Choose one of three approaches. The steps for that option appear below.
Choose an option above. On a phone, swipe the tabs left or right to see all options →
ChatGPT builds the join audit table, drafts the report and cross-checks it. You check the table and the final draft.
Assemble the exports
Build audit table
Check the table
Draft and cross-check
Check and decide
Use: You
Assemble the campaign cost exports, conversion exports and supplied attribution definitions, retaining authorised source IDs and dates.
Check each item is dated and permitted; list anything missing as still to confirm.
Use: ChatGPT
Ask ChatGPT for a join audit table retaining campaign IDs, periods, currencies and unmatched rows, with source references and uncertain facts.
What to paste into the tool
Task: Join audit table retaining campaign IDs, periods, currencies and unmatched rows; source references; uncertain or absent facts; Source packet: Campaign cost exports, conversion exports and supplied attribution definitions. Use only the supplied packet. Build a labeled table with source IDs, dates, uncertainty and permitted-use limits on each row. Check join keys, duplicate IDs, unmatched rows, periods, time zones and currencies against raw exports; independently recompute totals and record named exceptions. Retain the source packet as an unchanged attachment in the output bundle. Keep unresolved facts and approval dependencies explicit. Do not resolve a missing fact by inventing one. Produce a draft only; do not publish, send, deploy, change accounts, spend or launch a campaign
Use: You
Compare every table row with the source materials and correct it; resolve or retain missing facts before drafting.
Check join keys, duplicate IDs, unmatched rows, periods, time zones and currencies against the raw exports; recompute totals and record named exceptions.
Use: ChatGPT
Ask ChatGPT to reconcile totals with attribution caveats, then cross-check the draft against the checked table and return corrections with a change list.
What to paste into the tool
Task: Draft campaign report with reconciled totals, attribution caveats and unresolved mismatches; Open questions and approval dependencies retained beside the recommendations; Reviewer-checked join audit table retaining campaign IDs, periods, currencies and unmatched rows; source references; uncertain or absent facts; Supplied constraints, retained open questions and approval dependencies with source IDs and dates; Source packet: Campaign cost exports, conversion exports and supplied attribution definitions. Join cost and conversion rows by campaign ID, period and currency, reconcile totals, retain unmatched rows and named exceptions, and state supplied attribution caveats; do not call attributed results incremental impact. Retain the checked table, constraints and source packet unchanged in the output bundle. Keep unresolved facts and approval dependencies explicit. Do not resolve a missing fact by inventing one. Produce a draft only; do not publish, send, deploy, change accounts, spend or launch a campaign. Second message, after your check: Task: Corrected draft with a change list and unresolved mismatches flagged for the reviewer. Compare the draft with the checked join audit table and the retained source packet. Correct every mismatch you find in the draft, keep source IDs, dates, uncertainty and named exceptions, and list each change. Flag anything you cannot settle from the supplied material; do not invent facts, figures, policies or claims. Return preparation material only; publishing, outreach, audience changes, tracking deployment and financial commitments are outside this task.
Use: You
Compare the corrected draft and change list with the checked table and exports, confirm the named exceptions, then decide; publishing stays outside.
Check totals match source exports or named exceptions, and attributed results are not called incremental impact. If a check fails, correct the draft or leave the workflow incomplete.
Claude builds the join audit table, drafts the report and cross-checks it. You check the table and the final draft.
Assemble the exports
Build audit table
Check the table
Draft and cross-check
Check and decide
Use: You
Assemble the campaign cost exports, conversion exports and supplied attribution definitions, retaining authorised source IDs and dates.
Check each item is dated and permitted; list anything missing as still to confirm.
Use: Claude
Ask Claude for a join audit table retaining campaign IDs, periods, currencies and unmatched rows, with source references and uncertain facts.
What to paste into the tool
Task: Join audit table retaining campaign IDs, periods, currencies and unmatched rows; source references; uncertain or absent facts; Source packet: Campaign cost exports, conversion exports and supplied attribution definitions. Use only the supplied packet. Build a labeled table with source IDs, dates, uncertainty and permitted-use limits on each row. Check join keys, duplicate IDs, unmatched rows, periods, time zones and currencies against raw exports; independently recompute totals and record named exceptions. Retain the source packet as an unchanged attachment in the output bundle. Keep unresolved facts and approval dependencies explicit. Do not resolve a missing fact by inventing one. Produce a draft only; do not publish, send, deploy, change accounts, spend or launch a campaign
Use: You
Compare every table row with the source materials and correct it; resolve or retain missing facts before drafting.
