As of 13 August 2026, AI can only partly report on your Google Analytics data.

What to actually do

1. Hand it to a person

   The route this page recommends

   A person who owns the outcome does this end to end, worth it when the failure is dear.

2. Use a tool built for this

   Second choice

3. Do it yourself

   The distant third

   A chat interface, chat-fluent skill, and roughly 30 minutes until you can act on the result.

   How to actually do it

   1. Open the relevant Google Analytics property and export the report covering the required date range, comparison period and dimensions, such as channel, landing page, device and conversion event.
   2. Record the property name, time zone, reporting dates, conversion definitions, recent tracking changes and the business question the report must answer.
   3. Remove unnecessary personal data, identifiers and query-string information from the export, then save the cleaned file with the reporting and comparison dates in its name.
   4. Open a chatbot or Julius AI, attach the cleaned export, paste the prompt and replace each bracketed slot with your recorded context.
   5. Ask the model to show the source rows or calculations behind every headline figure, then correct any calculation that does not match the exported data.
   6. Compare the report's totals, dates, filters and conversion definitions with the corresponding Google Analytics reports and ask a colleague familiar with the property to check the interpretations before sending it.

   Prompt

   Report on the attached Google Analytics export for [website or business name]. The reporting period is [date range], and the comparison period is [comparison date range]. The purpose of the report is [decision or question]. The intended reader is [audience].
   
   Use only the data provided. Do not invent figures, causes, campaigns, events or business context. Separate the report into: 1) data and date range used, 2) key findings with exact figures, 3) comparisons with the previous period, 4) notable changes by channel, landing page, device or other available dimension, 5) limitations and possible tracking or attribution issues, and 6) practical actions to investigate next.
   
   For every percentage, rate or change, show the underlying figures and the calculation. Distinguish clearly between an observed result, a calculation and an interpretation. If a requested metric or dimension is absent, say so. Do not treat correlation as causation. Flag unusual values, missing values, sampling, consent-related gaps and changes in tracking configuration where the data suggests them, but do not claim a cause without evidence.
   
   Write in plain British English. Begin with a short executive summary, then provide a concise table of the main metrics and a detailed report. End with five specific questions that a colleague should answer before this report is used for a business decision.

   Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.