Streaming television is the easiest advertising spend you will ever approve and the hardest to defend three months later.
The pitch writes itself. Your face, your listings, on the big screen in the living rooms of the neighbourhoods you actually work, without buying the whole market the way you would have had to in the broadcast days.
Then the quarterly review arrives, somebody asks what it did, and the room goes quiet in a very particular way. The channel with the most beautiful creative has the least convincing receipts.
It resists measurement by design
Nobody clicks a television. That is not a reporting bug you can configure your way out of, it is the medium.
So the industry substitutes. It measures the things it can see and presents them as if they were the things you asked about. And because the substitutes have confident names and arrive in a tidy dashboard, they get treated as evidence.
It gets worse in housing specifically, because you cannot ring a neighbourhood any more. No ZIP targeting, a radius floor on both major platforms. So the geography you are actually buying is far broader than the geography in the pitch, and a good deal of that beautiful impression count landed somewhere you do not work.
The numbers that are theatre
Completion rate. Ninety-something percent of your advert was watched to the end. Of course it was. It was unskippable and the remote was on the arm of the sofa. That is a measure of the format, not of your message.
Impressions. A television was on and your advert was on it. Whether a person was in the room is not a thing anybody is measuring.
Households reached. Real, and quietly enormous, and almost never cross-referenced against whether those households are inside the area you can actually service.
Lift. Sometimes genuinely rigorous. Frequently a comparison between people the system chose to show your advert to and people it chose not to, which is not a control group. It is a selection effect with a chart.
How to actually know
Three things, and none of them come from the vendor.
Ask. Every enquiry, every listing appointment, in your own words. "How did you come across me?" Write the answer down somewhere you will read it again. It is unscientific and it is the only measurement in the whole stack that nobody is grading themselves on.
Watch what happens to your own name. Brand advertising does not produce clicks, it produces people typing your name in a fortnight later. If you cannot see searches for your name and your team name over time, you have no instrument capable of detecting the thing you bought.
Turn it off. Deliberately, for a defined stretch, and see whether anything moves. This is the only honest experiment available to you, and the reason almost nobody runs it is that the answer is binding.
I am not telling you to stop
I want to be fair here, because this is where these pieces usually get lazy.
Being seen matters in a business where people choose a person, not a product. A seller picking between three agents they have already met is making a decision about familiarity and confidence, and that is precisely the sort of thing television has always been good at. There is a real case for it.
The case is just not the one on the dashboard. Buy it as brand building, on purpose, with a number you have decided in advance and a way of noticing whether your name is being searched more. Do not buy it as performance advertising and then act surprised when the performance evidence is decorative.
Any road up
Have you thought this all the way through? If the honest answer to "what did it do" is "we think it helped," that is allowed. It is a perfectly respectable answer for brand spend.
What is not allowed is a report full of completion rates being waved about as though it answered a different question. Know which one you bought.