Check join keys, duplicate IDs, unmatched rows, periods, time zones and currencies against the raw exports; recompute totals and record named exceptions.
Use: Claude
Ask Claude to reconcile totals with attribution caveats, then cross-check the draft against the checked table and return corrections with a change list.
What to paste into the tool
Task: Draft campaign report with reconciled totals, attribution caveats and unresolved mismatches; Open questions and approval dependencies retained beside the recommendations; Reviewer-checked join audit table retaining campaign IDs, periods, currencies and unmatched rows; source references; uncertain or absent facts; Supplied constraints, retained open questions and approval dependencies with source IDs and dates; Source packet: Campaign cost exports, conversion exports and supplied attribution definitions. Join cost and conversion rows by campaign ID, period and currency, reconcile totals, retain unmatched rows and named exceptions, and state supplied attribution caveats; do not call attributed results incremental impact. Retain the checked table, constraints and source packet unchanged in the output bundle. Keep unresolved facts and approval dependencies explicit. Do not resolve a missing fact by inventing one. Produce a draft only; do not publish, send, deploy, change accounts, spend or launch a campaign. Second message, after your check: Task: Corrected draft with a change list and unresolved mismatches flagged for the reviewer. Compare the draft with the checked join audit table and the retained source packet. Correct every mismatch you find in the draft, keep source IDs, dates, uncertainty and named exceptions, and list each change. Flag anything you cannot settle from the supplied material; do not invent facts, figures, policies or claims. Return preparation material only; publishing, outreach, audience changes, tracking deployment and financial commitments are outside this task.
Use: You
Compare the corrected draft and change list with the checked table and exports, confirm the named exceptions, then decide; publishing stays outside.
Check totals match source exports or named exceptions, and attributed results are not called incremental impact. If a check fails, correct the draft or leave the workflow incomplete.
You run the join and reconcile the totals locally. Julius AI only suggests mismatch explanations, which you check.
Run the join
Suggest mismatch causes
Check and write report
Use: You
Assemble the exports and definitions; run the join by campaign ID, period and currency locally, retain the formulas or SQL queries, and reconcile the totals.
Check each item is dated and permitted; check join keys, duplicate IDs, unmatched rows, periods, time zones and currencies against the raw exports; recompute totals and record named exceptions.
Use: Julius AI
Give Julius AI only the mismatch summary and attribution definitions; ask for explanations with visible analysis code, without changing totals.
What to paste into the tool
Task: Mismatch hypotheses linked to supplied row IDs, generated analysis code and open questions; Human-reconciled totals, retained formulas or SQL queries, attribution definitions and mismatch list keyed by campaign, period and currency. For each mismatch pattern, suggest possible causes such as duplicate IDs, currency, period boundaries or attribution windows. Identify covered row IDs and show any generated code. Treat each explanation as a hypothesis. Do not recompute or change a total. Say when the supplied material does not explain a mismatch. Return the input summary unchanged with the hypotheses and code. Keep unresolved facts and approval dependencies explicit. Do not resolve an open fact by inventing a fact. Produce a draft only; do not publish, send, deploy, change accounts, spend or launch a campaign
Use: You
Check each suggested explanation and code against the local table; resolve or retain missing facts before drafting the report from your reconciled totals.
Check each suggested explanation and code against the local table. Decide approval and any subsequent execution separately.
Best when: You want AI proposals for table, draft and a first cross-check
What it does: ChatGPT builds, drafts and cross-checks; you check both gates
Strong at: Analyses uploaded files; writes and rewrites text
Limit: A table error can carry into the draft; check at the checkpoint
Free plan: Yes, limited (file upload and analysis caps)
Best when: You want AI proposals for table, draft and a first cross-check
What it does: Claude builds, drafts and cross-checks; you check both gates
Strong at: Drafts and edits text in many formats
Limit: Usage capped in 5-hour and weekly windows
Free plan: Yes, limited (5-hour and weekly usage caps)
Best when: You want to own the queries and report; AI only suggests causes
What it does: Julius AI suggests explanations; you run the join and write
Strong at: Totals come from your retained formulas or queries, not AI
Limit: A plausible explanation still needs checking against the raw exports
Free plan: Yes, limited (files stored for only 7 days)
Before you paste into any AI tool, remove any confidential information. Put it back at the end.
Deciding between ChatGPT and Claude for this? See the comparison.
The named reviewer compares working records and final drafts against retained source evidence; unresolved questions stay visible.
These are proposed preparation workflows; execution, publication, sending, deployment, account changes and spending are outside scope.
Check the result against your original inputs. Approve it before sharing or acting.