As of 13 August 2026, AI can analyse your email campaign results.

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 your email platform's campaign reporting area and export the results for the relevant campaigns as a CSV or spreadsheet, including delivery, bounce, open, click, unsubscribe and conversion fields where available.
   2. Remove recipient names, email addresses and other unnecessary personal data, then keep campaign names, dates, audience segments and aggregated performance figures needed for comparison.
   3. Write down the campaign objective, date range, audience or segment definitions, conversion definition and any changes to tracking, consent or email-platform reporting during the period.
   4. Open an approved AI chat or Julius AI, upload the cleaned export, paste the context and metric definitions, and run the prompt.
   5. Compare every headline figure and percentage-point change in the response with the original platform export, checking that the denominators and date ranges match.
   6. Ask the model to recalculate any disputed figure and to label each explanation as observed, possible or unproven before accepting the report.
   7. Send the checked findings and proposed tests to the campaign owner or marketing team, and record which recommendation will be tested rather than treating correlation as proof.

   Prompt

   Analyse the attached email campaign results for [campaign objective] over [date range]. The data came from [email platform] and the platform defines the metrics as follows: [paste metric definitions if available]. The intended audience was [audience and segment details].
   
   First, check the data for missing values, duplicate rows, inconsistent date ranges and metrics that cannot be compared fairly. State any data-quality problem before drawing conclusions. Then:
   1. Summarise deliveries, bounces, opens, clicks, click-to-open rate, unsubscribes and conversions where those columns exist.
   2. Compare campaigns, subject lines, send dates, audience segments and message types where the data supports a fair comparison.
   3. Separate directly observed results from possible explanations. Do not claim causation from this dataset alone.
   4. Identify the strongest and weakest results, unusual changes and patterns that may be misleading because of small or different-sized groups.
   5. Give up to five practical next actions, each linked to a finding and labelled as a test, an operational fix or a decision requiring more evidence.
   6. Show the calculations behind every percentage-point change and use only figures present in the data. Do not invent benchmarks, campaign details or reasons for recipient behaviour.
   
   Present the result under these headings: Data checks, Key results, Comparisons, What the data does not prove, Recommended next tests, and Questions to investigate. Keep the language suitable for a UK marketing team.

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