Bathroom design and fitting for homeowners across Hertfordshire and Bedfordshire, with a showroom in Hitchin. 32 staff.
Verdict: 6 ads. Spend: £9,300 last month, £10,000 next; conversions: 68 last month, 80 expected next. A conversion matched to your records is worth £315 of gross profit once enquiries that don’t buy, and customers who’d have come anyway, are taken out (less where there’s a discount). The ads net £7,130 a month now and should net £9,232 next month: £26,806 over a year (£12,348 from refreshing 1 tired ad, £8,683 from pausing, £6,092 from moving budget, -£318 from testing new ads). After £4,240 of new creative in the year (refreshing again every 3 months), the first year nets £22,566; £17,585 a year if every change does half as well. The best ad now is “Search: bathroom fitters near me”.
| Ad | Spend last month | Cost per conv. | Net a month | Next month | Net next | Change a month |
|---|---|---|---|---|---|---|
| Before-and-after carousel Meta · click rate 0.6% (was 1.4%) · frequency 6.8 · platform figures · tired | £2,400 14 conv. | £171 | £687 | Refresh £2,400 next | £1,716 | £1,029 |
| Remarketing banner Google Display · click rate 0.15% (was 0.4%) · frequency 9.5 · platform figures · 20% truly extra · tired | £900 2 conv. | £450 | −£724 | Pause £0 next | £0 | £724 |
| Showroom video tour Meta · click rate 1.1% (was 1.2%) · frequency 2.1 · platform figures | £1,800 16 conv. | £113 | £1,728 | Keep £2,800 next | £2,236 | £508 |
| Search: bathroom fitters near me Google Search · click rate 6.5% (was 6.8%) · checked | £3,000 30 conv. | £100 | £6,450 | Keep £3,000 next | £6,450 | £0 |
| 15% off spring offer Meta · click rate 0.9% (was 1%) · frequency 4 · platform figures · 15% discount | £1,200 6 conv. | £200 | −£1,011 | Keep £1,200 next | −£1,011 | £0 |
| Customer story video (test) Meta | – 0 conv. | – | – | New £600 next | −£159 | −£159 |
The carousel’s click rate has more than halved and people have seen it nearly seven times: a refresh is counted at half the click rate it wins back, because it tires again within three months. The remarketing banner mostly reaches people who were coming anyway (20% truly extra) and is paused. The 15% offer gives £1,350 off every job, including to customers who’d have come anyway: it loses about £1,000 a month and should be paused too, though the owner kept it. The search ad’s conversions are matched to Oakfold’s enquiry records; the Meta ones aren’t yet, so they count at 70%.
Spend, click rates, frequency and conversions come from the ad platforms for last month; 20% of enquiries become customers, and only 50% of those are treated as ones who wouldn’t have come anyway. Platform conversions not checked against the business’s records count at 70%. A refresh wins back 50% of a tired ad’s lost click rate, lowering its cost per conversion in proportion (half that on average as it tires again), and is repeated every 3 months; added budget costs 30% more per conversion; new ads start at their platform’s average, 50% counted, and run 2 months before they’re judged. Discounts come off every order. Remarketing’s help to other ads isn’t valued, nor the weeks a job takes to close. One month is a short period: results swing.
Re-run monthly with three months of figures, and match platform conversions to the enquiry log so the 70% can be replaced by the real figure.
Oakfold chose the guided route. At the last session the agent was set up in an AI assistant the business already uses, connected read-only to its Meta and Google ad accounts and its enquiry log (its IT confirmed this before it committed). On the first Monday of each month it loads last month's results into a new dated copy of this review, runs the checks and sends the owner a short note. It uses the rules on Inputs: an ad is tired when its click rate has fallen 30% from its first two weeks and it is seen 4 times or more, a test is judged after 2 months, and a tired ad is flagged only if it spent £250 or more. Nothing left the business, and no decision changed, without one of Oakfold's own people approving it. These events continue the fictional example.
| Date | What the agent found | What people decided |
|---|---|---|
| 2 Nov 2026 Note only | October's results: the refreshed carousel's click rate rose from 0.6% to 1.0%, as the review expected at its peak. No ad newly tired. | Nothing needed. The owner read the monthly note. |
| 7 Dec 2026 Flag | The showroom video tour, with its budget raised to £2,800, was newly tired: click rate 0.75% against 1.2% in its first two weeks (down 38%), seen 4.4 times. | The owner briefed the designer for a refreshed tour of the new winter displays, live from 4 January, and held the budget at £2,800 until then. |
| 7 Dec 2026 Flag | Data problem on the same run: Google's new report layout had renamed the conversions column, so the search ad's row failed its check. The note named the check and gave no figures. | The marketing assistant re-exported the report. The rerun passed the checks the same morning and the monthly note followed. |
| 4 Jan 2027 Flag | The customer story video had run its two-month test: £1,450 spent, 11 conversions, 8 matched to the enquiry log, about £132 a conversion against the £150 Meta average the review assumed. A decision was due. | The owner kept it from January at £600 a month, reviewed with the other ads from now on. |
| 1 Feb 2027 Note only | January's results: the refreshed showroom tour back to a 1.1% click rate. The carousel, refreshed in October, was down 20% from its new first two weeks: under the 30% line, so listed, not flagged. | Nothing needed. The owner pencilled in the carousel's next refresh for March. |
At the check-in on 30 November, after the first monthly run (about day 45), Geoff and the owner read that run's log together and left the rules as they were. After that the agent, its accounts and its log were Oakfold's.
What the agent added. The method is unchanged: the same spreadsheet, the same checks and the owner's own rules. The agent made sure the review was rerun every month, and caught the showroom tour on the first Monday after it crossed the line. The renamed column was stopped before any wrong figure reached the owner. The monthly review dropped from the 2–4 hours of gathering and entering to reading the note and deciding (estimate: about 30 minutes a month). The months after add nothing to the figures above.
Fictional and labelled: these events continue the worked example. The same events are the sample rows in the Agent log of the worked-example spreadsheet. The agent is part of the guided and done-for-you routes; the self-implement kit does not include it.