For the last few years, audience targeting has become an increasingly central part of Google Ads.
Remarketing, Interest category, custom affinity, in-market audience, customer match, detailed demographic… the list continues to grow.
But as the options and apparent precision continue to increase, how accurate – actually – is it?
One interesting way to gauge the accuracy Google’s audience categorisation, is to take a look at the interests Google has earmarked for you…
You can see what interest categories Google associates you with at https://adssettings.google.com
You’ll see something like this. (Prepare to see my bared soul…)
There is no clear demarcation of which of these would feed into your in-market, interest/affinity or demographic categories, but if you click into each one, you’ll see further details, starting
This advertiser shows you ads based on:
Google estimates this interest, based on:
Google estimates this demographic because:
relating to the remarketing lists, interest and in-market categories, and demographics you’re pegged for respectively.
If you want to mould your profile you can turn off any of these categories – or opt out of personalised ads altogether.
What does this mean for PPC Advertisers?
Seeing how hit and miss your own interest categories are will give you some idea of how far from perfect category targeting is.
You’ll almost definitely see some head-scratchers in there…
Take this sample of mine:
Of these 10, five are either genuine interests or topics that I’ve knowingly browsed recently.
Two I can put down to my children borrowing my phone.
Two will have come from my activity relating to a sports brand client (basketball equipment? Have you seen me..?)
…and one is a complete mystery. (I like a beach and island as much as the next man but they’re not really on my radar at the moment.)
So on an individual level – the list doesn’t seem to be a particularly accurate reflection of my interests.
But it’s a lot better than random.
So on aggregate (taking the ‘signal strength’ from a whole interest category group) the categories become a very usable tool… just don’t imagine they’re anywhere near 100%.
And remember there are some targeting variables (like gender/device) where you can treat the labelling much more reliably than others.
Final Note on Audiences
When using audiences with search campaigns, it’s important to remember the distinction between what you do for data, and what you do for performance (I have seen this blurred even with high-level PPC management).
Adding relevant audiences on observation will only confirm or deny whether the members of those audiences are responding well to your ads.
That’s useful if you then go on to upweight their bids once you find that they do.
But on the other side of the coin, low-performing audiences are just as (or more) useful to add and identify, in order to downweight or exclude them.
And if you’re using smart bidding, optimising with manual audience bid adjustments isn’t an option… which means it’s doubly valuable to add less desirable audiences on observation, to root out and exclude the low-performing ones.