The GEO PDCA cycle: monitor and improve mention share (2026)

Why generative engine optimization (GEO) needs a PDCA cycle

GEO is not something you do once and then finish. AI answers change from day to day, and your brand's mentions keep shifting as competitors launch initiatives and the information landscape changes.

Running a PDCA (Plan-Do-Check-Act) cycle gives you the following benefits:

Quantify the impact of your initiatives: Identify which initiatives helped raise your mention rate
Optimize resource allocation: Focus resources on high-impact initiatives
Respond to competitor moves: Spot changes in competitors' mention rates early and respond
Adapt to AI algorithm changes: Understand the impact of AI model updates and adjust your strategy

Plan: Set KPIs and design initiatives

Key KPIs for GEO

First, clearly define the metrics you will use to measure GEO results.

Primary KPIs

AI mention rate: the share of AI answers for a specific keyword (job) that mention your brand
- Formula: number of answers that mention your brand / number of valid answers x 100
- Target example: 15% now → 30% in three months
Mention share: your brand's share among all the tracked brands mentioned for the same keyword
- Formula: your brand's count of answers that mention it / sum of that count across all tracked brands x 100
- Target example: 10% now → 20% in three months

Secondary KPIs

Mention position: Where your brand comes in the order of brands named in an AI answer (the earlier, the better)
Mention context: The share of mentions that appear in a positive context
Mention rate by AI platform: The mention rate on each of ChatGPT, Gemini, Claude and Perplexity
Mention rate by keyword: The mention rate for each job (keyword) you track

Planning initiatives

Once your KPIs are set, plan the initiatives that will help you reach your targets. They are easier to organize if you group them into the following categories.

Content initiatives: Create and improve content that AI recommends
Technical initiatives: Implement structured data and improve site speed
PR initiatives: Issue press releases and build media relations
Review initiatives: Earn and manage reviews and word-of-mouth
Branding initiatives: Build authority

Do: Execute and record initiatives

Managing initiative execution

To make sure planned initiatives are actually carried out, manage them with the following framework:

Initiative name: The specific name of the initiative
Owner: The person responsible for execution
Deadline: When it must be done
Target keyword: Which keyword (job) the initiative relates to
Expected impact: Which KPI it should affect, and by how much
Execution status: Not started / In progress / Completed

Example initiatives

Each month, run initiatives like the following in parallel:

Weekly: Post on social media (2–3 posts per week)
Every other week: Publish blog posts (at least 2 per month)
Monthly: Issue press releases (1–2 per month)
Monthly: Add and update FAQ pages
Quarterly: Create and update comparison content
Quarterly: Audit and update structured data

Recording what you did

To measure the impact of each initiative accurately, keep a detailed record of what you did, when and where. Without this record, you cannot analyze which initiative led to which result in the Check phase.

Check: Monitoring and analysis

Monitoring frequency and methods

Daily monitoring

Scope: How AI mentions your brand for your main keywords (your 3–5 most important jobs)
Method: Review the data that AIOPulse tracks automatically in its dashboard
What to watch for: Sudden swings

Weekly monitoring

Scope: AI mention rate and mention share for all tracked keywords
Method: Check the mention rate trend charts to see which way things are moving
What to watch for: Competitor movements, and any brands that have newly started to be mentioned

Monthly monitoring

Scope: A comprehensive review of all KPIs
Method: Create a monthly report and check progress against your targets
What to watch for: Correlation between results and the initiatives you executed, and ROI calculation

Angles for analysis

When you analyze monitoring data, look at it from the following angles:

Trend over time: Is the mention rate rising, falling or flat?
Correlation with initiatives: Did the mention rate change after a specific initiative?
Differences by AI platform: Are there gaps, such as your brand being mentioned by ChatGPT but not by Gemini?
Comparison with competitors: Is the gap between competitors' mention rates and yours narrowing?
Differences by keyword: What separates the keywords where your initiatives are working from those where they are not?

