Gemini named 155 different brands. ChatGPT named 131 for the same prompt set. An AI-search dashboard should preserve each platform's results before combining them.
These counts come from Dageno's enterprise AI report. It collected 6,623 answers from ChatGPT, Gemini, and Google AI Mode in the United States from July 20 to 27, 2026. The answers contained 41,273 brand mentions and 73,980 cited links.
Store the answer, platform, and question together
Give each answer an ID. Save the sampling date, AI platform, original question, question type, and answer text.
Link product mentions and cited pages to that answer ID. This lets a reviewer open the source answer from a chart and see what the AI said.
About 17% of answers named no brand. Keep those answers in the dataset so a reviewer can inspect their questions.
Keep product and company totals separate
Microsoft Copilot Studio led with 1,899 mentions, or 4.6% of all brand mentions. Microsoft's six products together accounted for 17.0%. Google's products together accounted for 5.5%.
Store the product name and its parent company in separate fields. Display a product ranking and a company total as two views.
Record position separately from mention count. Copilot Studio's average rank was 3.69, while Azure AI Foundry's was 3.63. Lower values mean earlier positions.
The most common pair, AWS Bedrock and Google Vertex AI, appeared together in 1,272 answers. That was about 23% of the 5,501 answers naming a brand. Label the denominator on the chart.
Filter by the type of buyer question
The six question types had different leaders by mention share:
- Use cases: Copilot Studio, 4.8%.
- Comparisons across several products: Snowflake, 5.9%.
- Best overall: Databricks, 4.4%.
- Pricing and value: Agentforce, 5.3%.
- Completing a specific task: Snowflake, 6.7%.
- Choosing between A and B: Copilot Studio, 6.9%.
Keep multi-product comparisons and A/B choices as distinct filters. Combining them would hide their different leaders.
Record the strengths AI assigns to a product
Governance and compliance appeared as strengths 5,175 times. Agent orchestration appeared 1,807 times. IBM watsonx held the largest share of that category at 19%. The category included 116 brands.
Save the strength, product, and answer ID. A team can then review the descriptions associated with its product.
For five comparable brands, the website's most emphasized strength differed from the one AI mentioned most. Their top-three lists shared an average of 1.2 strengths. Show those two lists side by side, with links to the pages and answers.
Store citations separately from recommendations
Vendor sites accounted for 16.1% of cited links, community and video platforms for 9.9%, and review and analyst sites for 1.4%. Other third-party sources accounted for 72.6%, the largest category.
Coworker AI's website was cited 910 times. The brand was recommended 96 times and ranked 89th among 155 brands.
A citation record needs an answer ID, URL, and domain. A product record needs an answer ID, product name, whether it was recommended, and its position. Joining the records shows citations and recommendations within the same answer.
Show dates beside every change
The top-three order changed five times during the eight days. The top ten recorded eight changes in brand membership. The number-one product stayed the same.
The report also compared the first four days with the last four. Qlik rose from 25 to 35 mentions, Supermetrics from 54 to 69, and GoSearch from 48 to 61. Fiddler AI fell from 83 to 55, Cohere from 56 to 38, and Tellius from 87 to 63.
That comparison included 104 brands with at least 20 mentions in the first half and at least 60 overall. Display those inclusion rules alongside the dates.
For each new run, keep the same field names and prompt set. Review results by platform, question type, product, company, and date. Keep the original answer one click away from each record.
Source: Dageno, Enterprise AI Platform GEO Report. Dageno helps teams review brand descriptions and cited sources in AI answers.
Top comments (0)