Service
GEO (Generative Engine Optimization)
Improve the public evidence AI systems use to understand and describe your brand.
The short answer
GEO (Generative Engine Optimization) is our name for the off-site half of the same work. No one outside the model providers can control what a generative system says, so we do not claim to. What we can do is improve the public evidence those systems read: consistent company and product facts everywhere your brand appears, accurate listings and profiles, earned coverage and authoritative mentions, corrected outdated claims at the source, and original research worth referencing. Then we observe how your brand is described across a fixed prompt set and report the wording we actually see.This page covers how geo (generative engine optimization) is run: deliverables, cadence, and process. Package inclusions and prices live on the pricing page, and category-specific detail lives on the industry pages.
Deliverables
What you actually get.
- Brand representation baseline: what each assistant says about you today, captured verbatim with dates
- Factual consistency pass across company name, category, product names, pricing, and positioning
- Public source audit: review sites, directories, profiles, roundups, and communities that describe you
- Correction workflow for outdated or wrong third-party claims, filed at the source
- Earned mention program: original data, expert commentary, and outreach that gives publishers a real reason to cite you. We do not buy undisclosed placements or paid links presented as editorial
- Where a paid or sponsored placement is genuinely worth it, we say so, price it separately, and keep the disclosure intact
- Observed share of mentions across a stable prompt set, reported with raw outputs
Outcomes
What changes in the business.
- The public record about your company is accurate and current
- More credible third-party sources describe you correctly
- You can see how assistants describe you over time instead of guessing
Process
How we run it.
01
Listen
Capture verbatim assistant output about your brand and category, with dates and prompts recorded.
02
Correct
Fix factual inconsistencies and chase the sources feeding wrong claims.
03
Earn
Publish original evidence and pursue disclosed, earned coverage on the sources buyers and models both read.
04
Observe
Re-run the prompt set monthly and report observed mentions, wording, and accuracy.
Where this sits
GEO (Generative Engine Optimization) is one stage of a connected system.
Run on its own it can help. Run inside the system, it compounds with the other seven stages. Step through to see how they connect.
Stage 03
Search and AI discovery
SEO, AEO, and GEO run together so buyers and answer engines both find and describe you correctly.
Measurement
The engines we track, and how often.
GEO is only real if the representation is measured. Every tier runs a fixed prompt set against named engines on a schedule, with the generated sentence about your brand captured verbatim.
ChatGPT
Chat answers and ChatGPT search results, with and without browsing
Google AI Overviews
AI Overviews and AI Mode on the queries your ICP runs
Perplexity
Answers and the cited source list, including follow-up prompts
Gemini
Gemini answers and Google's generative surfaces
Claude
Claude answers with and without web access
| Tier | Prompt set | What is covered |
|---|---|---|
| Ignite | 5 observed prompts | Category and brand prompts in one market and language, checked monthly. |
| Momentum | 20 observed prompts | Category, comparison, and problem prompts across assistants, checked monthly with observed mention rates. |
| Category Leadership | 50+ observed prompts | Full buyer-stage prompt set, competitive mention comparison, and a correction workflow for wrong or outdated claims. |
| Custom | Prompt set per statement of work | Multi-product, multi-brand, or multi-market prompt sets sized to the portfolio. |
- Weekly written update in a shared Slack channel: what shipped, what moved, what is next.
- Monthly performance report covering observed citations, observed mentions, ranking movement, and pipeline, including months with no movement.
- Quarterly strategy review that re-scores the prompt set and retires prompts that stopped mattering.
No tasks, no seats, no overage fees or surprise usage metering. The prompt counts above are scope, not a meter. See published pricing.
Sample scorecard
What lands in your inbox every month.
This is the reporting artifact, metric by metric. Nothing here requires a login to a dashboard we own.
| Metric | Definition we use | How it is reported |
|---|---|---|
| Observed citation rate | Share of the observed prompt set where a page on your domain appeared as a cited source, by assistant, on the dates we checked. | Percentage plus the prompt-level list, with the raw answers archived |
| Observed mention rate | Share of the observed prompt set where your brand was named at all, cited or not, against a fixed competitor set. | Percentage against up to five named competitors |
| Prompt coverage | Share of your observed prompt set that has a clear, quotable answer published on your own domain. | Covered, partially covered, and uncovered counts |
| Description accuracy | Whether the sentence an assistant generated about your category position, product, and pricing was factually correct. | Verbatim quotes with each inaccuracy and its likely source |
| Engine split | The same observations broken out per assistant, because changes rarely arrive on all of them at once. | Per-engine table with month-over-month change |
| Pipeline endpoint | Sessions, conversions, and qualified conversations attributable to AI search referrers and to the pages we build. | Reconciled to your analytics and CRM, with the attribution limits stated |
Metric definitions matter more than the chart. We publish ours in the AI visibility metrics guide so you can hold any vendor, including us, to the same wording.
Stack
The tooling we operate on your behalf.
Tooling is necessary and insufficient. Here is what we run, so you know the work is instrumented rather than manual.
Search and ranking data
Google Search Console, Semrush, and server log analysis for crawl and index behavior.
AI answer monitoring
A fixed prompt set run against each assistant on a schedule, with verbatim answers and cited sources archived so month-over-month change is auditable. We report what we observed, and we do not claim the outputs are controllable.
Models in production
GPT, Claude, and Gemini class models used for research, drafting support, and structured data generation, always with a human editor before publish.
Publishing targets
Webflow, WordPress, Contentful, Sanity, HubSpot, and modern React codebases, or we build and own the site ourselves.
Measurement
GA4, one attribution tool, and your CRM as the reporting endpoint rather than a dashboard we control.
Comparing us to a self-serve visibility platform? Read platform vs agency and the honest AEO tools landscape.
How this differs by industry
GEO (Generative Engine Optimization) is not run the same way in every category.
The queries buyers run, the trust signals they need, and the review process we work inside change by category. Each industry page sets out what changes and what stays the same.
B2B SaaS
Horizontal SaaS built for teams, sold through pipeline.
Vertical SaaS
Industry-specific SaaS: legal, healthcare, construction, hospitality, and more.
Developer Tools
Developer-first products with PLG motion, docs, and community.
Fintech
Regulated financial software: payments, lending, wealth, and infrastructure.
AI & ML Products
AI-native products, model APIs, and ML platforms.
Consumer Software
Secondary capability. Consumer apps where search, content, and paid drive growth.