A marketing director shared her screen on a call a few weeks ago and parked the cursor on one row. Streaming. A wall of impressions, almost no clicks, almost no sales attributed. "I think we call it," she said. "Nobody clicks on a TV ad."
I asked her a different question. Where do your customers actually buy? She went quiet, and then she started counting on her fingers. Her own website, for some of them. A large online marketplace, for plenty. A phone call, for others. A physical location, for the rest. She stopped at the fourth finger and looked at the screen again. "The report only shows the first one," she said.
That sentence was the whole call. The ad had not failed. It had been graded in the one room where it was never going to pay out.
People find you on one screen and buy on another
Think about how a streaming ad actually works on a person. Someone is watching a show. The spot runs. They do not reach for the remote. They reach for a phone, or they remember a name, or they mention it at dinner. Discovery happens on streaming, on podcasts, on social. The purchase happens later and somewhere else.
More and more, that somewhere else is a marketplace or a retail media environment. Those places keep the customer inside their own walls. They hand back aggregate numbers at best. For some brands, a large share of total ad spend is now aimed at sales that close inside a walled garden. The ad runs in the open. The result lands behind a door you do not have a key to.
The four places a streaming ad pays off
I count four. The first is your website and your branded search. A viewer sees the spot, types your name into a search bar, and lands on your site. Standard analytics can observe this one, though even here it tends to credit the search click and not the show that made the person search.
The second is the inbound call or the booked appointment. A person watches something on Tuesday and phones on Thursday. The call log has a timestamp and a duration. It has no memory of the show.
The third is the physical store or the dealer lot. Someone walks in because they saw you on a screen at home. Nobody scans anything on the way through the door. If a salesperson asks how they heard about you, the answer lands in a notebook, not a dashboard.
The fourth is the marketplace. The buyer leaves the show, opens the marketplace app, and buys your product there. You will see total sales rise in the seller account. You will not see a path from the ad to the order.
Only the first of those four is reliably visible to standard analytics. The other three finish outside the pages it instruments.
One meter on four pipes
Picture a tank fed by four pipes. You install a flow meter on one of them. A month later you read the meter, see a trickle, and conclude the tank is barely filling. The meter is accurate. It is also attached to the wrong pipe, or at least to a quarter of the plumbing.
That is what a channel report does when most of a channel's work lands in the other three places. The report is honest about what it counted. The mistake comes when someone reads the count as the whole of the effect and moves budget on it.
When you cannot follow the person, measure the system
The usual response is to try harder to track. Better tags, tighter integrations, one more pixel. I think that road ends at a wall, because the wall is by design. A marketplace will not give you the individual. A store will not hand you a click. A phone call does not carry a cookie.
So stop asking each doorway to testify. Measure the system. Look at what happened to total sales, across all four places, when the spend went up and when it went down. Model the whole outcome instead of reconstructing one customer's walk from screen to checkout. That approach does not need to watch anyone cross any threshold. It needs the business results you already hold, set against the media you ran.
Map where your sales physically happen
Before you trust any channel report, map where your conversions physically occur. If you did the afternoon exercise from the last issue, take it one step further.
Write down the four places. Next to each, put a rough share of your sales. Your marketplace seller account, your call logs, your store or dealer sales reports will get you most of the way. Mark which of the four your analytics can see. Then look at where your discovery media is aimed, and ask how much of that spend is pointed at an outcome you have just marked as invisible.
That gap is the portion of your advertising you have been grading blind. It tells you where you need an independent read, and it tells you which reports to stop quoting in the budget meeting.