Westhive
Eight locations, three languages, week one. There is nothing to report yet and that is the point: this is what the start of a multi-location, multilingual AEO programme actually looks like.
Client engaged 16 September 2026. Figures describe the scope of the programme, not its results. Read on for why there are no result numbers on this page yet.
Who they are
Westhive runs eight serviced-office buildings across three Swiss cities, selling to established companies taking real space rather than to freelancers buying a desk. Their buyers search in German, French and English, and an answer that names a competitor in the wrong language is still a lost deal.
Where they started
Multi-location businesses lose AI answers in a specific way: the engine knows the brand but cannot tell which building answers a question about office space in one particular district, so it names an operator whose individual locations are legible instead. Multiply that by three languages and the surface area is eight times larger than a single-site account, with eight times as many ways for the facts to disagree with each other.
What has been built
Every item below is published and live on the client's site.
A baseline that will still be valid in ninety days
150 buyer questions across seven language and market pairs, measured and published before a single change was made. A question set that moves between runs cannot show progress, so this one is fixed now and left alone.
Seventeen competitors, tracked per market
Who gets named instead of Westhive, separately in each language, because the answer is not the same in German as in French.
Every location audited against the catalogue
All eight Google Business Profiles pulled and checked location by location, because in this category the profile is the entity as far as a local answer is concerned.
The language rules written down before the first draft
Swiss German spelling, the informal form of address in German, the formal one in French, and the local convention of putting the amount before the currency. We got the form of address wrong in the first examples we sent and it was corrected the same day.
What we are not claiming yet
There is one measurement run on this account and no second point to compare it to. Anything published as a result here today would be a baseline wearing a result's clothes. The page exists so it can be updated honestly later, and so there is a public record of what the starting position actually was.
We publish a case study when the work is real, not when the numbers flatter us, and we add the numbers when they survive our own comparability rule: two runs are only comparable if the question set barely moved between them. That rule costs us a headline on this page today. It is the same rule that stops us reporting a fake win to the client, which is worth more.
Scope figures read from the SourceRank AI delivery database, 18 September 2026. No customer of theirs is named anywhere on this page.
See the rest of the programme
All case studies
Two of our own properties with published numbers, and the client programmes still in flight.
See all ->See your own visibility gap
Get a real baseline across every answer engine before you decide what to do next.
Start a conversation ->