Most media buying teams have a routine. Few have written it down, and the ones that have not tend to discover the gaps the expensive way: the thing nobody was responsible for checking is the thing that was broken for two weeks.
This is a working checklist, split by cadence. The split matters more than the items. Something checked daily and something checked monthly are answering different questions — daily catches breakage, weekly catches drift, monthly catches structure — and putting an item on the wrong cadence is how it gets either ignored or over-monitored into noise.
The principle behind the cadence
Match the check frequency to how fast the underlying thing can hurt you.
- Daily is for anything that can lose money within hours and that a person can fix the same day. Breakage.
- Weekly is for anything that needs several days of data before the signal is trustworthy. Drift.
- Monthly is for anything about structure, cost base, access, and decisions that should not be revisited constantly. Structure.
A corollary worth stating: if an item appears on the daily list and nobody has ever found anything wrong with it, either move it to weekly or automate it. A daily checklist that never surfaces a problem stops being read within a month.
Daily
The daily pass should take one focused block, not an hour of clicking. If it takes longer, the checks are not aggregated enough.
Account health
- Zero-spend on live accounts. Any account marked active, with an enabled campaign and available budget, that spent nothing in the last several hours. Treat this as the first item, every day, before anything else.
- Suspensions and restrictions. New account-level suspensions, payment holds, or verification requests.
- Billing status. Cards close to expiry, failed charges, prepay balances that will not survive the next 48 hours. Payment failures are the single most preventable cause of downtime and they are almost always visible a day in advance.
Spend and pace
- Pace against budget per account. Both directions. Accounts significantly under budget on a profitable offer, and accounts running ahead of plan.
- Day-over-day spend deltas at the account level. Not portfolio total — the total will look normal while individual accounts swing.
- Any account that crossed a spend ceiling you set as a risk limit.
Performance
- CPA versus the account's own trailing baseline. Compare each account to its own recent median, not to a shared team target. Only act on sustained moves with enough conversion volume to be meaningful.
- ROAS on the top spenders. Your largest accounts deserve individual attention daily; the tail does not.
- Conversion tracking sanity. Conversions still arriving, postbacks still landing. A tracking break masquerades as a performance collapse and gets treated as one, which wastes a day.
Creative and approvals
- New disapprovals since yesterday, grouped by reason. Repeated reasons mean a systematic issue with a template or landing page, not a run of bad luck.
- Ads still pending review past the normal window.
- Limited (not disapproved) ads. These are the easy ones to miss because the account keeps spending.
Infrastructure
- Landing domain checks. HTTP status, redirect chains, page load, form or click-through working. A domain that starts failing damages every account pointed at it at once.
- Proxy and access health, if your setup depends on it.
The one habit that matters most
End the daily pass by writing down what you changed. Not a formal report — a line per action. Without it, the weekly review has no way to connect a performance change to a cause, and you will spend Monday arguing about what happened on Thursday.
Weekly
Weekly is where you look at trends rather than events. Give it a fixed slot and a fixed agenda.
Performance review
- ROAS and CPA by buyer. Direction over the last four weeks, not just the last week's number.
- ROAS by offer and vertical. Offer decay is systematic — when a payout changes or an angle burns out, every account running it moves together.
- ROAS by geo. Auction pressure and payout shift by market.
- Performance by landing domain. The classic hidden correlated failure. A degraded domain spreads its damage across buyers so no individual buyer sees enough to raise a flag.
- Winners and losers list. Explicitly name the accounts to scale and the accounts to cut. A review that ends without a list of decisions was a status meeting.
Budget reallocation
- Shift budget toward accounts and offers that held up over the full week.
- Cut or pause the persistent bottom of the distribution. The most common failure here is inertia: accounts that should have been cut two weeks ago keep running because nobody made the call explicitly.
- Check that scaling accounts have enough budget headroom and billing headroom to keep scaling.
Creative pipeline
- Creative age. How long have your top performers been running? Fatigue shows in CTR before it shows in ROAS.
- Test throughput. How many new creatives went live this week, and did any of them get a fair volume of data?
- Disapproval rate by template. If a specific creative pattern draws repeated rejections, fix the pattern rather than resubmitting.
Team and ownership
- Every account has a named owner. Unowned accounts are where problems live longest.
- Alert load per person. If a buyer received sixty alerts this week, they are reading none of them. High alert volume is a threshold-tuning problem, not a discipline problem.
- Blocked work. Anything a buyer needs from finance, from an admin, or from a platform review.
Monthly
Monthly is about structure and cost base. These items rarely produce urgent findings, which is exactly why they need a fixed date rather than good intentions.
Financial reconciliation
- Platform spend versus internal records. They will disagree. Find out why, every month, while the discrepancy is still small enough to explain.
- Revenue reconciliation against postbacks. Confirm reported revenue matches what actually settled.
- Full P&L including the costs that are not media spend. Tooling, proxies, domains, hosting, contractor fees, payment processing. Teams that only track spend and revenue systematically overstate margin.
- Payout reconciliation per buyer, if compensation is performance-linked.
Access and security review
- Who has access to what. Every Google Ads account, every internal tool.
- Remove access for anyone who left. This slips more often than any other item on this list.
- Check for shared logins. They destroy attribution, make incident review impossible, and turn one compromised credential into a full-portfolio problem.
- Confirm role assignments still match reality. People change teams; permissions rarely follow.
Infrastructure and portfolio hygiene
- Domain inventory. Which domains are live, which are expiring, which are flagged, which are pointed at nothing.
- Account inventory. Dormant accounts, accounts kept "just in case", accounts with residual balances.
- Alert rule review. Which alerts fired this month, which ones led to an action, which ones were ignored every single time. Delete or retune the ignored ones. An alert nobody acts on is worse than no alert, because it teaches the team that alerts are ignorable.
Strategy
- Vertical and offer mix. Concentration risk: what share of profit comes from a single offer, a single geo, or a single account?
- Capacity planning. Accounts per buyer, and whether the number is still workable.
- Documentation of what you learned. The angles that worked, the policies that caused rejections, the geo-specific behaviours. Most teams re-learn the same lessons annually because nobody wrote them down.
Making the checklist survive contact with reality
Three failure modes kill operational checklists, and all three are avoidable.
Length. A 40-item daily list gets skimmed and then abandoned. Anything that can be automated should be automated out of the daily list entirely — zero-spend detection, billing warnings, disapproval counts and pace anomalies are all mechanical checks. Reserve human daily attention for judgement calls.
No owner. Every item needs a named person, not a role in the abstract. "The team checks domains daily" means nobody checks domains.
No trail. If the daily pass leaves no record, you cannot audit whether it happened, and you cannot reconstruct causes later. A line per day is enough.
The direction of travel for any mature team is the same: push mechanical checks into automated alerting, and reserve the human cadence for the decisions — what to scale, what to cut, what to test next. That is the split AdsTracker is designed around, with account health, spend, approvals, domains and finance in one view, and alerts to email or Telegram handling the checks that should never have needed a human in the first place.