---
name: ben-heath-meta-ads
description: Plan, audit, test, scale and troubleshoot Meta ads using a study summary of Ben Heath's public methodology. Covers offers, campaign structure, targeting, creative diversity, testing, budgets, lead quality, Pixel and Conversions API, account diagnostics and omnipresent content. Includes an audit checklist and separately labelled adaptations for Egypt and GCC markets. Use with ad-account exports or business data; distinguish CRM evidence from Ads Manager metrics.
---

# Ben Heath Meta Ads Method

> **Unofficial.** A study summary of Ben Heath's free public YouTube videos and podcast
> interviews, compiled by Hossam Ahmed (https://7ussam.net). Not affiliated with or endorsed
> by Ben Heath or Heath Media. Watch his channel for the originals:
> https://www.youtube.com/@BenHeath

This skill applies Ben Heath's publicly taught Meta ads method (YouTube channel "Ben Heath",
Heath Media) to client accounts. His advice changes quickly. Where his old and new advice
differ, use the newest dated advice in `references/framework.md` (section "What changed").

Reference files (load the one you need):
- `references/framework.md`: the full method by topic, with numbers and dates.
- `references/creative.md`: offer, angle, format, hook, copy and proof playbook, plus creative testing.
- `references/audit-checklist.md`: a step-by-step audit of an existing ad account.
- `references/sources.md`: every video, podcast, program and book used, with URLs.
- `references/gcc-adaptation.md`: our own notes for Saudi, GCC and Egypt clients. These are not Ben's words.

## Response language and evidence

- Reply in the language requested by the user; otherwise match their current message. English instructions and Arabic examples do not force the response language. Keep metric names, identifiers and code unchanged when useful.
- Before an audit, state the data source, reporting period, timezone, currency and missing fields. Separate observed facts, hypotheses and proposed tests.
- CRM orders and website analytics are not Ads Manager insights. A Facebook/Instagram source does not prove paid attribution. Do not calculate CPA or ROAS without matching spend and attributed results, or infer campaign settings, creative fatigue or tracking health from orders alone.
- Distinguish placed, paid, delivered, cancelled and refunded orders; avoid double-counting split orders. Label gross order value separately from collected revenue or profit. Compare like-for-like periods and allow time for fulfilment.
- With partial data, complete the supported parts of the audit and list the minimum missing inputs. Do not invent metrics or recommend budget increases without profitability evidence. Treat thresholds below as diagnostic heuristics, not proof of a cause.

## Principles to apply every time

1. **The offer comes before everything.** No conversions usually means an offer problem.
   Creative is the second most important lever. Targeting matters much less than most
   advertisers think: avoid big mistakes and leave the rest broad.
2. **Meta optimises literally.** The *performance goal* is the most important setting. Use the
   Sales or Leads objective, optimise for the deepest event you can track accurately with
   enough volume, fire that event only for qualified leads, and optimise for value when
   order values vary.
3. **Consolidate.** Use as few campaigns and ad sets as possible: often one campaign and one
   ad set per product range, offer or market. Put cold and warm audiences together, and run
   many diverse ads per ad set (20+ is fine after the Andromeda update).
4. **Creative diversity, not variations.** Use different angles, formats, styles, avatars
   and awareness stages. To test small tweaks (hooks, overlays), use Meta's Creative Testing
   tool or push delivery.
5. **Judge on one north-star metric.** Use ROAS when you can track value, otherwise cost per
   result. Use secondary metrics only to find out *why* something wins or loses.
6. **Patience is measured in conversions, not days.** Don't tinker. Batch changes on a fixed
   schedule that matches the account's conversion volume.
7. **Scale existing campaigns gradually.** Never duplicate a campaign to scale it, and never
   make a 10x jump. Expect ROAS to fall as spend grows, so scale only from headroom.
8. **Win beyond the buttons.** The edge comes from things outside Ads Manager: the offer,
   proof, brand, creators, lifetime value (LTV) and data accuracy.

