
The order to run targeting prompts when launching a brand-new client. The prompts themselves are canonical in Targeting Audit Workflow — this page is the launch-specific routing layer (what to run, when, and how to act on the output).
Owner: Tasneem. Reference call: walkthrough with Sean on 2026-05-13.
NOTE
The audit workflow (Targeting Audit Workflow) is for fixing live targeting on an existing client. This page is for setting up correct targeting on Day 1 of a new client. Same prompts, different order and emphasis.
| Step | Prompt | When to use | Output |
|---|---|---|---|
| 1 | Prompt B — Messy Location → Radius | Only if the client gave a messy geo (zip list, county list, screenshot of overlapping bubbles). Skip if they gave a clean radius. | Clean primary radius + zip add-ons + exclusion radius (if needed) |
| 2 | Prompt C — Location Qualifier (Question Meta Prompt v2.0) | Always. Every new launch. | Lead form Q1 (the "Are you a homeowner in [Region]?" question) |
| 3 | Prompt A — Zip Exclusion Research (3-tier model) | Always. Every new launch. | Tier 1/2/3 zip lists with audience math |
| 4 | Apply | Always. | Final FB ad set audience |
All three prompts live in full in Targeting Audit Workflow — copy/paste from there.
From the new client's Discord channel: - Stated geo (radius, zip list, county list, city list, or "around X miles from Y") - Any client constraints ("we don't service downtown", "stay out of XYZ zip", etc.) - Service area boundaries (franchise lines, county-only restrictions, state-only)
If anything's unclear, tag Griffin before running prompts. Don't guess on geography.
Run when: client gave a messy geo input. Examples: - A screenshot of overlapping bubbles in FB Ads Manager - A list of 25+ zip codes - A list of 5+ cities - Conflicting verbal vs. survey input
Skip when: client said something clean like "50 miles around Las Vegas" or "Hamilton County only."
Where the prompt lives: Targeting Audit Workflow → Prompt B — Messy Location → Radius Optimization.
What you'll get back: a primary radius pin, optional secondary radius, zip code add-ons for outliers, exclusion radius if there's a franchise/territory carve-out, and a city callout recommendation for ad copy.
Then: use the resulting [Primary City + Radius] as the input for Steps 2 and 3.
Why first (before exclusion research): the location qualifier question shapes the lead form, and you're building the lead form before the campaign (see Media Buyer Launch Playbook § "Build the lead form FIRST"). So Q1 needs to be ready early.
Where the prompt lives: Targeting Audit Workflow → Prompt C — Lead Form Qualifier (Location Question Meta Prompt v2.0).
Inputs: primary radius from Step 1 (or the client's clean stated radius).
What you'll get back: the exact Are you a homeowner in [Region]? question text + answer options. This text drops directly into the lead form (Lead Form Standard Questions § Q1).
TIP
Prefer regional nicknames ("Treasure Valley", "East Texas", "Lowcountry") over county lists when a strong regional name exists. The prompt is calibrated for this — if it returns a county list when a regional name would work, push back and re-run.
Where the prompt lives: Targeting Audit Workflow → Prompt A — Zip Code Exclusion Research.
Inputs: center city + radius (in miles).
What you'll get back: three tiers of zip codes to exclude, with population math, area names, and reasoning per tier. Plus a "Top 5 Money Zips" callout for context.
Use Claude when available (Sean's preference for citation quality). ChatGPT and Gemini are fine fallbacks — manually verify a couple of zips' poverty/income against unitedstateszipcodes.org if you're suspicious of any output.
Use the audience size after exclusions to decide how aggressive to be. This is the "750K floor" rule from Targeting Audit Workflow § "Audience Size Floor Logic":
| Audience size after exclusions | What to do |
|---|---|
| 1M+ | Apply Tier 1 + Tier 2. Consider Tier 3 if you want to push quality further. |
| 750K – 1M | Apply Tier 1. Selectively add Tier 2 zips (only the worst). |
| 500K – 750K | Apply Tier 1 only. Stop excluding. Focus on creative quality. |
| Below 500K | Don't exclude. Consider interest-narrowing instead (Tasneem's approach — see note below). |
NOTE
For audiences below 500K: instead of brute-force zip exclusion, layer in interest targeting to filter by income proxies (e.g., interests correlated with high-earner status). This is Tasneem's pack-testing approach — see Targeting Audit Workflow § "Audience Pack Adjustment". Run the detailed-interest pack in parallel with the broad pack, not as a 50/50 split.
For the ad set audience:
Audience naming:
[CompanyName]_[City]_1.0
Bump the version when you change exclusions/inclusions. Don't edit the locked 1.0 — duplicate and increment.
Client: Call on the Double (Lubbock, TX) — a new launch pending onboarding survey completion as of 2026-05-13.
Are you a homeowner in West Texas? — used locals' regional name rather than a county list. Drops into the lead form.