# Plan a capped Meta creative experiment Canonical page: https://agents.paretoid.com/workflows/plan-a-capped-meta-creative-experiment/ Category: Meta Ads Practice relevance: cross-practice. A general operational method. Review the stated tool and jurisdiction limits before use. Jurisdiction/provider scope: Write an offline test plan with a supplied spending ceiling and an inconclusive-result option. Status: Proposed / not tested. The public source describes the problem. Documentation supports only the stated method components; no result from this recipe has been verified. Verification: Source methods and reported problems have been reviewed. We have not run the adapted workflow. The sample below describes expected behavior on synthetic data, not a customer result. ## The problem Write an offline test plan with a supplied spending ceiling and an inconclusive-result option. Why it matters: Small differences in ad results can tempt a firm to scale before the evidence supports it. Predeclared metrics, costs, and stop rules make the decision clearer while allowing the result to remain inconclusive. ## Sources and evidence - pain: [Pain evidence: Law firm, small ad budget](https://www.reddit.com/r/FacebookAds/comments/1ucpbxd/law_firm_small_ad_budget/) 2026-06-22, self-described law-firm campaign operator Late-Spite-7865, Law firm, small ad budget. The firm has a £900 first-month budget. The poster asks how many ads to create and “How would you identify winners (based of what metrics)” before scaling. The reported budget is context, not a recommended budget or validated result. Same author as M02; this is an experiment-design problem rather than a production problem. Date verification: the indexed original-thread text explicitly displayed “Monday June 22 2026.” The separately opened cached page displayed “23d ago” despite a three-month-old crawl; that stale relative label was not used to calculate the date or claim recent-30-day evidence. - method: [Method: Meta Blueprint: Designing A/B tests, creative tests and lift studies](https://www.facebookblueprint.com/student/path/219763-ab-test-conversion-brand-lift-course) Meta Blueprint: Designing A/B tests, creative tests and lift studies explicitly describes A/B tests, creative tests, conversion-lift and brand-lift studies, Ads Manager/Experiments, setup requirements and conclusive versus inconclusive results. The public overview does not establish eligibility or budget minimums for a particular account. No lift study is promised for a small budget. Source checked: October 15, 2026. Pain accounts are self-reported. Documentation supports components; this adapted recipe and prompt have not been tested. ## Suggested approach Start when: Staff have two approved creatives and a fixed experiment allowance 1. predeclare the one question, primary metric, measurement window, budget ceiling, safety stops and result rule 2. Keep source references and report unknowns instead of filling gaps. 3. The responsible reviewer checks the report before any separately approved action. ## What you need - Existing authorized access to the correct source account and records; no new login, credentials, installation or permission grant is included. - Responsible lawyer/operations reviewer approves scope, definitions and any later action. - AI optional; use only approved minimized data and retain originals separately. - Use synthetic inputs first. Keep real data in firm-approved accounts and processing, with appropriate access, confidentiality and retention controls. Aliases alone do not anonymize records. Never paste credentials, API keys or unnecessary client details. ## How to use Planning path only; current account-specific Meta setup/export UI unverified; launch blocked; not tested - Current exact Meta experiment setup and export controls were not publicly verified. This remains a planning recipe, not an operational launch guide. Access and setup: - Meta help page redirected to login; no login attempted. Public Blueprint overview confirms feature family only. ### 1. Acquire an approved experiment brief and native eligibility evidence Tool: Firm marketer / Meta Ads Manager Location: Existing approved creative brief; Ads Manager Experiments feature (exact live submenu/report-export route unverified) Collect two approved creative IDs, one variable, primary metric/source, attribution window, all-in currency allowance, fees/taxes, dates, stop authority and historical aggregate volume. Staff with existing access captures available experiment options and displayed eligibility requirements. Do not create, publish or start a test. Expected output: Verified planning inputs and account-option capture, or a launch-blocked brief. Next step: Prepare labeled ApprovedCreatives, Budget and Measurement sections plus full Scope with source evidence IDs. Paste this complete nonconfidential packet into Acrobat. Missing actual controls, currency/fees or measurement evidence stays visibly blocked. Check before continuing: Meta Blueprint verifies experiments and conclusive/inconclusive results. Its public lesson links and a help page did not expose full current setup/export controls without login; no exact eligible option, report export or hard all-in cap is claimed. Execution status: Public official documentation checked 2026-10-15; no login, export, live query, import, installation, configuration, message or client workflow was executed. Step documentation: https://www.facebookblueprint.com/student/path/219763-ab-test-conversion-brand-lift-course ### 2. Draft from the approved source packet Tool: Acrobat PDF Spaces / AI Assistant, optional Location: Acrobat home > PDF Spaces > Create a PDF Space > Paste copied text > Add to PDF Space > chat Paste only the approved nonconfidential packet, source IDs and full Scope note. Paste the exact prompt into the Space chat. Do not upload raw client records; this is a downstream drafting assistant, not an AI in the source application. Prompt: ```text Draft an offline experiment brief using only ApprovedCreatives, Budget, Measurement and Scope. Compare the two approved creative IDs and vary only the supplied creative element. State the question, primary metric, source, dates, equal-treatment assumptions, all-in ceiling, stop owner and triggers for broken measurement/privacy/policy issues. If fees/taxes, platform eligibility, budget controls or metric definitions are unknown, mark launch blocked. Include an inconclusive result and no automatic scaling. For any existing result capture, show observed spend/outcome arithmetic and the platform’s actual confidence fields only; never invent confidence, sample sufficiency, power or a winner. Do not target sensitive legal circumstances or start/spend/publish. ``` Expected output: Approval-ready experiment specification, reporting template and blockers. Next step: Copy reviewed output into an internal review document or the named downstream tool, preserving evidence IDs. Check before continuing: A lower observed cost per result alone does not establish a winning randomized test. Execution status: Public official documentation checked 2026-10-15; no login, export, live query, import, installation, configuration, message or client workflow was executed. Step documentation: https://helpx.adobe.com/acrobat/desktop/explore-pdf-spaces/create-pdf-spaces.html https://helpx.adobe.com/acrobat/desktop/use-acrobat-ai/generative-ai-features/ai-get-answers.html ### 3. Prepare the evidence-reading handoff Tool: Marketer and Excel reviewer Location: Existing Meta Experiments result view; manually captured aggregate results if no verified export is available After a separately authorized experiment exists, the operator supplies its actual test ID, dates, variant allocation, metric definitions, spend, outcome counts and platform confidence/result labels. Manually capture visible results with screenshots if an export route is not verified. This future evidence is not fabricated now. Expected output: A checked results ledger or a results-unavailable blocker. Next step: Use Excel for visible arithmetic, then return the platform’s actual conclusive/inconclusive status to the review document. Check before continuing: Ordinary unequal ad delivery is not assumed to be a randomized experiment. Native account/budget controls may not cap taxes and fees. Execution status: Public official documentation checked 2026-10-15; no login, export, live query, import, installation, configuration, message or client workflow was executed. Step documentation: https://www.facebookblueprint.com/student/path/219763-ab-test-conversion-brand-lift-course ### 4. Human review and final action Tool: Responsible firm reviewer Location: Internal Review worksheet / review document Attorney/marketer approve creatives, measurement, budget and any future launch separately. Save the plan with all unknowns. Expected output: Capped planning brief; launch blocked until account evidence and approvals exist. Next step: Authorized marketer verifies current official controls inside the actual account before building. Check before continuing: No live campaign, ad, audience, conversion-data upload or spend change. Execution status: Public official documentation checked 2026-10-15; no login, export, live query, import, installation, configuration, message or client workflow was executed. Step documentation: https://www.facebookblueprint.com/student/path/219763-ab-test-conversion-brand-lift-course ## What you will get an approval-ready experiment brief and results-reading template that can return “inconclusive. ## Human review and limits Budget is user-supplied, never invented or raised. No guaranteed statistical sufficiency. No prediction of retained clients, revenue or ROI from clicks. A privacy-safe measurement path must already exist; this card does not set up Pixel/CAPI or offline lead uploads. If the desired metric cannot be measured safely, the plan remains blocked or uses a separately approved non-sensitive metric with its limitations explicit. The assistant drafts or reports only. Check source records and review the output before a human performs any separately approved send or change. No automatic legal decision, money movement, account connection or publication is included. What to measure: In a synthetic pilot, count correctly found exceptions, false alarms, and unresolved inputs. No time or cost saving is assumed. ## Original prepared-input schema (1) two already attorney-approved creative IDs; (2) one business question and one primary metric definition, including the source and attribution window; (3) the firm's total all-in allowance, currency, maximum experiment portion, dates, expected fees/taxes if known, reviewer and stop authority; (4) current account screenshots for eligible experiment options and any displayed requirements; (5) aggregate historical volume if available, or “unknown”; (6) optionally an existing experiment export with variant, spend, primary outcomes, dates and the platform's actual confidence/result fields. Do not provide raw leads or case outcomes tied to people. Synthetic example input: ```text Monthly allowance GBP900; staff-set experiment allowance GBP200 all-in; approved A/B creative IDs; day14 review. Fees and eligibility unverified. Illustrative supplied report: A