A full year of spend that came back as itself
Before anything could be improved it had to be measured honestly. Across the preceding full year the account spent 41,368 dh and produced 42,126 dh in revenue — a blended return of 1.02 and a return on investment of 1.83%.
At 1.02 the advertising is not a growth channel. It is a break-even treadmill: money goes in, roughly the same money comes out, and none of it funds inventory, salaries or expansion. Worse, it hides the problem, because revenue is technically being generated so it looks like something is working.
- The allocation was backwardsPaid social absorbed 72% of total spend and returned 0.76 — every 100 dh sent to it came back as 76 dh. Meanwhile search was the better platform at 1.67 and was being under-funded.
- Nobody owned the account day to dayWithin search, a single campaign spent 981 dh to produce one conversion, a 0.16 return. That is what an unmonitored campaign looks like when responsibility is split.
The starting year, exactly as the data reported it
| Channel | Spend | Purchases | Revenue | CPA | ROAS |
|---|---|---|---|---|---|
| Search | 11,722 dh | 105 | 19,617 dh | 111.64 dh | 1.67 |
| Paid social | 29,646 dh | 193 | 22,509 dh | 153.61 dh | 0.76 |
| Combined | 41,368 dh | 298 | 42,126 dh | 138.82 dh | 1.02 |
Both the ads and merchant accounts were suspended
This was the hardest part of the engagement and it had nothing to do with advertising skill. Both accounts were suspended under a circumventing-systems policy violation — one of the harder violations to recover from, because the platform treats it as deliberate rather than accidental. Shopping and Performance Max both depend on the merchant feed, so for a retailer whose catalogue is the product, the entire acquisition engine went down at once.
Rather than treating it as a support ticket, we audited the full account estate. The client had worked with several agencies over time and each one, on arrival, had created fresh ads and merchant accounts for the same business instead of requesting access to the existing ones. Several live accounts promoting one domain looks exactly like an advertiser rebuilding to escape prior enforcement. The system was working correctly. The account structure was not.
- We consolidated before we appealedWe inventoried every account tied to the business, identified which carried the legitimate history, made it the master and closed every duplicate. Most reinstatement attempts fail because they appeal first and clean up later — an appeal filed while duplicates are still live argues against a violation that is still true.
- Then we fixed the sourceThe merchant account was consolidated the same way, and we brought the website into compliance so product data and policy pages met requirements. Feeds and landing pages are re-reviewed continuously, so a site that scrapes past the first check gets caught on a later one.
- Both accounts were reinstated — and have stayed reinstatedOne consolidated master account, a compliant site behind it, and no repeat suspension since.
This work generated no impressions, no clicks and no revenue while it was happening. It was also, measurably, the single highest-return work of the entire engagement — because every result that followed is downstream of it.
We rebuilt rather than resuming
When access was restored we deliberately did not switch the old campaigns back on. The previous setup had produced a 1.02 return; restarting it would only have reproduced 1.02 more reliably.
Search spend was consolidated into a single, properly fed Performance Max campaign instead of being split across fragmented search and shopping campaigns. Splitting a modest budget starves each campaign of the conversion signal it needs to learn — one well-fed campaign learns, four hungry ones do not. That single campaign went on to deliver 636 conversions and 147,489 dh of revenue at a 3.28 return.
Paid social was restructured away from broad interest-theme targeting and onto catalogue-driven campaigns with dynamic product retargeting. For a retailer with a large SKU count the catalogue itself is the targeting signal — showing a specific bottle to someone who has already looked at that bottle beats guessing at interest categories.
Before and after, side by side
| Metric | Prior full year | Six months after rebuild | Movement |
|---|---|---|---|
| Ad spend | 41,368 dh | 50,099 dh | +21% |
| Revenue | 42,126 dh | 176,063 dh | 4.2x |
| Purchases | 298 | 758 | 2.5x |
| Cost per purchase | 138.82 dh | 66.09 dh | −52% |
| Blended ROAS | 1.02 | 3.51 | +244% |
Four times the revenue of the entire preceding year, in half the time, on 21% more spend.
Performance earned the budget, not the other way round
As results became visible, other agencies began approaching the client with pitches and audit reports of his own account. Rather than evaluating them privately or using them as leverage, he forwards them to us and asks whether their findings are accurate.
A competitor’s audit is designed to make the incumbent look negligent. Handing those documents to the incumbent and asking for an honest assessment is something a client does only when he is confident the answer will be straight even where it is unflattering.
On the strength of that performance the client approved close to a doubling of daily search investment — an increase of about 88%. Budget is the clearest expression of confidence a client has available, and it was earned by the efficiency work rather than requested ahead of it.
I don’t want to go with any other agency. All your data-driven strategies are working for me.
The bottleneck moved off the ad account
The account is now in a position most performance engagements never reach: the campaigns are no longer the constraint. Click-through rate has risen and cost per purchase has halved, but on-site conversion rate still sits below 1%, and the largest remaining loss of revenue now happens after the click.
So we have recommended and begun a redesign of the website, focused specifically on the stages where users drop out of the purchase journey rather than on a general visual refresh. It is being decided the same way every other decision in this engagement has been — from behaviour, not preference — and it will be measured against on-site conversion rate and blended ROAS, not against whether it looks more modern.
- The honest caveatWe are stating an expectation, not a guarantee. Scaling spend usually costs some efficiency as campaigns move beyond the highest-intent audiences, and a redesign takes time to prove out. The correct measure of this next phase is whether blended return holds at a materially higher spend level — not whether it sets a new single-month record.