Before trusting ad numbers, check where your sales come from
Anyone who spends money on ads asks the same question at the end of the month: which money brought in customers, and which simply vanished? To answer it, the path between the click and the payment has to preserve the information about where the person came from. When that information is lost, a sale can end up credited to a different channel than the one that brought the first contact.
The problem of the customer who “appears out of nowhere”
Picture a hardware store that advertises on Instagram and also shares a link on WhatsApp. A customer clicks the ad, looks at the catalog, leaves, and later comes back by typing the address to buy.
This is an illustrative example. If the system doesn’t preserve the source of the first contact, someone looking only at the record of the return visit may see a direct visit and leave the ad out of the picture. The decision to cut or increase the budget then rests on an incomplete view.
What to record on every click
Campaign links can carry tags, called UTMs, that identify the source, the medium, and the campaign. They work like the label you put on merchandise when it enters your inventory: they help you recognize where it came from.
In a recent campaign tracking project, I took care of two points:
- Store the first recorded source. Preserving the information that comes with the first contact helps you understand how the person arrived, even when a later visit happens through a different route.
- Carry that information all the way to payment. The source has to follow the navigation and reach the purchase records. Storing it only on the landing page leaves part of the path uncovered.
When interpreting this data, you have to separate source from classification. A visit coming from Facebook may have started from an ad or from a link that was not boosted. The social network alone doesn’t prove paid traffic. Google Analytics itself distinguishes organic and paid social traffic.
When there isn’t enough data to identify the campaign or the channel, the report should show that limitation. Filling in the field with a guess gives an appearance of precision that the records don’t support.
Tie the spend to the right campaign
There is another part of this check: separating the spend. In an account with campaigns for different offers, I made the connection through each campaign’s identifier. This lets you match records without depending only on the name shown in the dashboard.
If the spend from different campaigns is added together and attributed to the same product, the cost calculated for that product is distorted. The decision to continue or stop a campaign then carries that error.
It’s worth asking whoever manages your ads how the spend is separated and which identifier connects each campaign to the data in the report.
I help owners of small and midsize businesses organize processes, track metrics, and automate routines. If you want to understand where to improve your operation, take a look at my management and process consulting. The first conversation is free and carries no obligation.
Test the whole path before you trust it
After the adjustments, I simulated a visit with the source recorded, went to the payment screen, and checked whether the information was still there after returning to the site. In the tests, it was preserved.
The events from this check were labeled as tests, so they wouldn’t contaminate the numbers used in the analysis. A historical test purchase was also kept separate, without being sent again as a new conversion.
This distinction matters for anyone reading the dashboard: a simulation helps verify that things work, but it shouldn’t show up as a sale won.
Reaching payment doesn’t confirm a sale
In that check, I went through the navigation up to payment without completing a real purchase. The test confirmed that the data was preserved in that part of the path. To validate the full sequence, I still needed to observe a controlled purchase through to the final record.
Visits, payment starts, and sales are different stages. Before drawing any conclusion about the return on ads, you need to know which of these stages the report is counting.
And identifying the source of a sale is only one part of the analysis. To evaluate what it leaves for the company, it’s also worth looking at contribution margin.
What you can check in your own company
- Follow one sale through the records you have. Ask to be shown which data links that purchase to an ad or another channel. Check what is known and where the gaps are.
- Understand how spend is separated. Verify that each campaign and each offer is tied to the right records.
- Investigate direct traffic. It may represent a real direct visit, but it also deserves a check when you suspect campaign tags are getting lost. The channel name alone doesn’t explain the whole journey.
- Separate testing from commercial results. Checking a visit through to payment and observing a completed purchase validate different parts of the process.
- Keep the limitations visible. A report should let you tell recorded information apart from interpretation.
The decision of where to put the next real of ad spend depends on understanding the results of the previous investment. It’s worth checking the quality of that information before using it to distribute your budget.