The impact this potentially creates can be defined into two schools of thought from a media value front.
Less targeting capability will increase the cost per result, because you won’t be able to target the users you deem most relevant to your product.
Ads will be less relevant to users. There’s a heap of data in the Facebook platform, but that’s where it needs to stay. We never had a problem with it being there so long as it wasn’t used against us politically. That seems to be a taboo area for consumers to deal with, even though they’re influenced to consider purchasing products every day.
From the market’s perspective, the ecosystem needs to be as closed off as Apple’s. Third-party integrations will be closed, resulting in a garden that is even more walled off than before.
Pulling the pin on third-party data means there will be a crossover of time where people will have to learn how to do their targeting differently. Back how it used to be done. Define a persona, develop a segment, model the audience. Get under their skin.
Short term: Less clicks, less relevance, less results, higher CPC
Long term: Marketers that understand their audience better than a cookie cut “xx persona” or “xx segment” and develop a better way to find and understand their audience.
Less targeting capability will decrease the cost per result, because you won’t be wasting money on data you didn’t actually need.
While the data targeting capability with partner categories is powerful, typically the way it has been used in the past hasn’t allowed the system to have the same optimisation capability as a machine learning algorithm would do on an open field. No surprises here for those that believe in the algorithm, but let me play this scenario out to you.
You have a target audience of mums. You use third party data segment to target people who have purchased baby products in the past 30 days and have been identified and matched within the platform.
This data then costs you from 10-20% on top of your existing CPM, according to Facebook. Which means that by you targeting this segment only, it needs to work 10-20% more efficiently than an open plane that a machine learning algorithm can find you an audience for.
That’s right: you thought that people that buy baby products are your right audience, but the machine found out that people that are into cycling are more cost-effective for the outcome you’re trying to achieve, but you never gave it that opportunity to find that audience because you thought you were smarter than a machine learning algorithm.
My prediction
I am predicting the latter will happen. Some marketers will become surprised to see that going super granular on their targeting doesn’t achieve as strong a result compared to going broader and focusing on the context of the user, not just their behavioural attributes.
Context rules, we just never noticed because we’ve been distracted by the depth of data we can get out of the platform.
Some may feel discombobulated with this change for a short while. Meanwhile, us relics will feel back at home and have an opportunity to train the art of developing a segment and understanding an audience again, and for it to be adopted out of necessity rather than opinion.
The understanding of the audience will get deeper and the next wave of marketers who use paid social media will get a better result out of the platform and for their clients.
Targeting has way more depth to it than just picking that segment and targeting it. The PEOPLE you are targeting have other interests beyond that which has defined them as a segment. Now you will need to dig deeper to understand that the mum who buys baby food also loves reading a specific type of content, and that content is the best place to reach them, not their purchase history or cookies.
Victor Condogeorges is digital investment director at Bohemia Group. This article first appeared on LinkedIn.
Word of the week: discombobulated
“Targeting has way more depth to it than just picking that segment and targeting it.”
You’re worrying and confusing a lot of people when you say that.
User ID not verified.
This guy has hit the nail on the head
User ID not verified.
My understanding of what Victor is saying is that rather than: “targeting pre-defined audience segments with similar attributes to hit them with ads”, try “targeting the right context to understand who is your audience then learn to define them better”.
I agree, there are no true cookie cutter definitions of audiences and we need to shake this way of thinking about – and I’m glad it’s capitalised – PEOPLE.
User ID not verified.
Partner Categories and third-party data never really performed well anyway. It’s a finger in the air approach. Best option for targeting is to go broad and use the Facebook pixel or algorithm to optimise to the right audiences in real-time. Machine learning. Do a split test in ads manager and I guarantee that your desired action will be more efficient. Aside from this advertisers can still bring their own data onto the platform.
User ID not verified.