Detecting and handling anomalies

If you detect anomalies like the following, investigate the cause quickly.

A sudden drop in mention rate: Possible causes include an AI model update, a new competitor initiative, or a technical problem on your site (downtime, removal from the index, etc.)
A sudden rise by a competitor: Possible causes include new PR activity, a product launch or a large content update by that competitor
Swings in only one AI: A possible cause is an algorithm change specific to that AI platform

Act: Adjust your strategy based on data

Strengthening initiatives that worked

Strengthen the initiatives that proved effective in the Check phase in the following ways:

Roll out what worked: Apply initiatives that worked for one keyword to other keywords
Invest more resources: Allocate more resources to high-impact initiatives
Standardize what works: Turn the way you ran effective initiatives into templates so the results are repeatable

Reviewing low-impact initiatives

Analyze the cause: Work out why the initiative did not deliver. Was the problem the initiative itself, how it was executed, or the timing?
Adjust the initiative: Change the approach based on the cause and test again
Decide whether to stop: If there is still no effect after 2–3 attempts, shift the resources to other initiatives

Adding new initiatives

As conditions change, it is also important to add new initiatives.

Respond to new AI platforms: Add initiatives when a new AI service becomes widely used
Respond to new competitor moves: Consider counter-initiatives when a competitor starts something new
Respond to industry trends: Review your keywords (jobs) as user needs and the market change

Example PDCA cycle schedule

Start of month (days 1–3): Plan

Review the previous month's report
Set this month's KPI targets
Decide which initiatives to run, and confirm owners and the schedule

Throughout the month: Do

Execute the chosen initiatives as planned
Record what was done and when
Review progress as a team every week

End of month (final week): Check

Create the monthly report
Check progress against your KPI targets
Analyze the correlation between initiatives and results

End of month to early next month: Act

Sort initiatives into those that worked and those that did not
Feed the findings into the plan for next month's initiatives
Adjust KPI targets as needed

Using tools to run the PDCA cycle more efficiently

To run the PDCA cycle efficiently, use tools like these:

AI mention monitoring: Use AIOPulse to automatically track how ChatGPT, Gemini and Claude mention your brand across multiple keywords, in answers to Japanese-language prompts. It charts mention share over time so you can quantify the impact of your initiatives
Project management: Track the status of initiatives in Notion, Asana, Trello or similar tools
Reporting: Generate monthly reports automatically with Google Sheets or a dashboard tool

A GEO PDCA success story

Case study: How a bedding brand raised its AI mention rate

Here is an example of the results a bedding brand achieved after running the PDCA cycle for six months.

Month 1 (baseline assessment)

AI mention rate for the keyword "better sleep recommendations" (in Japanese): 8%
Mention share: ranked 5th (8% share)

Months 2–3 (foundational initiatives)

Implemented structured data on the brand's own website
Created a new FAQ page (30 questions)
Created comparison content
Result: AI mention rate rose to 12%

Months 4–5 (stepping up PR initiatives)

Issued press releases twice a month
Began contributing articles to industry media
Ran initiatives to earn reviews
Result: AI mention rate rose to 22%

Month 6 (optimization)

Strengthened the highest-impact initiatives
Expanded to new keywords
Result: AI mention rate reached 28%, and the brand rose to 2nd place in mention share

Summary

Here are the key points for running a GEO PDCA cycle effectively:

Set clear KPIs: Make AI mention rate and mention share your primary KPIs
Execute initiatives as planned and record them: A record of what you did and when is the foundation for analysis
Monitor regularly: Check on three cadences (daily, weekly and monthly)
Improve based on data: Reinforce high-impact initiatives and review low-impact ones
Keep going: GEO is a long game. Keep running the PDCA cycle for at least six months

Start by getting an accurate picture of how AI mentions your brand today. To see how ChatGPT, Gemini and Claude answer Japanese users' questions about your brand, contact us.