, and prominence, and that businesses with complete and accurate information are more likely to show up in local search results (Google Business Profile Help). Relevance is the one your church controls directly. It is built from the details you publish: your name, your address, your category, your hours, your service times.

Google is equally plain about the AI half, and its plainness is a kind of relief. There is no secret AI trick to buy. In Google's documentation for its AI features, a page qualifies to appear as a supporting source in AI Overviews and AI Mode by being indexed and eligible to appear in ordinary search with a snippet. There are, in Google's words, no additional technical requirements. The best practices Google lists are unglamorous: let Google crawl the site, put important content in readable text rather than only in images or PDFs, keep structured data consistent with the words on the page, and keep the Business Profile up to date (Google Search Central, AI features and your website).

Structured data deserves one plain paragraph, because it sounds more mysterious than it is. It is a small block of machine readable facts attached to a page, stating in a standard format that this is a church, at this address, gathering at these times. The standard vocabulary of the web, Schema.org, has a dedicated type simply called Church, with fields for address, location, and opening hours (Schema.org, Church). Google describes structured data as giving explicit clues about the meaning of a page, so the search engine does not have to guess (Google Search Central, structured data). Guessing is the risk. A service time buried inside a scanned bulletin PDF, or drawn into a graphic, is a fact a machine may never pick up. The same time, written as ordinary text on the homepage and repeated in structured data, is a fact it cannot easily miss.

The people on the other side of the search

It would be easy to treat all of this as a technical matter for whoever runs the website. The research on who is searching says otherwise.

Pew's study of Americans choosing a new congregation found that about half of U.S. adults have looked for a new place of worship at some point, and the most common reason was a move. Among those searchers, 70 percent said location was an important factor in their choice, and 79 percent said feeling welcomed by clergy and lay leaders was important. Younger searchers went online as a matter of course: 59 percent of adults under 30 incorporated online searches into looking for a congregation, compared with 12 percent of those 65 and older (Pew Research Center, 2016).

Lifeway Research found the mover's search is even more online than the average. In its 2023 study of Protestant churchgoers who had changed churches, those whose switch followed a residential move used church websites at a rate of 40 percent, and movers were more than twice as likely as other switchers to have been first introduced to their church through its website, 20 percent compared with 8 percent. Lifeway also found that 41 percent of switchers had stopped attending church for three months or more between congregations (Lifeway Research, 2023). That last number is the one to sit with. The months after a move are a fragile window. A family that cannot find a church quickly may not find one slowly either. Some of them drift, not because they stopped believing, but because nobody was easy to find at the moment they were looking.

A witness issue, not a visibility trick

Jesus describes his people as a city on a hill, and says a lamp is put on a stand so it gives light to everyone in the house. "In the same way, let your light shine before others, so that they may see your good works and give glory to your Father who is in heaven" (Matthew 5:16, ESV). The point of the lamp is not the lamp. Nobody lights one to admire it. The point is that people in the house can see.

Findability works the same way. No pastor should lose sleep over rankings as a scoreboard, and this article is not arguing that a church should chase an algorithm the way a retailer chases a sale. The argument is narrower and older. Your town is full of people who, at some point in a hard year, will go looking for a church. They will look the way people in 2026 look for everything, by asking Google or asking an AI. When they ask, the answer they receive is assembled from the public facts your church has published about itself. Keeping those facts accurate, complete, and readable is closer to putting the right name on the sign by the road than it is to advertising. A sign with the wrong service time, or no sign at all, is not humility. It is a missed neighbor.

The lost sheep in Luke 15 is found because the shepherd goes looking. The modern version runs in the other direction as well. The newcomer goes looking, phone in hand, and the question is whether the church has made itself findable by someone searching honestly.

What keeping findable actually involves

None of this requires a large budget or a staff member assigned to search engines. It requires accuracy, kept up in a handful of places.

• One set of facts, everywhere. Your church name, address, phone number, and service times should read identically on your website, your Google Business Profile, your Facebook page, and the major directories. Disagreement between sources is the most common reason machines hedge or omit a church.
• Service times as text on the homepage. Not only in a bulletin PDF, not only inside an image. Text a guest can read is also text a search engine can read.
• A claimed and current Google Business Profile. Hours updated for holidays and special Sundays, photos that look like your actual building, and reviews responded to with grace.
• Structured data that matches the page. The machine readable facts and the visible facts must say the same thing. Google asks for that consistency directly.
• Pages that answer real questions. What time, where to park, what happens with children, what a first Sunday looks like. Written in sentences, these are the pages both a guest and an AI summary can draw from.
• A seasonal check. Service times change, ministries move rooms, pastors retire. The facts need an owner and a rhythm of review, or they quietly go stale.

Read the list again and notice what it is. It is not a growth tactic. It is the digital version of keeping the sign painted, the parking lot lights working, and the front door unlocked before the first guest arrives.

The quiet conclusion

Your church will be described to newcomers in the next year, by Google and by AI tools, whether you take part in that description or not. The only choice is whether the description is built from accurate, complete, cared for information, or from whatever stale fragments the internet happens to hold. Shepherding a church in this decade includes caring for the facts by which a stranger first meets it.

If you want to see what your church looks like when those facts are gathered, ordered, and put on a homepage built to be found, Tov will build you a free demo of your homepage, and you can judge it before you spend anything.