This playbook walks through the complete Gumshoe workflow used by top-performing customers: set up a tuned Visibility Audit (formerly known as a report), let it run on a regular cadence, publish first-party content, fix technical issues, build a third-party strategy, and repeat the cycle to keep improving how AI models see your brand.
The steps below work the same whether you're running this for your own brand or managing it for a client. If you're an agency, treat each client as its own project with its own product category, personas, and topics, so their data never mixes with another client's.
Step 1: Scope your Visibility Audit to one product category
Each project should be scoped to one product category. Combining multiple product categories in a single project muddies the data and reduces actionability. If a brand sells more than one product line, serves multiple regions, or supports multiple languages, create a separate project for each.
Examples of good projects, each scoped to a single product category:
- Best sustainable children's clothing
- Best second-hand kids' clothing
- Best American-made children's clothing
- Best sustainable children's clothing in the UK
- Best sustainable children's clothing stores in East LA
- Best sustainable children's clothing stores in East LA, Spanish language
To set up a project, work through the project-creation wizard:
- Set your product category. Pick one product category, or enter your own if none of the AI-suggested ones fit. Add a geographic focus or language if the audience is shaped by either.
- Review your personas. Gumshoe automatically generates personas based on your brand and product category, up to your plan's limit (currently up to 8 for the monthly Visibility Audit). Each persona simulates a real buyer asking AI models for advice. You can edit, delete, or add personas within that limit, and even add your own defined ICPs alongside the automatically generated ones.
- Review your topics. Topics are the themes and questions that map to how your audience actually shops or evaluates options. Adjust them so they reflect real buying questions.
- Review your prompts. Prompts are generated automatically by combining topics with personas, up to 10 prompts per persona, giving models enough surface area to reveal real patterns. You can regenerate prompts after tuning your personas and topics.
Step 2: Let your subscription set the schedule
AI search shifts every week, and your competitors are moving too. Your subscription sets the cadence automatically, so you don't have to remember to rerun anything: the full Visibility Audit runs monthly, and the recurring Snapshot Audit runs weekly on Starter or daily on Pro to catch movement in between full runs. Even when you're not actively optimizing, those regular runs matter, because the AI models and the brands around you are changing constantly.
If your goals change, for example moving toward more active, fast-moving optimization, upgrading your plan is what changes your cadence, not a setting you toggle project by project.
Step 3: Understand your model coverage
Relying on a single model gives you an incomplete picture. Each model has its own training data, sources, biases, and update schedule, so visibility can vary dramatically from one to the next. Your plan selects a set of models for you (currently up to 6), chosen from providers like OpenAI, Anthropic, Google, Perplexity, xAI, and DeepSeek, so you get a durable, multi-model read without having to hand-pick each one.
A few reasons multi-model coverage matters:
- Different models cite different sources. What ChatGPT pulls from is rarely identical to what Gemini, Claude, or Perplexity rely on.
- Different models have different blind spots. Multi-model coverage surfaces differences you'd never see by checking only one.
- Foundational and live-search models behave differently. Foundational models lean on training data that updates in batches. Live-search models like Perplexity crawl in real time. Most major models now use a mix of both, depending on the query and the user's settings.
- Updates roll out independently. One model may improve citations the same week another drops them. Multiple models protect you from misreading a single-model fluctuation as a trend.
- Real buyers mix tools. Your audience moves between standalone apps, browser tools, and built-in AI search. Your strategy needs to cover the full ecosystem.
How do I take action with the three levers?
Once your Visibility Audit is running on its regular schedule, it's time to improve the score. Customers who see the biggest gains pull on three levers in parallel: first-party content, technical health, and third-party strategy.
Lever 1: Generate and publish first-party content
Publishing fresh, structured, targeted content on your own site is the strongest lever you control. Use Content Generation in Gumshoe to draft pieces aimed at your weakest persona and topic combinations.
Available formats:
- FAQs
- Knowledge articles
- How-to guides
Gumshoe content is modeled on the formats LLMs already cite, which makes it easier for models to read, understand, and reuse. Aim to publish weekly or monthly to stay inside the LLM freshness window.
Lever 2: Improve technical health with your Technical Audit
Strong content doesn't help if AI models can't read your pages. Your Technical Audit identifies the technical issues that block models from parsing, understanding, and citing your site.
Technical Audit covers issues across:
- Metadata
- Heading structure
- Markup quality
- Page performance
- Structured data
- Internal linking
- Crawlability
- Retrievability
Fixing these makes your site fully accessible to the models that drive visibility.
Lever 3: Build a third-party content strategy
AI models pull from a wide mix of fresh, authoritative sources. Your own site is only part of the picture. Use your Citations (or your Citation Audit on Pro, which adds deeper page-fetch enrichment) to find the third-party domains shaping your category, then build outreach around them.
Use Citations to identify:
- Websites the models already cite for your topics
- Publications that influence your category
- Competitor coverage you're missing
- Domains that repeatedly shape model answers
- Times models cited your own content
Build your strategy around:
- Targeting cited publications for placements or commentary
- Contributing expert content or quotes
- Securing listings on authoritative directories
- Collaborating with writers, journalists, and creators
- Strengthening your presence on high-trust sites
Third-party coverage helps AI models confirm and reinforce what they learn from your own site.
How do I run the loop again?
The Gumshoe workflow is continuous: monitor your Visibility Audit, find new opportunities, make improvements, and let the next scheduled run measure impact. AI search is a fast-moving space, and your competitors are working on it too, so consistent measurement and iteration is what separates brands that gain ground from those that fall behind.
Below is a recommended cadence based on patterns from thousands of audit runs, content batches, and Technical Audits across customer brands. Small, consistent improvements add up.
Weekly
Focus on freshness, technical clarity, and targeted content, where you can move fastest.
- Review visibility trends across models, competitors, personas, and topics
- Publish at least one new batch of first-party content
- Rewrite and republish a batch of content older than 90 days (LLMs heavily favor fresh content)
- Fix high-priority technical issues identified by your Technical Audit
Goal of the week: chip away at the to-do list and keep momentum.
Monthly
Step back to measure changes, plan strategic focus, and expand your footprint across first-party and third-party sources.
- Review month-over-month visibility changes across your Visibility Audit
- Set third-party outreach or PR targets for the coming month based on Citations insights
- Review how often models cited your site or published content, and set content goals for the next month
- Set new technical improvement goals for the coming month
Goal of the month: confirm what's working, course-correct what isn't, and strengthen authority signals on-site and off-site.
Quarterly
Take on bigger structural shifts and longer-term strategic adjustments.
- Update personas, topics, and prompts based on new visibility trends
- Decide whether new projects are needed (new product lines, new regions, new competitor clusters)
- Review your overall content strategy
- Re-review your overall Technical Audit health and strategy
- Review your overall third-party strategy
- Compare multi-model performance to identify where gaps are widening or narrowing
Goal of the quarter: adapt your strategy to shifts in your business, your competitors, and the AI models themselves.
How do I stay in touch?
The Gumshoe team is continually improving the platform and supporting customers and agencies through every stage of the workflow. If you would like a product demo, you can book directly here: https://cal.com/jim-watson-kdgftu/30min. For support, use the chat widget in the bottom-right of your account or email support@gumshoe.ai.