ChatGPT now stores conversation memory and uses it to personalize future answers. When a homeowner asks for a contractor recommendation after previously discussing home repairs in ChatGPT, the memory influences which businesses get named. Contractors with stronger entity signals across the web, more consistent business data, and more recent mentions in directories are the ones that show up in memory-shaped recommendations. This is not a settings page you can flip. It is a reason to treat every piece of online information about a business as something ChatGPT might remember next time.
TL;DR
OpenAI rolled out a feature called "memory" that lets ChatGPT store details from past conversations and use them in future ones. In late 2026, OpenAI expanded this with something called "dreaming," a process where ChatGPT reviews stored memories during downtime and builds richer associations between facts. For trades businesses, this means the old approach of optimizing for one query ("best plumber in [city]") captures only part of the picture. The other part is what ChatGPT associates a business with across every conversation it has ever had about that trade in that area.
What ChatGPT Memory Actually Does
OpenAI rolled out a feature called "memory" that lets ChatGPT store details from past conversations and use them in future ones. In late 2026, OpenAI expanded this with something called "dreaming," a process where ChatGPT reviews stored memories during downtime and builds richer associations between facts. The company described it as making recommendations more contextual and personal.
Here is how that affects a trades business. A homeowner asks ChatGPT in March about a banging noise in their pipes. ChatGPT explains air hammer, mentions a few plumbers in the area. Three months later, the same homeowner asks who to call for a water heater replacement. ChatGPT remembers the prior conversation. It factors in the location, the type of home issue discussed before, and the plumbers it previously mentioned. The recommendation may shift based on that accumulated context rather than starting from a clean search.
OpenAI's own announcement page showed examples of restaurant recommendations shaped by memory. The same mechanic applies to plumbers, electricians, and HVAC companies. Memory means the conversation history becomes part of the recommendation engine.
Why Memory Makes Local Recommendations Less Predictable
Before memory, a contractor could reason about AI visibility this way: a homeowner asks ChatGPT "best plumber near me," ChatGPT does a web search, pulls in Google Business Profile data and directory listings, and names three plumbers. The same query from a different homeowner in the same town got similar results.
Memory changes that. Two homeowners in the same ZIP code can get different plumber recommendations because one of them mentioned a boiler issue in a previous ChatGPT conversation while the other asked about a clogged drain. The accumulated memory steers ChatGPT toward businesses it associates with those specific services.
For a contractor, this means the old approach of optimizing for one query captures only part of the picture. The other part is what ChatGPT associates a business with across every conversation it has ever had about that trade in that area. Those associations come from a business name appearing next to specific service descriptions in directories, on its website, in news articles, in reviews, and on social media.
What Shapes ChatGPT Memory for Contractor Recommendations
Several factors determine which contractors show up in memory-influenced recommendations.
Service specificity in online listings. A plumbing company whose Google Business Profile lists "water heater installation," "boiler repair," and "sump pump service" as separate categories is more likely to be remembered in connection with those specific services. A company that just says "plumbing" gets a thinner association.
Consistency of business data. If a company's name, address, and phone number match across Google Business Profile, Yelp, Angi, Thumbtack, its website, and the Better Business Bureau, ChatGPT builds a stronger memory of that business as a single entity. Inconsistent data fragments that memory.
Recency of mentions. ChatGPT's memory weighting favors recent information. A contractor mentioned in a local news article last month has a stronger memory signal than one mentioned in a 2023 blog post. This is why publishing fresh content, getting fresh reviews, and maintaining active social media profiles matters more than it did for traditional SEO.
Review content, not just review count. ChatGPT reads the text of reviews. A plumber with 50 reviews that say "fixed my water heater" builds a memory association with water heater repair. A plumber with 50 reviews that just say "great service" builds a vaguer association. The specificity of what customers write matters.
How to Check What ChatGPT Remembers About Your Business
There is no dashboard that shows what ChatGPT has stored about a company. But there is a practical way to test it.
Open a fresh ChatGPT conversation and ask: "What do you know about [business name] in [city]?" The answer reveals what ChatGPT has associated with that business name from its training data and browsing. If the answer is vague or wrong, entity signals are weak.
