As of 13 August 2026, AI can only partly optimise your Meta ads.

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, power-user skill, and roughly 1 hour until you can act on the result.

   How to actually do it

   1. Open Meta Ads Manager and export the relevant campaign, ad set and ad level results for a comparable recent period, including spend, impressions, reach, frequency, clicks, CTR, CPC, CPM, conversions, conversion value and cost per result where available.
   2. Gather the campaign objective, conversion event, target locations, budget limits, margin or lead value, recent creative changes and any known tracking problems.
   3. Remove names, email addresses, phone numbers, customer IDs and other unnecessary personal data from the export, then paste the context and data into the prompt.
   4. Ask the model to separate measured results from hypotheses and to show every calculation using only the figures supplied.
   5. Compare its calculations against the original Meta Ads Manager export and reject any recommendation based on a missing, inconsistent or incorrectly attributed metric.
   6. Choose one or two low-risk changes or tests, record the starting metrics and budget, and apply those changes manually in Meta Ads Manager.
   7. After the planned observation period, export the same metrics again and ask the model to compare the periods without declaring a winner unless the test design supports that conclusion.

   Prompt

   Act as a cautious paid-social analyst. I will paste an anonymised export from Meta Ads Manager and supporting business context below. Do not invent missing figures, attribution, customer information or test results. If the data is insufficient, say exactly what is missing.
   
   Business objective: [for example, profitable sales, qualified leads or awareness]
   Primary conversion: [conversion event]
   Target audience: [audience]
   Geography: [target locations]
   Campaign period: [period]
   Budget constraints: [daily or total budget and limits]
   Commercial context: [average order value, gross margin, lead value or other figures, if known]
   Recent changes: [creative, offer, website, tracking or targeting changes]
   
   Meta Ads Manager export:
   [Paste the relevant campaign, ad set and ad level data here, including spend, impressions, reach, frequency, clicks, CTR, CPC, CPM, conversions, conversion value and cost per result where available.]
   
   Analyse the account without pretending that correlation proves causation. First state the data limitations and any tracking concerns. Then:
   1. Summarise performance by campaign, ad set and ad, separating results by objective and conversion event.
   2. Identify the three most credible opportunities and explain the evidence for each.
   3. Identify waste, fatigue, delivery problems, audience overlap or creative weaknesses only where the data supports them.
   4. Recommend what to keep, pause, change or test, with a reason, expected risk and the metric that would decide the next action.
   5. Suggest a test plan that changes one meaningful variable at a time and does not claim a winner before enough evidence exists.
   6. Give exact calculations using only the supplied figures and show the formula for each calculation.
   7. End with a short action list for a human to apply in Meta Ads Manager. Do not make changes, guarantee results, recommend unlawful targeting or use sensitive personal data.

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