I expect you’ve seen ad targeting by hashtag. If you’ve not got your own example top of mind, you will if you start looking. Leave a comment below or tweet me up if you have.

My recent example started with a short afternoon walk outside my office. Yesterday, I saw an odd scene in a nearby park, so I shot and shared it.

sweatpants, tree, Colorado Springs, Instagram, ethanbeute

The image went up on Instagram with this description:

“#Sweatpants in a #tree. #Really.”

The intent of the image is to conjure either a comical or a sordid story for its viewers.

The image is generally unfavorable toward sweatpants; it doesn’t likely produce sweatpants-desiring thoughts or behaviors.

But …

Ad Targeting by Hashtag

With what was I greeted this morning on Facebook? An ad for sweatpants (15% off!). And not just once – every time I refreshed the page. And not just one – they’re A/B testing two treatments.

Facebook, ad, advertising, hashtag, Instagram, targeting, advertising, testing, A/B test

I’d like the A/B test better if they either: used the same image and tested the effect of the Betabrand name or used the Betabrand name on either or both and tested the two images against eachother. As they are, any effect can be attributed neither to the specific image nor to the use of the Betabrand name.

More importantly, though, the hashtag used yesterday obviously provides insufficient context for this Facebook ad targeting today.

Beyond the facts that I’ve not work sweatpants since I was a young child and I’d never consider wearing “dress sweatpants,” the use of #sweatpants in describing the Instagram photo was not favorable.

This is obvious to any human who saw the image together with the hashtag. This is not so obvious to an ad-serving machine looking for hashtags related to their product or service category.

The Bottom Line

People employ hashtags with too much subtlety and irony for this targeting to be highly accurate.

Perhaps not too much for the advertising to be cost effective, but too much for them not to serve regularly ads that lack relevance beyond comic coincidence.

It’s a fine effort from Betabrand – you have to try out the tools available. And, again, it might be effective!

Here, though, the result was as comical or sordid as sweatpants in a tree.

For meaning and relevance, context matters.