## Procedure A: Audit an existing account

Use `references/audit-checklist.md` and work through it in this order:
1. **Business numbers.** Find target CPA or ROAS, breakeven, customer value or LTV, margins,
   and the lead-to-customer rate. If unknown, request them and continue the evidence-supported
   checks; defer profitability and budget decisions until they are available.
2. **Tracking.** Check pixel + Conversions API, event match quality, and that purchase or
   lead events don't fire on page load (Pixel Helper). Compare Meta's numbers with the
   CRM, Shopify, Salla or Zid. Check that engaged audience and existing customers are
   defined in Advertising settings.
3. **Objectives and goals.** Every direct-response campaign should be Sales or Leads, with a
   conversion performance goal. Flag link-click, landing-page-view, reach or impression goals
   inside those campaigns, and any use of Traffic or Engagement "to warm the pixel".
4. **Structure.** Count campaigns and ad sets, and conversions per ad set per week (target 50,
   or at least 20–25 for a purchase goal). Flag duplicate campaigns, splits between cold and
   warm or between interests, and splits by funnel stage.
5. **Targeting.** Controls should hold only location, minimum age (only if a law requires
   it), language (only if relevant) and needed exclusions. Everything else should be a
   suggestion. Check value rules, and whether "further limit reach" is used without a reason.
6. **Creative.** Count active ads, and their formats and styles. Check diversity, hooks,
   proof, a clear CTA, landing-page match, creative enhancements and personalized
   destinations (Meta re-enables these when ads are edited), and fatigue (frequency rising
   while results fall).
7. **Testing and scaling.** Is new creative actually getting spend? Is there a testing
   mechanism, and how are budgets changed?
8. **Diagnosis.** Use the decision table below and write the top 3–5 fixes, ordered by leverage.

## Procedure B: Plan a new campaign

1. Fix the offer first: differentiation, guarantee, urgency or scarcity, and proof.
   Use Hormozi's value equation (see framework, section "Offer and LTV").
2. Define the ideal customer and list the buying reasons. Each reason becomes one ad angle.
3. Choose the strategy:
   - **Direct-to-offer**, the default for most businesses.
   - **Omnipresent content** alongside a conversion campaign, only for high-value or
     expertise purchases with LTV of roughly $1,000+.
4. Check tracking: pixel + CAPI, with qualified-lead or purchase events and values.
5. Build the campaign:
   - Sales or Leads objective, with daily budget, conversion goal and Advantage+ placements.
   - One campaign per product range, offer or market. One ad set.
   - Controls: location, legal minimum age, and language only if it matters.
   - Suggestions: all custom audiences, plus optionally an age range and one to a few interests.
   - Turn on value rules if the numbers support them.
6. Creative: at least 4 genuinely different ads to start. Aim for 10–20+ over time across
   angles and formats, including both statics and video. Use 2–5 copy variations inside each
   ad rather than separate ads.
7. Budget: an amount the client can afford to lose. For accounts under about $3,000 a month,
   use the small-budget rules (one offer, one campaign, one ad set, direct conversion goal).
8. Set a review schedule based on conversion volume (from roughly every 5–10 days up to monthly)
   and write it down before launch.

## Procedure C: Test creative

Test in this order: offer → angle → format and style → hook → headline, primary text, CTA
and other small details. Isolate one variable and start each test with a hypothesis. To test
a format or style, use a batch of ads, not a single ad.

Choose the mechanism by conversion volume (optimised event per week):
- **Under ~50 a week:** add clearly different new ads to the existing ad set. For a specific
  new ad, use **push delivery** (ad level; default 20% of budget, 7 days, one ad at a time).
- **Dozens to hundreds a week:** use the **Creative Testing tool** (ad level). It takes 2–7
  ads (older UI: 2–5) and you set the budget share and duration. Set the comparison metric to
  CPA, cost per lead or ROAS, never cost per engagement. It is the right tool for hook and
  overlay variants that Andromeda would otherwise merge.
- **50+ a week (comfortable at 100–200+) with a strong evergreen winner:** a separate testing
  campaign is acceptable. Graduate winners into the scaling campaign.
- **Do nothing** during one-off sales (BFCM, single promotions), on tiny budgets, or with huge
  audiences that are slow to fatigue.

Testing budget: 10–20% when the account is very profitable, 20–50% (up to 60%) normally, and
up to 100% when nothing converts. Accept a short-term hit of 10–20% in exchange for the chance
of finding an ad that is 50% better.

Deciding a test:
- Judge on the north-star metric with enough conversions. Three conversions is not enough.
  Use a significance calculator and act at 85–90% confidence.
- Kill early only when the ad is far off target, for example at 2x the target CPA.
- If in doubt, be patient.
- Never turn off the account's best ad to force spend onto new ones.

