How long does it take AI systems to recognise a business rebrand?
A live longitudinal study on a real rebrand, with the Day 0 baseline captured before any external profile was touched.
Status
Collection in progress. Partial data only. Day 0 baseline captured 2026-09-03. Later intervals are not yet due. No conclusions exist yet and none are offered.
The question
A business changes its name. Its website is updated correctly and consistently. Its old domain redirects to the new one. How long before search and AI systems understand that the two names are the same business, and what actually causes them to accept it?
We could not find this measured anywhere with a published method. There is plenty of advice about rebranding for search, most of it reasonable and none of it evidenced with timings. So we are measuring it on ourselves.
Why this case is unusually clean
The subject is our own business, which trades as A Social Solutions and formerly as A Social Strategy. That means we control the site, we know exactly what changed and when, and we do not need permission to publish the results including the unflattering ones.
More usefully, the variables separate. The website side of the rebrand is complete and verified: the name, the former name as an alternate name, the founder, the redirect from the old domain, and consistent facts across 105 pages. The external side has not started. No Google Business Profile exists, and the LinkedIn company page still carries the old name.
So the interval that begins now isolates one question: what does a correct, consistent website achieve on its own, before any external profile corroborates it? That is a question most rebrands cannot answer, because everything moves at once.
What was measured on Day 0
Seven fixed questions, run unpersonalized in a US locale, recorded verbatim. They cover both names, the relationship between them, who runs each, and which company each domain belongs to.
The Day 0 result is stark and it is the reason this study is worth running. Asked whether the two names were the same company, the answer was no. Not an absence of information: an explicit negative. Asked who runs the new name, the answer named four executives of three entirely unrelated companies that share a similar name, and did not mention the actual owner at all. Asked the same question about the old name, the answer was correct, well sourced, and specific.
Every fact about the old entity is externally corroborated. Every fact about the new entity that depends on continuity with the old one is currently absent or denied. The website has asserted otherwise, correctly and consistently, since before this baseline was taken.
Schedule and what happens next
Day 0, then day 7, 14, 30, 60 and 90, same questions, same conditions.
The external profile migration will happen during the study and will be recorded with its date, because it is the intervention. Measurements before and after it are reported as separate phases rather than as one series, since they are answering different questions.
If an interval is missed it will be recorded as missed. Nothing will be reconstructed after the fact.
What this study cannot tell you
Published alongside the method, not buried after the results, because a study that hides its limits is advertising.
Sample size is one business. A single rebrand cannot establish a typical timeline, and no typical timeline will be claimed from it.
The Day 0 baseline captured web search results as a proxy for the retrieval layer. Direct assistant answers must be captured by a human and were not available at Day 0 for all platforms; any missing platform is recorded as not captured rather than inferred.
The subject is the publisher. We have an obvious interest in the outcome, which is exactly why the method and the raw Day 0 answers are published before the results.
The external migration is an uncontrolled intervention with its own timing, and it will confound the clean before-and-after unless it is clearly separated. It will be.
The name is unusually collision-prone. Several unrelated businesses have similar names, which may make this case harder than a typical rebrand rather than representative of one.