Gemini stopped calling them a staffing agency

The problem

Rotaly sells shift scheduling software to clinics. It helps a team organize the people it already employs. Yet Gemini described the company as a healthcare staffing agency, and ChatGPT often repeated the same interpretation.

That mistake changed the meaning of almost every follow-up question. An answer about how Rotaly worked could discuss finding temporary staff rather than planning shifts. Only 22% of the answers in our starting sample described the company correctly.

The first job was to fix the identity of the business. More category recommendations would have been of limited value while an engine was associating the name with a different service.

Gemini

· June · “What is Rotaly”

“Rotaly is a healthcare staffing agency that connects clinics with temporary nurses and medical assistants…”

Wrong category

Finding the source of the confusion

We ran 210 prompts daily, including direct questions about Rotaly and questions that added context about clinics, scheduling and software. Comparing those answers helped separate a broad misunderstanding from an isolated phrasing issue.

The audit traced the mix-up to an old press release and a staffing business with a similar name. Some surrounding profiles reinforced the ambiguity. The work therefore needed to cover the company’s own explanation and the external records a reader might use to identify it.

What we did

Days 1–14. We documented the wrong category descriptions and assembled a reference set of current company facts. The queue distinguished incorrect information about Rotaly from accurate information about the similarly named staffing company. Those were different problems and required different corrections.

Weeks 3–5. We rewrote the homepage and About page to explain plainly what Rotaly did and who used it. The opening description connected the company name with shift scheduling software for clinics. Supporting copy explained the workflow without relying on a reader to infer the category from a broad promise about healthcare teams.

We added organization schema to reinforce the same identity. The structured information followed the visible page copy; it was not treated as a substitute for explaining the business clearly to a person.

Weeks 5–8. We corrected profiles on 18 directories and had the old press release updated. Each request identified the category mistake and linked to a current public reference. Where the confusion involved the other business, we made the distinction explicit rather than asking a publisher to remove unrelated, accurate information.

The prompt checks continued throughout. We kept unresolved descriptions in the issue log even after the page or profile behind them had changed, because publishing a correction and observing a corrected answer are separate events.

What moved

Answers describing Rotaly correctly rose from 22% to 94% over eight weeks. Gemini corrected itself last in this engagement, roughly two weeks after the directory updates.

The result was a clearer account of the business across the tracked questions. It did not establish recommendation share, referral volume or new customer revenue. We kept those questions separate from the category-accuracy measure the work was designed to improve.

What we’d do differently

We would check for name collisions on day one. The similarly named business explained a substantial part of the confusion, but we did not identify it until the second week. Finding it earlier would have made the initial correction requests more precise.

If an engine repeatedly puts your business in the wrong category, begin with identity. Review the company description, old announcements and external profiles together. A consistent, specific explanation gives both buyers and publishers a better reference than another general awareness campaign.