Skip to main content

Product tour

See how GEO Style works, in four steps.

Follow an example CRM brand through the app, on example data, from a buying question it loses to proof that its work changed the answers.

Request accessStart at step 1

Sign-up opens soon. Request access and we will set you up.

  1. Step 1 TrackWhich brand the answers pick for each buying question.
  2. Step 2 ExplainThe answers, and the sites behind each pick.
  3. Step 3 ActOne ranked list of fixes, drafted for you to approve.
  4. Step 4 ProveEach result measured against questions nobody acted on.

Step 1 · Track

Track the questions buyers ask. Each one becomes a decision.

Add your website and GEO Style suggests competitors and buying questions. Each question is asked every day on the assistants you choose, from the markets you sell in, and becomes a decision: won, competitive, weak or lost.

  1. 1

    Your assistants and market

    Each question is asked on its schedule, daily by default, on the assistants you choose, from the market you sell in.

  2. 2

    A status, not a mention count

    Won, competitive, weak or lost: the most common result of the last 7 runs, so one odd answer does not flip it.

  3. 3

    Who the answers pick instead

    The brand the answers recommend, and how many of the 21 answers name you at all.

In the app

Example data: The Decisions list: 5 buying questions, asked on 3 assistants in the United States over the last 7 runs, each with its status, impact, the brand the answers pick and how many of the 21 answers name you. The first, Which CRM is best for a 50-person sales team, is lost: the answers pick Northwind and name you in 9 of 21.

Step 2 · Explain

See why the answers pick someone else. Read the answers and their sources.

Open a decision to read what each assistant answered and the sites it cites. See where a competitor is described and you are not, and what the answers say for and against you.

  1. 1

    What each assistant answered

    Over the last 7 runs: the brand each one picked, and how often.

  2. 2

    The sources behind the pick

    The sites the answers cite, in how many of the 21 answers, and which are rising or new. Where an assistant reports them, the pages it read but did not cite.

  3. 3

    What the answers say

    The arguments for and against you, how many answers make each, and the site it comes from.

In the app

Example data: A decision opened: Which CRM is best for a 50-person sales team, lost. Over the last 7 runs ChatGPT and Perplexity pick Northwind and Gemini names your brand third. The answers cite g2.com (12 of 21), techradar.com (8 of 21), northwind.com (7 of 21), reddit.com/r/sales (4 of 21); one answer read yourbrand.com/features but did not cite it. For you: easy to set up. Against you: no built-in forecasting.

Step 3 · Act

Fix what matters first. One ranked list of actions.

One list ranks every action by priority: mentions to earn on the sites the answers cite, pages to refresh, threads to answer and site issues that block assistants. Playbooks draft the content for a person to approve.

  1. 1

    One list, ranked by priority

    Mentions to earn, pages to refresh, threads to answer and site fixes, across all your decisions.

  2. 2

    What each action is about

    Its kind, and the site, page or community it is about. Open it to see the decision it serves.

  3. 3

    Drafted in steps, approved by a person

    A playbook writes the draft from your Brand Kit and approved facts, then waits for review. Approved content goes to WordPress, Webflow or one of 9 other CMSs, such as Contentful, Shopify and Framer.

In the app

Example data: The Actions list: 4 open actions ranked by priority, each with its kind and what it is about. The top one, at 86: Get listed on G2’s “Best CRM for small sales teams”, about g2.com. The refresh of yourbrand.com/compare is open beside the list: it serves the decision Your brand vs Northwind for B2B sales, and the Page refresh playbook has written a draft that waits for approval before it goes to WordPress as a draft.

Step 4 · Prove

Prove what your work changed. Against questions nobody acted on.

Each finished action is compared with up to 8 similar questions nobody acted on, over the same weeks, and shown with a likely range. Wrong claims are followed until each assistant gets them right.

  1. 1

    Against questions nobody acted on

    Your question next to 8 similar questions you did not act on, from 28 days before the action to 60 days after, so a shift across the market is not credited to you.

  2. 2

    A result with its likely range

    +14 pts, likely +6 to +22. It counts as Helped only when the whole range is above zero.

  3. 3

    A clock on wrong claims

    A claim that contradicts a fact you approve is followed until each assistant gets it right, with the days it took.

In the app

Example data: The Impact page for a finished action, “Added a forecasting section to the features page”, on the decision Best CRM with built-in sales forecasting. Helped: +14 points in the share of answers naming you, likely +6 to +22, compared with 8 similar questions you did not act on, from 28 days before to 60 days after it was done. By kind of action: Content, 2 of 4 measured helped; Earn a mention, 1 of 2 measured helped.
Example data: The correction clock on the Reputation risk page: the wrong claim “Your brand has no built-in forecasting”, checked against the approved fact that forecasting is included in every plan. Perplexity was right after 9 days, Gemini after 12 days; ChatGPT is still wrong on day 14.

Get started

See it on your own brand. The first answers arrive shortly after setup.

GEO Style reads your site, suggests competitors and drafts the questions to track.

Request accessSee plans and prices

Sign-up opens soon. Request access and we will set you up.