ChatGPT recommends Porch 419 times. Porch sells insurance now.

Porch showed up 423 times in our citation data. Four hundred and nineteen of those came from ChatGPT. Three came from Perplexity. One came from Google AI. That is 99 percent of a source's total visibility riding on a single engine, which is strange enough on its own. Then we opened the site to see what ChatGPT was actually sending homeowners to, and it got stranger. Porch.com in 2026 leads with home insurance.

The short answer: Porch has not shut down and its Pro Network still runs, but porch.com now sells home insurance, warranty and moving as its front door, and every contractor signup path we tested returns a 404 — so ChatGPT is confidently describing a marketplace experience that no longer matches what a homeowner finds.

What we actually checked, and what we found

This is the last post in a ten-part series, so I want to be careful here rather than clever. An earlier version of this article said Porch had shut down. That was wrong, and it got corrected before it ran. Here is what is verifiably true as of today.

So: the lights are on, the marketing page is live, and an existing pro can log in and buy leads. But the front door sells insurance, the parent company earns roughly five dollars of insurance revenue for every dollar of consumer services revenue, and a new plumber trying to join through the obvious route hits a dead page.

A small aside that makes the point better than I can. Porch Group's own companies page still lists an insurance carrier it sold into its reciprocal exchange in early 2025. If the company itself cannot keep its own pages current, think about what chance a language model trained on a scrape of the internet has.

What ChatGPT thinks Porch is

When ChatGPT points a homeowner at Porch, it is describing something close to 2019. A busy contractor marketplace where you post a project, get matched with three or four local pros, compare quotes. That was a real product. Parts of it still exist. But it is not what the brand leads with, it is not where the money is, and it is not the experience the homeowner gets when they land on the homepage and see a pitch for member-owned home insurance.

The model is not lying. It is repeating the version of the internet it learned, and that version is several years old. Nothing in the model's process involves checking whether the page still says what it used to say. It reproduces the old answer in exactly the same confident tone it uses for things that are true today, because confidence is the only tone it has.

That is the whole lesson, and it is worth more to you than anything about Porch specifically.

Now apply that to your own company

If ChatGPT is years behind on a venture-backed public company with a marketing department and an investor relations page, consider how current it is on a six-truck plumbing company in a mid-size metro.

Here is what "years behind" looks like when it is about you:

• The phone number you dropped in 2022, still being handed to homeowners who ask for a plumber near them.
• The old shop address, three miles from where you actually are now, which quietly puts you outside the service area someone is searching in.
• A one-star review from 2021 about a tech who has not worked for you in four years, summarized as if it were representative.
• Hours that say you close at five when you have run 24/7 emergency service for two years.
• Or the most common one: nothing at all. The model has never encountered your company in a context it can repeat, so it names three competitors instead.

None of that is hypothetical. It is the same failure mode we just documented on Porch, aimed at a smaller target with less ability to correct it.

The audit: find out what AI currently says about you

Do this yourself. It takes about twenty minutes and you do not need us for it.

1. Open a fresh chat in ChatGPT with memory and personalization turned off, so you get the generic answer a stranger gets, not one shaped by your history.
2. Ask it: "Who are the best plumbers in [your city]?" Write down every company it names. Note whether you are one of them.
3. Ask it directly about your company by name. "Tell me about [your company] in [your city]."
4. Check what it produces against reality, field by field: phone number, address, hours, service area, whether you do emergency work, whether you handle water heaters, sewer, repipes.
5. Repeat all of that in Perplexity, in Gemini, and in Google's AI Mode. The answers will differ, sometimes a lot. Note which sources each engine cites — that is your target list.
6. Search your company name and city in a plain Google search and open every result on page one. Directories, old listings, aggregator profiles. Every one of those with wrong information is a vote for the wrong answer.
7. Fix the sources, starting with the ones the engines actually cited. If your Google Business Profile is unverified or unclaimed, start there — verifying your Google Business Profile is the single highest-leverage thing on this list.
8. Redo the whole audit in ninety days. This is not one-and-done. The web moves, the models re-crawl, and your competitors are not standing still.

If the answers come back thin or wrong, that is not a crisis. It is a to-do list.

Ten sources, one job

Across this series we walked through the ten sources AI leans on hardest when it decides which plumber to recommend: Angi, the BBB, Today's Homeowner, Yelp, Thumbtack, HomeAdvisor, Yellow Pages, Expertise.com, HomeGuide — and Porch, the one that turned out to be pointing at a business that had moved on.

Seven of those ten you can claim for free in an afternoon. Two of them, Today's Homeowner and Expertise.com, you cannot sign up for at all — you earn your way in through your public record. And the tenth is a lesson rather than a task.

The pattern underneath all ten is the same and it is duller than people want it to be. AI does not have opinions about plumbers. It reads what the web says about plumbers and repeats the consensus. So the job was never to game the models. It was to make the consensus about you accurate, consistent and abundant: correct name, address and phone everywhere it appears, a real review pace on the profiles that get cited, a website that answers the questions homeowners actually type, and enough independent mentions that a model has something to say when your name comes up. If you started at the hub post for this series, go back and check off the ones you skipped.

The shelf life on this is the part I would take seriously. Right now most plumbing companies have done nothing about any of it, which means the bar is low and the wins are cheap. That will not last. The same thing happened with Google reviews in 2015 and with Local Services Ads in 2020 — a two-year window where showing up early was worth a fortune, then a crowded field where it was just the cost of playing. We are somewhere in year two of this one.

None of it replaces the fundamentals, by the way. The reason AI can find you at all is usually that your plumber SEO foundation is solid and your listings are clean, and the reason your phone rings tomorrow morning is more often that your Local Services Ads are running properly. AI visibility is a compounding layer on top of that work, not a substitute for it. Anyone selling it as a substitute is selling you something. While you are auditing, it is also worth reading the five dumbest mistakes plumbers make on Google, because at least two of them are probably costing you calls right now.

Run the audit yourself first. Seriously — twenty minutes, a fresh browser tab, and a notepad. If what comes back is fine, you have saved yourself a phone call and you can get back to work. If it comes back wrong, or if it comes back with three competitors and no mention of you, book a call and we will go through what we find on your listings, your site and your ad accounts, and tell you which of it is worth fixing first and which of it does not matter. We start by telling you the truth, which occasionally means telling you your marketing is fine and the problem is somewhere else.