Tactic 1: Fix objective mismatch first
If your campaign is set to Traffic or Awareness but your goal is leads or purchases, Meta is optimizing for the wrong signal. It will find people who click or scroll, not people who convert. Switching to Leads or Sales objective is usually the single highest-impact change in an underperforming account.
Check every active campaign: does the objective match the business goal? If not, duplicate the campaign with the correct objective, run both for a week, then pause the underperforming one.
Tactic 2: Rotate creatives every 10–14 days
Creative fatigue is the most common driver of rising CPL. As frequency climbs above 3.5, CTR drops and Meta's delivery algorithm deprioritizes the ad. CPL rises not because the audience changed but because the creative stopped performing.
Set a rotation schedule rather than waiting for performance to decline. New creative introduced proactively maintains CPL at a lower level than reactive swaps made after frequency peaks.
- Pause any creative with frequency > 4.5 and CTR down more than 25% from its 7-day peak
- Keep 4–6 active creatives per ad set to give Meta options for delivery
- Test one new format each cycle — static, video, carousel — to find what your audience responds to
Tactic 3: Expand and refresh audiences
Interest audiences exhaust over time, especially at smaller budget levels. If an audience has been running for 60+ days without a refresh, the algorithm has likely reached the most responsive segment and is now cycling through lower-quality users.
Build lookalike audiences from your best converters. A 1% lookalike from Customer Match or pixel purchasers typically outperforms broad interest targeting once you have 500+ seed records.
- Lookalike 1% from past customers or high-intent events (InitiateCheckout, Lead)
- Broad targeting (no interest layers) with Advantage+ Audience enabled
- Retargeting: website visitors last 14 days, video viewers 25–50%, lead form openers who did not submit
Tactic 4: Improve post-click conversion rate
Cutting CPL through bidding and creative is limited if the landing page converts at 2%. A landing page improvement from 2% to 4% effectively halves CPL without touching Meta at all.
The fastest fixes: remove navigation links that let visitors leave without converting, match the headline to the ad copy, add a number-backed proof point above the fold, and reduce the number of form fields to the minimum required.
Tactics 5–8: Bidding, placements, timing, and lead forms
Switch to Cost Cap bidding once you have 30+ conversions to anchor Meta's spend around your target CPL. Below that threshold, Lowest Cost usually outperforms because Cost Cap needs conversion data to work well.
Use instant lead forms for cold traffic — they pre-fill the user's contact information and remove the friction of leaving the platform. Accounts that switch from link-to-site to instant forms for lead generation typically see a 30–60% CPL reduction on cold audiences.
- Exclude placements that historically convert below 50% of account average
- Daypart if your data shows conversion rates drop significantly outside business hours
- Test a shorter instant form (2 questions vs 5) — fewer fields almost always improves completion rate
- Use lead quality questions only if you have genuine downstream qualification needs
Doing this with Claude connected to the account
Eight tactics is a lot to check by hand every week, and most accounts only need two of them at any given time. With Meta Ads connected through MCP Ads, the first job is finding out which two, and Claude can do that from the account rather than from a checklist.
Prompt 1: "Diagnose why cost per lead went up in the last 14 days versus the 14 before. Check objective per campaign, frequency per ad set, CPM trend, placement split, and landing page conversion rate from the pixel." Claude calls meta_ads_campaign_performance for the objective and cost trend, meta_ads_frequency for fatigue, meta_ads_placements for the placement split, and meta_ads_pixel_stats to see whether leads dropped at the click or after it. What comes back is a ranked list of causes with the numbers behind each. As an illustrative example: frequency on the main retargeting ad set rose from 2.8 to 5.1, CPM is flat, and pixel conversion rate is unchanged — so the answer is creative fatigue, not the landing page, and tactics 2 and 3 are the ones to act on.
Prompt 2: "Show every ad with more than $150 spend, sorted by cost per lead, and flag any where CTR has fallen more than 30 percent since its first week." This is meta_ads_ad_performance with a week-over-week comparison. The response is the pause list and the refresh list, already separated. Pausing runs through meta_ads_update_status, which is a write, so Claude names each ad and waits for approval.
Prompt 3: "Duplicate the best ad set into a broad-targeting version with the same budget, and set an automated rule that pauses any ad set whose frequency passes 4." Claude uses meta_ads_duplicate_adset and meta_ads_create_automated_rule. Both are writes with the same confirmation step, and the duplicated ad set starts paused so you can check the targeting before it spends. If you would rather watch than automate, meta_ads_frequency_watcher runs the frequency check as a read and just tells you.
The creative fatigue detector in the skills library runs prompt 1 and prompt 2 weekly and sends you the two lists.
Where the manual way still wins
The API can tell you an ad is fatigued; it cannot tell you what to replace it with. Creative direction — the hook, the offer, the format — is still a person's job, and the difference between a good refresh and a bad one is bigger than any bidding change. Claude can draft copy, but the decision about what the brand is willing to say belongs to you.
Landing page conversion rate through the pixel is also only as good as the pixel setup. If the Lead event fires on the form load instead of the submit, every diagnosis downstream is wrong, and nothing in the ad account will reveal that. Test the form once in a browser before trusting the funnel numbers.
