Owner, trigger and scope
The paid media manager owns this procedure and makes pause decisions within the agreed thresholds. The account lead owns the threshold document itself, approves any pause that falls outside it, and reviews the pause log each month. Splitting these roles stops thresholds drifting to suit whatever the manager wants to do that week.
There are two triggers. The routine one is the weekly optimisation review, where every active ad, keyword and ad set is compared with its target. The ad hoc one is an alert: an automated rule, a spend spike or a sudden drop in conversions. Alerts trigger this SOP, but they do not shortcut it.
The scope is pausing and un-pausing individual ads, keywords, ad groups and ad sets. It does not cover restructuring a campaign, changing bids or budgets, or replacing creative. Those decisions are part of the wider account management covered in the paid search account management playbook, and the logic of when to stop a test or scale a winner sits in the testing and scaling guide. When the trigger is a budget cut rather than routine pruning, the guide to cutting a Google Ads budget decides which spend to remove first and when trimming a campaign beats pausing it.
Before you start: inputs and access
| Input | Owner | Why it is needed |
|---|---|---|
| Threshold document | Account lead | States the target CPACost of each conversion. or ROASRevenue generated per unit of ad spend., minimum data and lookback window |
| Performance report for the lookback window | Paid media manager | The data the decision is made on |
| Conversion tracking status | Analytics owner | A pause on broken tracking is a pause on nothing |
| Learning status of each ad set | Platform | Avoids resetting learning without need |
| Pause log | Paid media manager | Records decisions and review dates |
You need edit rights on the account and read access to analytics. Keep the CPA calculator and the ROAS calculator open to check figures that the platform reports differently from your target definition. If there is no threshold document, stop and ask the account lead to write one; this SOP cannot run without it.
The procedure
Confirm the threshold
Open the threshold document and find the rule that applies to the item: the target cost or return, the minimum volume of spend, clicks or conversions before a judgement is allowed, and the lookback windowThe time after an ad interaction in which a conversion can still be credited to it.. Confirm the rule is current, since targets change with offers and seasons. If no rule fits the item, for example a brand campaign judged on reach rather than cost, escalate to the account lead.
Check data volume and tracking
Check the item has reached the minimum volume in the agreed lookback window. If it has not, the answer is keep, whatever the numbers look like. Then confirm conversion tracking was working throughout the window: look for flat lines, sudden drops across every campaign or a gap between platform and analytics figures. If tracking looks broken, stop and raise a tracking incidentA live error in marketing work, data or systems that reaches customers, spend or reports, such as a wrong price, a wrong segment or a broken purchase tag. instead of pausing anything.
Check learning status
On platforms that use a learning phaseThe period after launch or a significant edit when an ad set's delivery is still being optimised., check whether the ad set is still learning or has recently re-entered learning after an edit. The learning phase guide explains what the platform documents about this state. Results during learning are less stable, so unless the item is plainly broken or off-target, record it as keep and review after learning ends.
Compare against the target
Compare the item's result with its target over the full lookback window, not the last few days. Look at the supporting signals as context: a falling CTRShare of impressions that result in a click. or rising FrequencyAverage times a person saw an ad. may explain a worsening cost and suggest a creative refresh rather than a pause. If a whole campaign has dropped rather than one item, work through the performance drop troubleshooting guide first.
Decide: pause, keep or escalate
If the item has enough data, working tracking, has left learning and misses the target by more than the agreed tolerance, pause it. If it is within tolerance, keep it. If pausing it would leave an ad set or ad group with nothing running, or would remove a large share of the campaign's delivery, escalate to the account lead instead of pausing alone. Pause, never delete.
Log the decision and set a review date
Write one line per decision in the pause log: item, date, metric value against target, data volume, decision, reason and a review date. Keep decisions are worth logging too when they were close, because they explain later why something stayed live. On the review date, check whether the pause still makes sense or whether conditions such as season, offer or landing page have changed enough to retest.
A weekly review that pauses one ad and keeps another
Satchelbloom Tutors, an invented tutoring service, has a target cost per lead of £40 and a rule that no ad is judged before £200 of spend over a 14-day window. In the weekly review, ad A has spent £260 at £71 per lead with tracking healthy and learning complete; it is paused and logged with a review date in 30 days. Ad B shows £95 per lead but has spent only £120 and its ad set re-entered learning 3 days ago after a budget change. It is kept, logged as "insufficient data", and re-checked the following week, when its cost per lead has settled at £38.
QA checks
The spot-check is not a second review of every item. It samples the decisions most likely to be wrong: the closest calls and anything paused in response to an alert.
Before the review is closed, the account lead or a second manager spot-checks the decisions:
- Every pause matches a rule in the current threshold document, or carries the account lead's approval.
- Data volume met the minimum for every paused item.
- Tracking was confirmed healthy for the lookback window.
- No ad set was paused while still in learning without a recorded reason.
- No ad group or ad set is left with nothing running unintentionally.
- Nothing was deleted instead of paused.
- Every decision has a log entry with a reason and a review date.
Hand-offs and records
Pauses that leave a gap, such as an ad set with only one remaining ad, are handed to the creative lead as a request for replacement assets, with the log entry attached so they can see what failed. Suspected tracking problems are handed to the analytics owner through the tracking incident process, and the review is re-run once tracking is confirmed. Decisions outside the threshold go to the account lead for approval before they are applied.
The pause log is the main record. It should live in a shared place, sorted by review date, so the next weekly review starts by checking which earlier decisions are due. The monthly reporting pack draws on it to explain changes in spend distribution. The account lead reviews the log each month for patterns, such as the same kind of ad failing repeatedly, which point to a strategy issue rather than an ad issue.
Common mistakes
- Pausing on a bad day. Short windows are noisy. Use the agreed lookback and minimum volume.
- Pausing on broken tracking. A tracking fault looks like poor performance everywhere at once. Check tracking first.
- Resetting learning repeatedly. Frequent edits keep ad sets unstable. Wait until learning ends unless something is clearly wrong.
- Moving the goalposts. Changing the threshold to justify a decision defeats the SOP. Only the account lead changes thresholds.
- Deleting instead of pausing. History disappears and mistakes cannot be reversed cleanly.
- No review date. Paused items are forgotten, including ones that would work under new conditions.
- Treating a campaign-wide drop as an ad problem. Investigate the campaign before pausing its parts.
