Ask for a report on a display, video or social campaign and you’ll get clicks. Click-through rate, cost per click, clicks by placement, clicks by creative, line by line.
Then ask how many people saw the ad long enough to take it in, how often, and whether they were the right people. The answers get shorter. Often there isn’t one.
That’s the wrong way round, and the reason is simple arithmetic.
Ninety-nine and one
In most display, video and social buying, well under one ad in a hundred is clicked. Everything else was shown and not clicked. So nearly all of the budget bought impressions.
Those impressions are also where most of the effect is. People see an ad, don’t click, and later search for the brand, walk into the shop or type the address in directly. Advertising worked that way long before there was anything to click on.
And yet the click gets nearly all the attention, because it’s the easy thing to count.
I’m not against clicks. For search, and for anything built to get an immediate response, the click is the point. My complaint is about treating it as the measure of everything else.
Five questions about the other ninety-nine
I first saw a version of this at a Google event years ago, and it changed how I look at media. An impression has to get through five checks before it can do anything. Each one removes some of what you paid for.
- Safety. Where did the ad actually appear? Next to content you’d be happy to be seen beside, or on sites made to harvest ad money? If nobody has set rules, the answer is “wherever was cheapest”.
- Viewability. Was it on screen at all? The usual industry bar is half the ad in view for one second. Plenty of impressions don’t clear it: they load below the fold, in a background tab, or after the person has scrolled past.
- Audience. Of the ones that were seen, how many reached the people you meant to reach? Broad buying is cheap partly because much of it lands on the wrong people.
- Reach and frequency. How many of the right people saw it, and how many times? Once in a quarter does nothing. Thirty times in a week is waste of a different kind.
- Attention. For how long, and did the creative earn it? An ad that is technically in view and never looked at is still a miss.
What survives all five is the part of the budget that had a chance of changing what someone thinks. That is what people mean by brand lift, and it’s usually a much smaller share of the spend than anyone expected.
Why it gets ignored
Clicks arrive in the report by default. The five questions don’t. Each needs something set up on purpose: placement rules, viewability measurement, audience definitions, frequency caps, and some way to judge attention.
There’s a commercial reason too. A click-through rate always exists and usually looks fine. A viewability figure can be embarrassing. Nobody volunteers the embarrassing number.
So the campaign is tuned for clicks. That favours placements and formats that get accidental taps, and audiences who click on everything. The click-through rate rises and the campaign gets worse at the job it was bought for.
What to ask instead
- What share of impressions were viewable, and what did we pay for the ones that weren’t?
- What share reached the audience we defined?
- How many people did we reach, and how often did each one see it?
- Where did the ads run, and what did we exclude?
- Did searches for our name, direct visits or sales move in the places and weeks we advertised?
The last one matters most. If awareness is the aim, look for it in what people do afterwards, not in whether they clicked at the time.
What I’d do
Keep the click report for the campaigns that exist to get a response. For everything else, put the five questions on one page and make it the first page.
Then split your time to match your money. If ninety-nine per cent of what you bought was impressions, they deserve a good deal more than one per cent of the meeting.