The people looking for you asked an AI bot first. Were you in the answer?
Customers, families, funders and reporters now ask an AI bot before they contact anyone. Here’s what it costs to be left out, and what decides who gets named.
August 6, 2026Last updated October 14, 2026
Someone looked for a business like yours this week. Maybe they needed a dentist who takes their insurance, or a legal clinic that’s open on Saturdays. They didn’t scroll through ten search results. They asked an AI bot and went with the answer it gave them.
If you weren’t in that answer, nothing happened that you could see. There was no bounce in your analytics and no half-finished form. A missing customer doesn’t show up on any dashboard, which is how this has gone unnoticed while most people watched their search traffic and assumed things were fine.
Here’s what’s inside:
- Who is asking. Five people who now check with an AI bot before they contact you.
- What it costs. The customer you never hear about, and why nonprofits feel it differently.
- Why smaller organizations get skipped. What an AI bot can and can’t read about you.
- What changes the answer. Four things that decide who gets named.
Five people asking about you before they call
A customer comparing options. Someone needs a physical therapist near work, or a bakery that can make a wedding cake on short notice. They ask an AI bot who’s good nearby and call one of the names it gives them. If the bot doesn’t know what you offer, you never hear about that customer.
A family at eleven at night. A parent looking for legal help or an after-school program their child can walk to. They’re on a phone and they’re tired, so they ask in plain words. They will take the answer they get.
Someone deciding where to send money. A funder building a shortlist, or a company choosing a vendor, often starts by asking an AI bot who does this kind of work well. By the time you hear from them, a filter has already run.
A reporter on deadline. They ask who can speak to a topic, and the organizations the answer names get quoted. Being quoted makes the next answer more likely to name you, so the gap grows in one direction.
A future hire or board member. People check before they say yes. A thin, vague answer makes an organization look small, whatever its real size.
A lost customer you never hear about
For a business, the cost is a sale that went to a competitor. You can’t follow up with someone you never knew existed, and that customer might have been worth years of repeat visits.
For a nonprofit it lands differently. The person who was looking carries the cost. Some of them needed something specific on a specific night and got sent somewhere that couldn’t help.
Either way, it shows up later as numbers that miss your projections. You won’t find the cause in any system you track, because it never appeared there.
Why smaller organizations get skipped
An AI bot trusts what other sources say about you far more than what you say about yourself. Larger companies collect that outside evidence almost by accident, through press coverage and review sites.
Small businesses have a different footprint. These are patterns we see again and again in our own work:
- A Google listing filed under the wrong category, so the business never matches the searches it should.
- A website with leftover template pages, including a placeholder address that contradicts the real one.
- A booking or delivery app that describes the business more completely than its own website does.
- Hours or prices that changed months ago on one platform and nowhere else.
Nonprofits have their own version. Years of proof sit in printed annual reports and funders’ PDFs, where an AI bot can’t easily connect them to you. Meanwhile an old address or a program that ended in 2019 is still easy to find.
In both cases, years of real work are hard for the bot to see, while a thin, outdated version of you is easy for it to read.
In Spanish, the gap is wider
Ask an AI bot about local services in English and you’ll usually get a specific answer. Ask the same question in Spanish and the answer is often shorter and vaguer. It’s more likely to point to a national hotline or a big chain, and to skip the local business or organization that actually does the work.
Nothing about the business changed between those two answers. What changed is how much Spanish-language material the bot can find. A translate widget doesn’t help, because a translation generated in the browser isn’t a page anyone can find.
This matters for any business or organization whose customers search in Spanish, in the US or anywhere else. Many of them have very little written about them online in Spanish, so the bot has little to work with.
Very few people are working on this in Spanish yet.
Four things decide who gets named
Be one organization. Many businesses and nonprofits look like several to an AI bot: a legal name, a nickname people actually use, an old domain and a listing with a manager who left in 2021. Pull that into one identity with one main web address. It’s the fastest fix on this list.
Say things an AI bot can repeat. Write down who you serve and where. Add how long you’ve been doing it, which languages you work in and what things cost. Put it in plain sentences on a page. Copy written to sound inspiring at an event summarizes to nothing, because there’s nothing concrete in it for the bot to carry.
Get confirmed elsewhere. This is the hardest step and the one that decides the most answers. For a business, that means a claimed and correctly categorized Google listing, reviews, local guides and press. For a nonprofit, it means funder grantee pages, coalition directories, a Candid profile and partner pages. Each one is someone else confirming that you exist.
Check it on a schedule. Ask the questions your customers would ask, in each language they use, and write down what comes back. Do it again next month. I wrote up the method in What does ChatGPT say when someone asks about your business?
What nobody can promise you
No one can guarantee you’ll appear in an AI bot’s answer. Anyone who says otherwise is describing something they don’t control. The answers also change from week to week, so a one-time audit is a snapshot of something that keeps moving.
You’ll also hear a lot about a file called llms.txt. AI bots almost never read these files, and Google has said it doesn’t use the format. Put one up if you like, since it costs nothing. Be wary of anyone who leads with it.
The work that holds up is the plain kind in section 05. If you want to see where you stand first, Monarch’s free AI visibility check sends you the results within 48 to 72 hours.
