Sample deliverable
Google Ads change log: a month of account work, one row per change
This is what account management looks like when it is written down. Each row records what changed, why it changed, and how the change will be checked, so the monthly report is a log of decisions rather than a screenshot of a dashboard.
Example, not a client's data
Invented setup: a service business running a brand campaign, a generic search campaign and a remarketing campaign in one market. Month one of the retainer. Conversion tracking existed before we arrived but had never been compared with the orders the sales desk actually logs. No real account and no real client.
How the log is kept
- One change, one row. A batch of edits made in a single sitting is still listed as the separate decisions it was.
- The reason is written at the time, not reconstructed at the end of the month. It is what makes a change reviewable later.
- The check is agreed in advance: which report, which window, and what would count as the change having worked.
- A deliberate decision not to change something is a row too, because the account should show why an obvious lever was left alone.
The month, change by change
| When | Campaign | What changed | Why | How we check it |
|---|---|---|---|---|
| Week 1, Day 1 | Account | Manager access accepted, conversion actions listed and defined | No optimization means anything while the measurement is unverified | Each action has recent events and a written definition the client agrees with |
| Week 1, Day 2 | Search, generic | Negative keywords added for job, free and do it yourself queries | The search term report showed spend on visitors who cannot buy | The same report a week later, read on the same window |
| Week 1, Day 3 | Tracking | Enhanced conversions and consent handling reviewed with the developer | Consent gated tags under report quietly, and the account then looks worse than it is | Test conversions fire and appear in the interface |
| Week 1, Day 4 | Search, brand | Brand terms split into their own campaign | Brand and generic demand behave differently and were sharing one budget | Cost per conversion read separately for each campaign from now on |
| Week 2, Day 1 | Search, generic | Responsive ad headlines rewritten to lead with the service and the city | The ads repeated the brand while the query asked for the service | Asset performance labels and click through rate inside the same ad group |
| Week 2, Day 3 | Display remarketing | Mobile application placements excluded | The placement report showed clicks from apps with no matching behavior on the site | The placement report over the following weeks |
| Week 2, Day 5 | Account | Bid strategy deliberately left unchanged while the campaign was still learning | Changing a strategy mid learning restarts it and delays every other read | Conversion volume watched until the learning label clears |
| Week 3, Day 1 | Search, generic | Budget moved from an ad group with no conversions to one with steady demand | Budget was spread evenly across demand that is not even | Conversion volume per ad group against the previous window |
| Week 3, Day 2 | Landing page | Asked the client to place a phone number that can be tracked | Phone orders were invisible, so the platform looked wrong when it was not | Calls appear as a conversion action and match the sales desk log |
| Week 3, Day 4 | Performance Max | Brand exclusion applied | Without it the campaign takes credit for demand the brand already had | Brand and generic conversions read separately after the change |
| Week 4, Day 2 | Account | Sales desk log reconciled against platform conversions | The two counts disagree by default, and the reason has to be named rather than averaged | The reconciliation below, repeated the same way every month |
| Week 4, Day 4 | Reporting | Monthly note written: what changed, what it means, what comes next | A report that only shows charts leaves the client to guess the decisions | Reviewed together on the monthly call |
Orders against platform conversions
The two counts almost never agree, and the useful work is naming the reason for each gap instead of picking the number that flatters the account. Counts are left out of this sample on purpose: what matters here is the shape of the check.
| Source | What the sales desk counts | What the platform counts | Direction of the gap | Likely reason |
|---|---|---|---|---|
| Search, brand | Orders taken by phone and by form | Conversions attributed to the campaign | The platform counts more | The same person submits the form and then calls, and both actions are recorded |
| Search, generic | Orders logged with the source noted by the operator | Conversions attributed to the campaign | The platform counts fewer | Calls to a number that is not tracked dynamically never reach the account |
| Display remarketing | Orders from returning customers | Conversions including view through | The platform counts more | View through conversions sit in the default column and credit a channel that assisted rather than closed |
| Performance Max | Orders that reached the sales desk | Conversions attributed to the campaign | The platform counts more | Automated form submissions are counted as conversions, and the algorithm is then optimized toward them |
What the manager does every week
- Reads the search terms and adds negatives, because a query list drifts on its own even when nothing was changed.
- Checks pacing against the agreed budget, so the month does not end with a week paused.
- Reviews invalid traffic and placement reports and excludes what does not belong.
- Looks at the assets carrying the ads and replaces the ones that are never served.
- Compares conversion actions with what the sales desk sees, so a tracking break is caught in the week it happens.
- Writes the rows for this log while the reason is still fresh.
What we never change without asking
- A meaningful shift in the ad budget. Billing stays in your account, and volume is a business decision, not an optimization.
- The bid strategy on a campaign that is still learning, unless something is measurably broken.
- Pausing a campaign mid learning because a single week looked weak.
- The landing page. We say what is missing and why, and the change belongs to whoever owns the site.
Synthetic example, not a client's data
- No figures appear anywhere: not spend, not conversions, not cost per conversion. A real log carries them, and the reconciliation is where they matter most.
- The month shown is invented and unusually tidy. Real months include a change that was reverted the following week, and the log records that too.
- Campaign names here are generic. On a real account the log uses your naming, so a row can be found in the interface without translation.
- The reasons in the reconciliation are the common ones. Yours are found by reading your account against your sales desk, not by picking from this list.
A real change log is delivered with the monthly report and the figures behind it, so every decision can be traced back to the report that prompted it. Packages and what each one covers are on the service page.
Google Ads SetupWant it handled every month?
Tell us what you run and what it needs. We reply with a plan and a monthly price; month to month, no lock-in.
Discuss a retainerNo commitment. Response within one business day.