"Having a chatbot" isn't "showing up in chatbots"
It's the most common confusion and the most expensive. When a client asks to "position my brand in ChatGPT", half the agencies answer "great, we'll build you a chatbot of your own". That's building a chatbot — a different product, with its own utility, but one that positions absolutely nothing. Positioning your brand in ChatGPT means working on how the model decides which brand to recommend when a user asks about your category: and that decision is made outside your site, based on what the model finds about you on the internet.
How ChatGPT decides between two competing brands
When a user asks "what's the best [X]", ChatGPT weighs three factors in order of impact:
- Density and consistency of mentions. How many sources mention your brand tied to that category, how often and with what description.
- Entity specificity. If your brand is clearly associated with a concrete use case, the model brings you up when that use case shows up in the prompt. Generic brands lose to specific ones.
- Recency and authority of sources. A mention in a relevant sector outlet three months ago weighs more than ten mentions in anonymous blogs from three years ago.
The three levers: entity, citations, schema
Everything you can do to position yourself falls into three levers. Anyone selling you something outside these three is selling extras.
Lever 1 · Clear, consistent entity
Your brand must be defined identically across every public place: site, LinkedIn, Wikidata/Wikipedia, sector directories, Google Business profile. Same name, same category, same value proposition. Consistency is what the model reads as "this is a solid, recognizable entity".
Lever 2 · External citations in sources that carry weight
Getting sector outlets, podcasts, partners and customers to mention you is the hardest lever to move and the one that pays back the most. One mention per month in an authoritative channel weighs more than ten mentions in aggregators with no traction.
Lever 3 · Schema on your site
Structured data (Organization, Service, FAQPage) is how you talk to the model in its own language. It's the cheapest lever, the fastest to ship, and the one that closes the loop: the entity exists, the citations back it up, the schema confirms it.
Realistic timeline (4 weeks, 3 months, 6 months)
| Timeframe | What happens | What to measure |
|---|---|---|
| Week 1-4 | Schema shipped + entity unified across channels | First sporadic mentions (5-10% of prompts) |
| Month 2-3 | First external citations + content reformatted for citability | Mention rate climbs to 15-25% |
| Month 4-6 | Sector authority consolidated + content iteration | 25-40% mention, top-3 position in category |
| Month 6+ | Maintenance + expansion into adjacent niches | Stable position, defense against competition |
How to measure your position vs. competitors
Your position is only readable in comparison. The three metrics that matter in GEO competitive monitoring:
- Share of voice (SoV) — what percentage of total mentions in your category are yours.
- Comparative coverage — in how many prompts you appear versus each direct competitor.
- Relative sentiment — what adjectives the model uses for you vs. for the others.