When reviewing ad accounts for SMEs, seven out of ten times the issue isn't "wrong strategy" but one of three mechanical failures. None of these will show an error message, and the account might even look good on the surface — which is why they are so costly.
Automated bidding doesn't question whether the signal is correct; it simply bids based on the numbers you provide. So, if the conversion tracking breaks, it won't stop — it will accelerate to buy "seemingly cheap" clicks that don't actually convert.
The most insidious scenario is when a "key event" definition is removed or altered in website data tracking. This change might look harmless in the analytics tool, but the bidding signal received by the ad platform will be immediately disrupted. The system won't show any error messages, and the numbers in the reports might even look better because the denominator has disappeared. By the time you notice, money will have been wasted for a while.
The solution is straightforward: compare the conversions reported by the ad platform with your actual orders in the backend, day by day. Only trust the data when it matches; if it doesn't, investigate immediately, don't wait until the end of the month. This is a step we always take with our clients before spending any money — if the tracking isn't verified, we don't increase the budget.
The second cause is harder to spot because the problem isn't with the ad itself, but with the report you're looking at.
If you use a rolling time window (e.g., "last 30 days") to view performance and you change the primary conversion action in the middle, the window will simultaneously include data from both the old and new metrics. The system won't give any warnings, and you'll just see a number that looks perfectly reasonable. Moreover, the direction is fixed: the results will always be overly optimistic because the new metric is usually calculated more broadly.
The solution is to lock the conversion action used in the report, don't rely on the platform's default settings, and don't compare cross-period numbers after changing the metric. Before making any "increase budget" or "pause campaign" decisions, confirm that you're comparing the same thing on both sides. This isn't a theoretical issue — in the accounts we've audited, this is the most common source of "good numbers but no business impact".
Shopping and target ROAS bidding need at least two to three weeks of data to be meaningful. And conversions themselves have a delay — a click today might convert three days later — so the ROAS for the most recent few days will always be low, and using it to draw conclusions is just scaring yourself.
Another common misjudgment is: seeing that the ROAS hasn't reached the target, and then raising the target ROAS even higher. In reality, if the target is set too high, the system will simply give up bidding, and you'll go from "making less" to "no exposure". To judge whether the target is too high, you should look at the rank-lost impression share (the percentage of impressions lost due to insufficient bid), not the ROAS for the past few days.
Ending a campaign after three days is the most expensive way to save money — not only do you not get any conversions, but you also buy a bunch of useless learning data, and when you restart, you'll have to start over.
If clicking on an ad leads to the homepage instead of the specific product, it's like buying a customer and then telling them to find their own way. If you have thousands of SKUs in your catalog, and all ads point to the homepage, no matter how clean your product feed is, it won't help.
Therefore, the product feed, landing page, and campaign should ideally be handled by the same team: price, inventory status, and title should be consistent on all sides, so that what the customer sees when they click on the ad is what they saw in the ad. When these tasks are split between two companies, the most common result is that both sides think it's not their problem, and the customer is lost in the middle.
When we do Google Ads, we handle the landing page as well, to avoid this kind of no-man's-land situation.
The sequence is fixed: tracking comes first, then spending. We verify every conversion event on the website and the data received by the ad platform from start to finish, match it with your actual backend orders, and then start the campaign. If the landing page isn't right, we'd rather fix the page first before running the ads.
Google Ads management starts from HK$3,800 per month, and if the monthly ad spend exceeds HK$10,000, we charge 20% of the ad spend. The advertising budget can be paid from your own account, or we can handle it for you, as long as it's clearly stated in the quote. If a channel doesn't pay for itself, we'll tell you directly to stop and suggest where to allocate the budget.
If you want to know whether your account is currently affected by any of the above issues, we offer a free ad and analytics account check: just provide read-only (viewer) access, one per company, and we'll write a short report for you after reviewing it.
Tell us in one email what you want to do online and we will come back with a written scope and a fixed quote.
Talk to us →