## Procedure D: Scale

Prerequisite: the campaign runs profitably with headroom. For example, if 3x ROAS is the
minimum, run at 4–5x before scaling. Add new creative as you scale.
- **Option 1, automated (for advertisers who tinker):**
  - Increase rule: raise the daily budget 3% once a day when CPA is below target (or ROAS above
    target). Use the last 7 days, or 3 days with high volume, or 14–28 days with low volume.
    Keep the "lifetime impressions > 8,000" condition, set a maximum budget cap, and run daily
    around midnight.
  - Paired decrease rule: lower the budget 3% when CPA is above a higher threshold, with a
    budget floor. Leave a buffer band between the two thresholds.
  - Up to 5–7% a day at most.
- **Option 2, manual stepped (for disciplined advertisers):** make larger steps less often. Wait
  5–10 days between steps (never under 5), and make the percentage smaller as spend grows. His
  example ladder: 50→100→150→200→300→400→500→650→800→1,000 per day (from +100% at the start to
  +20% at $800+). At about $100k a month, even +20% in one go can be too much. Hold or step back
  when results slip past target.
- **Never:** duplicate campaigns to scale horizontally, jump 10x in one go, or use lifetime
  budgets (except where a feature requires them).
- **Find the scaling ceiling:** the budget at which ROAS reaches the minimum. To go past it,
  improve the offer and creative.
- If results collapse as soon as you scale, the ads only convert warm or hyper-responsive
  people. Fix the ads and the offer, not the budget.

## Decision rules (diagnosis table, his 2026 framework)

| Symptom | Likely cause | Action |
|---|---|---|
| No conversions at all | Offer | Rework the offer (value equation, guarantee, urgency, proof) |
| Some conversions, hook rate < 5% and link CTR < 1% | Creative | New hooks, angles and formats |
| Hook rate > 10% but link CTR < 0.5% | Weak offer or too little proof | Add testimonials and proof, sharpen the offer |
| High hook rate, low average watch time | Misleading hook or weak body | Rework the body, align the hook |
| Good CTR, poor conversion rate (e-com < 2%) | Landing page / post-click | Match the page to the ad; fix price shock, shipping, reviews and speed |
| Link clicks vs landing page views gap > 50% | Slow page or accidental clicks | Fix page speed and placements |
| Lots of conversions, unprofitable | Business model | Raise AOV or LTV (upsells, bundles, subscriptions, pricing) |
| Lots of conversions, profitable | Ready to scale | Procedure D |
| Results fall while frequency rises | Ad fatigue | New creative. Cold frequency should stay ≤ ~2.5; warm 6–8 is normal |
| Learning Limited | Too few events | Consolidate, improve performance, go higher-funnel only if 50/week is unrealistic |
| New ads get no spend | Meta favours proven ads | Push delivery, the Creative Testing tool, or a testing campaign; never turn off the winner |
| Leads are cheap but poor | Meta optimising for the cheapest leads | Higher intent form + SMS OTP, autofill off, conditional logic, qualified-lead event, CRM conversion leads |

Additional rules:
- An ad that gets spend but few conversions inside a mixed ad set may be doing top-of-funnel
  work. Turn off only the ads Meta isn't spending on.
- Compare results month on month, or year on year for seasonal businesses, never day by day.
- Never kill an ad that has a good ROAS because of a low CTR.
- Lead-gen gender skew: change the *messaging* rather than the targeting.
- Ignore Meta's warnings and Meta reps only when you are following a deliberate strategy
  (for example omnipresent content or true retargeting) and know why.

## Output format for client work

When auditing or planning, return:
1. The numbers used (target CPA or ROAS, breakeven, LTV, conversions per week).
2. Findings grouped by leverage: offer → tracking/optimisation → structure → creative → budget.
3. Specific changes, each with the rule behind it (cite the video date from `sources.md`
   when the rule is time-sensitive).
4. The test plan and review schedule.
5. Anything adapted for the GCC/Egypt context, marked as **our adaptation**.

Caveats:
- Some numbers come from auto-captions. The ambiguous ones are flagged in the references.
- Hyros, ClickMagick, HubSpot, Higgsfield and Hollow appear in his videos as **sponsors**.
  Don't present them as part of his method.
- Meta UI labels and limits change often. Check them in the live account before giving
  click-by-click steps.
