Fix Wrong Answers and Track Your Progress

When the AI says something false about you

You found these in your baseline: the wrong hours, the address you moved out of, a service you dropped, a price that's years old. Your first instinct is to look for a way to report it. There isn't one, not really. AI engines are reading something, somewhere, and repeating it. So the fix is detective work: find what they're reading, correct that, and make sure it gets re-read.

The correction workflow

• •Capture the wrong claim exactly: which engine, the precise wording, and the question that produced it.
• •Find the source. If the engine shows citations, start there. Otherwise, check the usual culprits: an outdated page on your own site, your Google Business Profile, a directory listing, a review site, or an old news article.
• •Correct it at the source. Update your own page, edit your Business Profile, or email the directory asking for a change.
• •Force a re-read. Request re-indexing in Google Search Console and Bing Webmaster Tools for any page you changed.
• •Leave feedback in the product too. The thumbs-down in ChatGPT and the feedback control on AI Overviews are worth thirty seconds, though they're a supplement to fixing the source, not a substitute.
• •Re-test. Ask the same question again in a fresh chat a week or two later, and keep asking until the answer flips.

Be patient across engines. A fix might show up in Perplexity within a couple of weeks while Copilot keeps repeating the old claim until Bing re-crawls the source. That's normal. Log the date you made each correction so you can tell the difference between “it didn't work” and “it hasn't propagated yet.”

Seeing AI traffic in your analytics

By default, most visitors arriving from AI tools get filed as Direct or Referral, so it looks like nothing is happening. In Google Analytics you can fix that: go to Admin, then Data Display, then Channel Groups, copy the default group, and add a new channel called “AI Search” placed above Referral. Set it to match the session source against this pattern:

• •chatgpt\.com|openai\.com|perplexity\.ai|claude\.ai|gemini\.google\.com|copilot\.microsoft\.com|meta\.ai
• •Expect the numbers to look small. Around 1% of sessions is a normal share right now, and those visitors typically convert much better than average, so judge this channel by leads and calls rather than by traffic volume.
• •If you use a CRM or a contact form, add a hidden field that records where the visitor came from. Knowing that three of last month's five best leads came from ChatGPT is worth more than any visibility score.

Your monthly rhythm: about 30 minutes

• •Re-run your baseline questions, word for word, in the same engines (15 min).
• •Update your share of voice and note any new wrong facts (5 min).
• •Fix one thing: a section rewritten answer-first, an FAQ question added, or a correction filed (10 min).

How to know if it's working, and when to stop

Give this two quarters before you judge it. If your share of voice is climbing and AI referrals are converting above your normal rate, keep going and consider investing more in content and third-party presence. If after two quarters the number hasn't moved and AI referrals are still negligible, that's real information too: put your effort back into the channels that are working for you. Being willing to walk away from something that isn't paying off is part of running this well, and it's the same discipline you'd apply to any other marketing channel.

You've built an engine

Look at what you have now: a measured baseline, a site AI crawlers can reach, pages written to be quoted, an FAQ built from real questions, a way to correct what's wrong, and a monthly habit to keep it all current. That's the whole system. The businesses that win here aren't the ones with the biggest budgets, they're the ones that keep showing up month after month with clear, honest, specific answers. That's a game a small business can win.