The first 90 days of a new e-commerce ad account follow the same order every time: fix measurement, then structure, then creative, then scale. Month one usually costs money because tracking changes what gets reported and restructuring resets learning. Month two is where structure starts paying. Month three is where scaling decisions get made against a known break-even.
The short version
- Week 1 is tracking. Not campaigns. Nothing else is worth doing on bad data.
- We do not pause inherited campaigns on day one. We watch first.
- Month one usually dips. We say that before we start, not after.
- By day 90 you should know your break-even, your best products and your ceiling.
Most agencies will not publish their onboarding plan, because it is the thing clients are actually buying and it is easy to copy.
We will, for two reasons. First, the plan is not the hard part. Doing it in the right order, on someone else’s money, while they are nervous, is the hard part. Second, if you read this and decide you can run it yourself, you probably should.
Days 1 to 14: make the numbers trustworthy
We do not touch campaign structure in the first fortnight. We fix measurement.
Conversion tracking end to end. Conversions API for Meta, enhanced conversions for Google, purchase value and currency passing in AED, and event IDs matching so the same purchase is not counted twice. Broken deduplication is the single most common fault we find, and it inflates reported revenue.
Reconcile against the store. We compare 30 days of platform-reported sales against store-reported sales and write down the gap. Under 30 percent is normal. Above 50 percent means something is broken. The full explanation is in why Shopify and Ads Manager never match.
Write down what a conversion means. Sounds trivial. It is not. We have inherited accounts optimising towards “add to cart” and wondering why revenue was flat.
Offline sales, if there are any. If the business also sells in a shop or takes bookings, we get offline conversion tracking in early, because every optimisation decision after this depends on the platform seeing all the revenue.
We also let the existing campaigns run during this fortnight. Switching things off before you understand them is how you kill the one campaign that was carrying the account.
Days 15 to 30: structure
Now we rebuild.
- Brand Search as its own campaign, cheap and exact, so we can see and separate it rather than letting it flatter everything else.
- Shopping or feed-only Performance Max on the products that actually sell, with brand excluded.
- Meta simplified. Most inherited accounts have twelve ad sets fighting each other for the same audience. Consolidation nearly always improves delivery.
- Retargeting cut back. It is usually oversized, it burns out fast on a small UAE audience, and it makes reported ROAS look better than the business is doing.
- Feed cleanup. Titles, images, GTINs, availability. Boring, and it caps everything above it.
This is the month where reported numbers often get worse before they get better. Learning resets. We say so up front, in writing, because a client who is surprised by a dip in week three will pull the plug in week four.
Days 31 to 60: creative and offer
By now the account is measurable and structured, so the constraint moves to what the ads actually say.
We test the offer before the creative. “Free delivery over AED 200” against “15 percent off your first order” against a bundle. The offer moves numbers far more than a new headline does, and almost everybody tests it last.
Then creative, three to five new concepts a month. Not variations. Concepts. In a market the size of the UAE the same people see your ads repeatedly, so fatigue arrives faster than it would in a large market.
This is usually the month the numbers turn.
Days 61 to 90: scale, or say so
By day 60 we know the break-even, the products that carry the account and the audiences that respond. Now budget goes up in steps against a written profit benchmark. Never all at once. The reference points we use are in what a good ROAS looks like in the UAE and what AED 20,000 a month should get you.
Here is what a settled account looks like after this process has been running a while. This one holds a 4x to 5x return on roughly AED 18,000 to AED 20,000 a month, and has done for over two years.
And if day 90 says the account cannot be made profitable at the budget available, we say that too. We would rather lose a retainer than spend a year taking a fee for something that was never going to work.
We used to rebuild structure in week one, at the same time as fixing tracking.
It made the first month unreadable. Reported numbers moved because the tracking changed, the structure changed and the learning reset, all at once, and there was no way to attribute the change to any of it. Now measurement gets a clear fortnight of its own, so month two has a clean baseline to be judged against.
Want to see what the first 90 days would look like on your account?
Send us access and we will send back the actual plan: what we would fix in week one, what we would restructure, and what we think it costs to find out whether the account can work. No obligation and no slide deck. Or see how we approach email marketing and retention.
Request a 90 day planOr see what the first 90 days produced elsewhere.
Frequently asked questions
How long before a new ad account produces results?
Month one is a rebuild and usually costs you money. Month two is where structure starts paying. Month three is where scaling decisions get made. Judging an account at week six tells you almost nothing useful.
What do you do in the first week of a new account?
Tracking, almost entirely. Conversions API or enhanced conversions, event deduplication, purchase value and currency in AED, and a written list of what each conversion action means. We do not touch campaign structure until the measurement is trustworthy.
Do you pause existing campaigns when you take over?
Not immediately. We let the account run for a few days while we watch, because the fastest way to lose a client’s money is to switch off the one campaign that was quietly carrying the revenue. We pause once we can see what each campaign is actually contributing.
What if the account gets worse in month one?
It often dips, and we say so before we start. Fixing tracking changes what gets reported, restructuring resets learning, and both happen in the same month. If month three is still below where you started, that is a real problem and a conversation.
Do you need a big creative library to start?
No, but you need a willingness to make new creative every month. Three to five new concepts a month beats one big production a quarter, especially in the UAE where the audience is small and fatigue arrives quickly.