spent GBP100 for 2 outcomes; B spent GBP100 for 3. No platform confidence field. No personal lead data. ``` Expected synthetic output: ```text Plan blocked until fees, eligibility and a safe measurement path are verified; no use of the remaining GBP700. Observed cost per outcome: GBP50 for A; GBP33.33 for B. Inconclusive; no invented winner or confidence. GBP200 reported spend reaches the synthetic ceiling. Staff review only; no automatic scaling or extra spend. ``` Failure check: Budget is user-supplied, never invented or raised. No guaranteed statistical sufficiency. No prediction of retained clients, revenue or ROI from clicks. A privacy-safe measurement path must already exist; this card does not set up Pixel/CAPI or offline lead uploads. If the desired metric cannot be measured safely, the plan remains blocked or uses a separately approved non-sensitive metric with its limitations explicit. ## Starter files Native blank scaffolds and original synthetic examples may use different schemas. Follow NATIVE-INPUT-MAPPING where supplied. These files do not connect accounts or install automation. - [Download starter kit](https://agents.paretoid.com/assets/starter-kits/m03-starter-kit.zip) — All inputs, synthetic examples and review notes in one ZIP. No automation is installed. - [Start here: get the sources and prepare the review](https://agents.paretoid.com/assets/starter-kits/m03/START-HERE.txt) — Documentation-checked acquisition, explicit manual mapping, exact prompts and human review limits - [Experiment input preparation checklist](https://agents.paretoid.com/assets/starter-kits/m03/INPUT-CHECKLIST.txt) — Input preparation and review instructions - [Synthetic experiment planning and results fixtures](https://agents.paretoid.com/assets/starter-kits/m03/SYNTHETIC-BRIEF-AND-RESULTS.txt) — Separate synthetic planning and result-reading examples - [Setup and fixture review guide](https://agents.paretoid.com/assets/starter-kits/m03/README.txt) — Recipe-specific setup, input rules, offline prompt, expected result and boundaries - [Native input scaffold: ApprovedCreatives](https://agents.paretoid.com/assets/starter-kits/m03/native-approvedcreatives-template.txt) — Approved text input for ApprovedCreatives; complete from verified evidence and include in the AI-visible packet - [Native input scaffold: Budget](https://agents.paretoid.com/assets/starter-kits/m03/native-budget-template.txt) — Approved text input for Budget; complete from verified evidence and include in the AI-visible packet - [Native input scaffold: Measurement](https://agents.paretoid.com/assets/starter-kits/m03/native-measurement-template.txt) — Approved text input for Measurement; complete from verified evidence and include in the AI-visible packet - [Native input scaffold: Scope](https://agents.paretoid.com/assets/starter-kits/m03/native-scope-template.txt) — Approved text input for Scope; complete from verified evidence and include in the AI-visible packet - [Native input mapping and scope guide](https://agents.paretoid.com/assets/starter-kits/m03/NATIVE-INPUT-MAPPING.txt) — Old-file-to-native-input compatibility map, acquisition requirements, privacy and blocking gates ## Optional original prepared-input prompt Destination: Your firm-approved AI assistant · prepared files This original prompt uses its own prepared-input schema. Do not mix it with native-step tables without the supplied mapping. ```text Draft an offline experiment brief comparing only the two approved Meta creative IDs supplied. State one question, the one variable changed, primary metric and source, fixed review dates, allocation assumptions, and a maximum experiment cost that does not exceed our supplied allowance. Reserve room for any supplied fees/taxes; if these or platform eligibility requirements are unknown, mark launch blocked pending verification. Include a stop rule for the supplied cap, a broken measurement path, a policy/privacy issue or the planned end date. Staff alone execute stops. Explain what must stay constant and why an ordinary unequal-delivery report is not automatically a randomized A/B test. Give a result template with 'inconclusive' and no automatic scaling. If a results export is included, report observed arithmetic and the platform's actual result fields without inventing confidence, sample size, statistical power or a winner. Do not start a test, change spend or send conversion data. INPUTS (1) two already attorney-approved creative IDs; (2) one business question and one primary metric definition, including the source and attribution window; (3) the firm's total all-in allowance, currency, maximum experiment portion, dates, expected fees/taxes if known, reviewer and stop authority; (4) current account screenshots for eligible experiment options and any displayed requirements; (5) aggregate historical volume if available, or “unknown”; (6) optionally an existing experiment export with variant, spend, primary outcomes, dates and the platform's actual confidence/result fields. Do not provide raw leads or case outcomes tied to people. OUTPUT an approval-ready experiment brief and results-reading template that can return “inconclusive. Use only these synthetic or firm-approved inputs and preserve source row/page IDs. Keep missing or conflicting evidence unresolved. Return a draft for the named reviewer. Do not send messages, connect accounts, change records or permissions, move money, or make legal decisions. Keep source files unchanged. List missing prerequisites before continuing. ``` Proposed / not tested · editorial prompt Published by Law Firm Agents. Proposed adaptation; not tested. Not legal advice.