Key takeaways
- Start every weekly review with a complete changes log before looking at any metric. Changes cause outcomes.
- Record the reasoning behind every decision taken in the review. Three sentences per call, written at the time.
- Flag anomalies against a consistent variance threshold and check the changes log for an explanation before investigating further.
- Creative fatigue and budget pacing drift are predictable. A weekly checklist catches both before they compound.
- The biggest time cost in most reviews is reconstructing context that was never captured. A memory layer removes that step.
Most weekly performance marketing reviews are backward-looking reports. They show what happened. This checklist is different: it starts with what changed, because changes cause outcomes. Once you know what changed, the numbers explain themselves.
Why most weekly reviews fail
The typical Monday review opens in the platform, sorts by spend or ROAS, and flags whatever looks different from last week. That approach misses the most important question: why is it different?
A campaign's CPA jumped 40 percent. Is that a signal or noise? You cannot answer that without knowing whether a budget cap was raised on Thursday, a creative set was paused, a competitor launched a promotion, or a landing page was quietly swapped. The number is the symptom. The change is the cause.
The second failure is that decisions made in the review rarely get recorded anywhere. The team agrees to pause a test, shift budget to a different audience, or hold creative rotation steady for another week. By the following Monday, two of those three decisions have been forgotten, and the reasoning behind all of them is gone entirely. When the same situation recurs in six weeks, the team debates it from scratch.
A good weekly review fixes both problems: it starts from changes, and it ends with a written decision log.
The seven-section checklist
1. Changes log (first, always)
Before you look at any metric, reconstruct everything that changed in the account over the past seven days. This includes:
- Budget adjustments, including dayparting and campaign-level caps
- Bid strategy changes or target adjustments (target CPA, target ROAS, manual bid overrides)
- Creative additions, pauses, or rotations
- Audience edits, exclusions added or removed
- Landing page or offer changes, even if they came from outside the marketing team
- Platform-side changes: algorithm updates, auction shifts, policy flags
- Business changes: pricing updates, inventory constraints, sales team feedback from the prior week
This log does not need to be long. It needs to be complete. Five minutes spent here saves thirty minutes of misdiagnosis later.
2. Anomaly scan
With the changes log in front of you, scan for metrics that moved more than your normal variance threshold. A reasonable starting point for most accounts is flagging any week-over-week movement above 20 percent in spend, CPA, ROAS, CTR, or conversion rate at the campaign level.
For each anomaly, the first question is: does the changes log explain it? If yes, note the explanation and move on. If no, it becomes an open investigation item.
Common unexplained anomalies worth deeper investigation:
- CTR drops with no creative change (audience fatigue or auction competition)
- Conversion rate drops with stable CTR (landing page, offer, or checkout issue)
- Spend underpacing with no budget reduction (bid competitiveness or quality score degradation)
- CPA spikes on previously stable campaigns (creative wear-out, audience overlap, or seasonality)
3. Open tests and experiment status
List every active A/B test, holdout, or experiment running in the account. For each one, record:
- What is being tested and the hypothesis
- Current sample size against the target
- Whether the test has reached statistical significance
- Scheduled end date
The most common mistake in this section is letting tests run past their end date because nobody closed the loop. The second most common mistake is pausing a test early because a metric moved in one direction during a short window. Both waste the learning.
If a test concluded last week, record the result here, including the reasoning for what you are doing with the winner. A test result without a decision is just a number.
4. Budget pacing
Review pacing against monthly and quarterly targets for every campaign and channel. Flag:
- Campaigns more than 10 percent ahead of pace (risk of exhausting budget before month-end)
- Campaigns more than 10 percent behind pace (lost impression share or a cap that was set too conservatively)
- Channels where pacing diverges from the agreed allocation
Note any business events in the coming week that should affect pacing: product launches, promotions, sales cycles, or external events that historically move conversion volume.
5. Creative age and rotation health
Creative fatigue is one of the most predictable sources of performance decay, and it is also one of the most commonly overlooked in weekly reviews.
For each ad set or ad group, note:
- The age of the current top-spending creative
- Frequency (for Meta) or impression share concentration (for Google)
- Whether any creative has been running long enough to warrant a refresh based on your account's historical fatigue curve
A rough benchmark for many direct response accounts: Meta creatives at frequency above 3.5 to 4.0 in a seven-day window are worth watching. Google responsive search ads with a single dominant combination serving more than 60 percent of impressions often benefit from asset diversification. Verify these against your own account's patterns, because the numbers vary significantly by category and audience size.
6. Decisions taken and reasoning
This is the section most teams skip, and it is the one that compounds the most value over time.
For every action taken or deferred in the review, write one to three sentences: what you decided, why, and what you expect to happen. Examples:
- Paused creative set B in Campaign X. CTR held but conversion rate dropped 35 percent over 14 days with no landing page change. Likely audience-level fatigue. Replacing with variant C from last month's test.
