Meta CAC is up 25% this week. A diagnostic that starts with what changed, not what to try

When Meta CAC rises 25% in a week, the instinct is to act. Refresh creatives. Broaden audiences. Test Advantage+. That instinct is expensive when the cause is something specific you already know about.

The diagnostic below starts with what changed, not what to try. It is built around the difference between a generic answer and a context-aware one, and that difference is worth showing before the checklist.

The generic answer versus the context-aware answer

Ask any AI tool why your Meta CAC spiked and you will get a reasonable list: creative fatigue, audience saturation, increased competition, landing page issues, iOS signal loss. All plausible. None of them tell you what actually happened in your account this week.

Now consider what a context-aware answer looks like when the account history is available:

• You are in week six of the current creative set. Frequency on your top three ad sets crossed 3.2 on Thursday. That is consistent with the fatigue pattern your account showed in August, when CAC rose 19% before you rotated creative.
• On Tuesday, daily budget was increased by 40%. That size of jump, in one move, reset the learning phase on two campaigns. Both are showing the elevated CPMs typical of re-entry into learning.
• In March, you tested broad audience targeting on a cold campaign. CAC rose 38% over three weeks before you reverted. The account notes from that test are still in the system.
• In August, UGC variants produced CAC that was approximately 22% lower than static creative across the same audience segments.

The context-aware answer does not just name possibilities. It names the ones that fit your account's actual history, ranks them by likelihood, and tells you what you already learned the last time each one happened.

That is the diagnostic you should be running.

Step-by-step diagnostic checklist

Work through these in order. Most CAC spikes resolve at step two or three. Reaching step five without a clear answer usually means the cause is external.

Step 1: Establish the timeline

Before touching anything, pin down when CAC started moving.

• Pull daily CAC for the past 21 days. Identify the first day the trend changed, not just the worst day.
• Note any events in the 48 to 72 hours before the inflection: budget changes, new creative launches, campaign edits, audience changes, landing page deploys.
• Check whether the spike is account-wide or isolated to specific campaigns or ad sets. A single campaign driving the average up is a different problem than a portfolio-wide shift.

The question to answer: Did something change in the account at roughly the same time CAC moved?

Step 2: Check for self-inflicted learning disruption

This is the most common cause of sudden CAC increases that teams overlook because the action felt routine.

• Any budget change above 20% in a single edit can reset the learning phase. A 40% jump almost certainly does. Check your change history for the past seven days.
• Campaign structure edits, audience swaps, and bid strategy changes all trigger re-entry into learning. Learning-phase CPMs typically run 15 to 35% higher than stable delivery, based on patterns commonly reported by practitioners.
• If two or more campaigns entered learning simultaneously, the account-level impact compounds.

What to do: If a budget increase caused the disruption, scale back to the pre-change level or to a level no more than 20% above the prior stable point. Let learning restabilize before scaling again. This is not a creative problem.

Step 3: Diagnose creative fatigue

Fatigue is real but it is also over-diagnosed. Check the signals before assuming it.

• Pull frequency by ad set for the past 14 days. Frequency above 3.0 on a cold audience within two weeks is a meaningful signal. Above 4.0 is almost always a problem.
• Look at CTR and hook rate trends on your top-spend creatives. A CTR decline of 20% or more week-over-week, with stable impressions, points to fatigue rather than audience or bid issues.
• Check how long the current creative set has been running. Six weeks at meaningful spend is a common fatigue threshold, though this varies by vertical and audience size.

What to do: If your account history shows which creative formats have recovered CAC after fatigue events, start there. If UGC variants have historically produced lower CAC than static creative in your account, that is where to go first, not a generic creative refresh.

Step 4: Review audience signals

Audience issues tend to show up more slowly than learning disruptions or fatigue, but they can accelerate CAC over a two to four week window.

• Check audience overlap between ad sets. Significant overlap means you are competing against yourself in the auction.
• Review CPM trends separately from conversion rate trends. Rising CPM with stable conversion rate means the auction got more expensive. Stable CPM with falling conversion rate means the audience quality or the offer fit changed.
• If broad targeting was recently introduced or expanded, check whether CPM and CAC moved in the same direction. In accounts where broad has been tested before, the historical result is your best prior.

What to do: Do not broaden audiences as a response to a CAC spike without checking whether your account has already run that test. If broad targeting raised CAC 38% the last time you tried it, that is not a hypothesis worth re-running without a specific reason to believe the conditions have changed.

Step 5: Check external factors

If steps one through four do not surface a clear cause, the problem may not be inside the account.

• Check auction competitiveness in your category. CPM benchmarks shift during retail events, competitor campaign launches, and seasonal demand periods. A CPM increase you did not cause is still your problem to manage.
• Review landing page performance. A conversion rate drop on the page with stable ad metrics means the issue is post-click. Check load time, form errors, and any recent page changes.
• Check whether any external business decision changed the offer, the price, or the product availability. A price increase that went live the same week as a CAC spike is not a coincidence.

What to do: External CPM pressure requires a different response than internal account issues. You cannot fix auction competition by refreshing creative. You can improve your creative efficiency so you win the auction at a lower effective cost, but that takes weeks, not days.

Step 6: Prioritize one change at a time

The most common way a diagnostic turns into a bigger problem is running multiple interventions simultaneously. If you change creative, revert the budget increase, and introduce a new audience in the same week, you will not know what worked.

• Identify the single most likely cause from steps one through five.
• Make one change. Give it five to seven days of stable delivery before evaluating.
• Document what you changed, why, and what you expected to happen. If the change works, that reasoning becomes part of your account's institutional knowledge. If it does not, you have ruled out a hypothesis rather than created noise.

What makes this diagnostic harder without account history

Every step above depends on knowing what your account has already tried. The question in step four, whether broad targeting raised or lowered CAC in your account, cannot be answered from the current week's data. It requires retained history.

Most teams hold that history in a combination of Slack threads, quarterly review decks, and the memory of whoever ran the account six months ago. When that person is on leave or has left the team, the history is effectively gone. The next CAC spike gets diagnosed from scratch.

This is the structural problem that generic AI tools do not solve. They can reason well about the 30% of performance marketing that lives in platform mechanics. They cannot reason about the 70% that lives in your account's specific test history, your team's past decisions, and the business context that shaped those decisions.

Where Almond fits

Almond's Business Context Engine is built specifically for this gap. It reads your Meta and Google ad accounts alongside your Slack, meeting notes, Notion docs, and CRM, and retains every test, result, and decision as it happens. When CAC spikes, the diagnostic does not start from a blank prompt. It starts from a system that already knows you ran broad targeting in March, that it raised CAC 38%, that UGC variants outperformed static in August, and that Tuesday's budget jump was 40% in a single edit.

The workflow does not replace the diagnostic above. It makes every step faster and more accurate because the account history is already there, versioned and ranked by relevance, rather than scattered across tools and people.

If you want to see how this applies to your specific account, early access is open at almond.ai. Every member gets a detailed 1:1 Meta and Google ad account audit with the CEO, recorded.