Brands that stopped being invisible.

The problem

Prospective students had stopped starting on Google. They were asking assistants to shortlist consultants for them, and Sovo appeared in none of those answers — not because its reputation was weak, but because nothing on its site told an AI engine what it was, who it served, or which outcomes it had earned.

What we did

• Mapped the prompt space students actually use, not the keywords the industry ranks for
• Rebuilt entity data so engines could resolve Sovo as a named organisation with a defined service area
• Seeded targeted assets against the twelve prompts where competitors were being cited and Sovo was not
• Tracked recommendation share weekly across five engines

The problem

AB7 published genuinely authoritative clinical material, and engines cited everyone but AB7. The content was correct and the structure was invisible: no schema, no clear authorship, and no machine-readable link between the claims and the organisation making them.

What we did

• Structured the clinical library with schema that names the author, the organisation and the evidence
• Consolidated duplicated topic pages competing with each other for the same answer
• Aligned each page to the specific question it should be the answer to
• Monitored which domains engines cited instead, and closed the gap page by page

The problem

Franchise locations inherit brand recognition and lose local specificity. Asked for the best salon in Mohali, engines returned aggregators and directory listings. The salon existed in every map and review database, and in none of the structured signals an assistant reads before it recommends somewhere.

What we did

• Reconciled name, address and service data across the sources engines actually draw on
• Separated the location’s identity from the parent brand so both could be resolved independently
• Built service-level pages matching how people phrase requests — by treatment, not by category
• Tracked the local prompt set weekly and watched which competitors surfaced instead

The problem

Lavoya won work through referral and repeat business, which left almost no public surface for an engine to read. Asked to suggest caterers for a wedding or a corporate event in the region, assistants had nothing to go on but listing sites, so that is what they returned.

What we did

• Documented cuisine range, event types and capacity as structured, machine-readable facts
• Built answers for the questions that precede a booking — guest counts, dietary coverage, lead times
• Established consistent business identity across the sources engines cross-check
• Measured which occasions the brand surfaced for, and which it still did not

The problem

Anthéa serves brides, walk-in salon clients and prospective students, and engines flattened all three into a single vague description. Someone asking about bridal makeup and someone asking about a makeup course were being given the same unhelpful answer, and neither converted.

What we did

• Split the entity into three resolvable service identities under one brand
• Built a distinct prompt set for each audience rather than one blended list
• Structured course, service and portfolio data so engines could tell them apart
• Tracked all three audiences separately so gains in one never masked losses in another

The problem

Buyers evaluating an AI partner now ask an assistant to name credible firms first. Welzin’s technical depth lived in engineering work rather than in anything an engine could read, so it was absent from exactly the answers its own expertise qualified it for.

What we did

• Translated delivered capability into structured, verifiable public claims
• Mapped the prompts technical buyers use when scoping a build, not the ones agencies market against
• Aligned service pages to those prompts and removed overlapping duplicates
• Monitored competitive citation share across engines month over month