- Held budget allocation steady despite ROAS softness in the brand campaign. Leadership confirmed a competitor price promotion this week. Expect recovery once it ends.
- Deferred audience expansion test. Current test still two weeks from target sample size. Expanding now would contaminate the result.
Three sentences per decision. Written at the time. This is the institutional memory that AI tools, new hires, and future-you all need.
7. One-week forward view
Close the review with a short list of what to watch in the coming seven days:
- Tests expected to conclude
- Budget milestones to monitor
- External events that could affect performance
- Any decision deferred this week that needs a trigger to revisit
This section takes five minutes and prevents the following Monday's review from starting cold.
How long this should take
For a single-channel account with one to three campaigns, this review should take 30 to 45 minutes. For a multi-channel account across Google and Meta with active creative testing, budget 60 to 90 minutes. If it is taking longer, the bottleneck is usually the changes log: context is scattered across Slack, email, and three people's memories, and reconstructing it takes longer than the analysis itself.
That reconstruction cost is worth naming. In most teams, a meaningful share of the weekly review time is not analysis. It is archaeology.
Making the review write itself
The checklist above describes what a good review contains. The harder problem is where the inputs come from.
The changes log requires someone to have tracked every budget adjustment, creative swap, and business decision as it happened during the week. In most teams, that tracking is partial at best. Budget changes live in the platform. Creative decisions live in Slack. Business context from Monday's leadership call lives in a meeting recording that nobody has time to transcribe.
This is the specific problem an agentic memory layer addresses. Almond (almond.ai) connects to ad accounts, Slack, Notion, email, meeting recordings, and CRM in read-only mode, captures decisions and context as they happen, keeps them current, and surfaces the relevant subset when you need it. The changes log that takes 20 minutes to reconstruct manually becomes something the system already holds.
More practically: the decisions section of the review, which most teams skip because it feels like extra work, becomes easier to maintain when the system is already capturing the context around each decision. The reasoning does not have to be typed from scratch. It has to be confirmed.
This is not about automating the review. The judgment in sections two through six still requires a practitioner who understands the account. It is about removing the archaeology so that time goes to the analysis.
Where Almond fits
Almond's Business Context Engine is built specifically for the problem this checklist surfaces: performance marketing context is created continuously, captured inconsistently, and lost permanently when it is not recorded. The weekly review is the moment when that gap is most visible.
The engine reads your ad accounts, Slack, meetings, and business data, classifies everything by relevance and recency, and keeps it current without anyone re-entering it. Canvas surfaces it as a live dashboard. Workflow Builder runs 30 prebuilt Google and Meta automations that act on context rather than triggers, so an automation does not scale a campaign your team decided to sunset on Monday.
Almond 2.0 early access is open at almond.ai. Every member gets a detailed 1:1 Meta and Google ad account audit with the CEO, recorded. Waitlist is work email only.
Frequently asked questions
- How often should a performance marketing account be reviewed?
- Weekly is the right cadence for most paid media accounts. Daily check-ins make sense for large-budget campaigns or active promotions, but a structured review that covers changes, anomalies, tests, pacing, and creative health belongs on a weekly schedule. Monthly reviews are too infrequent to catch creative fatigue or budget drift before they cost real money.
- What should be included in a weekly PPC review checklist?
- A complete weekly PPC review covers seven areas: a changes log from the prior week, an anomaly scan against normal variance thresholds, open test status, budget pacing against monthly targets, creative age and rotation health, a written log of decisions taken and the reasoning behind them, and a short forward view of what to watch in the coming week.
- How do you track decisions made in a paid media review?
- Write one to three sentences per decision at the time it is made: what was decided, why, and what outcome is expected. Store this in a shared doc, Notion page, or a system that stays connected to the account. The goal is that anyone on the team, including a new hire or an AI tool, can read the log and understand the reasoning without asking the person who made the call.
- What is a good variance threshold for flagging anomalies in a PPC account?
- A common starting point is flagging any week-over-week movement above 20 percent in spend, CPA, ROAS, CTR, or conversion rate at the campaign level. The right threshold varies by account size and historical volatility. Smaller accounts with lower volume will show wider natural variance, so a 30 to 35 percent threshold may reduce false positives.
- How do you know when a paid media creative needs to be refreshed?
- Watch for frequency above 3.5 to 4.0 in a seven-day window on Meta, or conversion rate decline with stable CTR on any platform. Creative age alone is not a reliable signal. A creative running for six weeks with stable conversion rate does not need replacing. One running for three weeks with declining conversion rate does. Track both age and performance together.
Almond 2.0 early access
Get a 1:1 Meta and Google audit with the CEO.
Every waitlist member gets a 1:1 audit of their ad accounts. Work email only.