Then ask: "What plumbers in [city] do you know about?" and see if the company appears. If it does not, the business is not strongly enough associated with its trade and location in ChatGPT's memory.
Run this test once a month. Track whether a business shows up, what services ChatGPT associates it with, and whether the information is accurate. Changes in these answers signal changes in AI visibility.
What Trades Businesses Can Do About Memory-Based Recommendations
The actions that improve standing in memory-influenced recommendations are the same ones that improve AI visibility overall, but with extra emphasis on specificity and recency.
- List specific services, not just the trade. On Google Business Profile, Angi, Yelp, and the company website, break out individual services as separate entries. "Tankless water heater installation" is better than "plumbing." ChatGPT's memory builds associations at the service level, not just the trade level.
- Keep business data identical everywhere. Same name, same address, same phone number, same hours. Inconsistencies make ChatGPT less confident that all these listings refer to the same business, which weakens the memory it stores.
- Get fresh reviews that mention specific services. Ask recent customers to mention what work was done. "Installed a new water heater and cleaned up afterward" builds a stronger memory signal than "would recommend."
- Update the company website regularly. A site that has not changed in two years signals a business that may be less active. A site with a recent blog post, updated service pages, or fresh photos signals a business that is operating and available.
- Get listed in the directories ChatGPT reads. Angi, Thumbtack, Yelp, the Better Business Bureau, and local chamber of commerce directories all feed ChatGPT's understanding of which contractors exist in a given area. If a business is not in these directories, ChatGPT has fewer sources to build a memory from.
The Catch: Memory Varies by User
ChatGPT memory is per-user, not global. Two different homeowners get different memories based on their own conversation histories. A contractor cannot optimize for every individual user's memory.
What a business can do is strengthen its entity across the web so that regardless of which memories ChatGPT pulls from, the business has the strongest association with its trade and services in its area. The businesses that show up most consistently across memory-influenced recommendations are the ones with the most coherent, specific, and recent online presence.
Think of it as building a business's reputation in a system that never forgets a conversation. Every directory listing, every review, every website update, every social media post adds to what ChatGPT might remember next time someone asks for a contractor in a given area.
Run a Free AI Visibility Audit
FirstCall Visibility runs a 12-point audit that checks exactly what ChatGPT, Google AI, Perplexity, and Gemini know about a trades business. The audit covers directory presence, business data consistency, service specificity, review content, and whether a business shows up in AI answers for its trade in its area. Run the free audit at firstcallvisibility.com.
FAQ
Does ChatGPT remember every conversation about my business?
No. ChatGPT stores memory per user account, not globally. It remembers details from that user's past conversations and uses them to personalize future answers. A business is not being tracked in a central database. But the information ChatGPT has access to across the web, directory listings, reviews, website, and mentions, shapes what it pulls into those memories.
Can I control what ChatGPT remembers about my business?
You cannot directly control ChatGPT's memory. What you can control is the information available about a business across the web. Strong, consistent, specific business data in directories, on the website, and in reviews gives ChatGPT better material to build accurate memories from.
Do all AI assistants use memory for local recommendations?
ChatGPT, Google Gemini, and Perplexity all use some form of conversation context to shape recommendations. ChatGPT's memory feature is the most developed for persistent, cross-session personalization. Siri and Apple Intelligence use on-device context but handle local business search differently, relying more on Apple Maps data.
How is ChatGPT memory different from regular SEO?
Traditional SEO optimizes for a single query producing the same result for every searcher. ChatGPT memory means the same query can produce different results for different people based on their conversation history. This makes AI visibility less predictable and more dependent on the overall strength and specificity of a business's online presence, not just a single keyword ranking.
Should I be worried that ChatGPT memory hurts my visibility?
Memory can work in a business's favor if it has strong, specific entity signals. A homeowner who previously asked about a service a company specializes in is more likely to get that business recommended later. The risk is for businesses with weak or inconsistent online presence, where ChatGPT has nothing specific to remember.