If you are a new advertiser encountering online market for the first time, when you are creating your
first advertisements, you will be given an opportunity to chose what audience your advertising will
appear to. Long ago, in the very early days of Facebook, new users were encouraged to complete
their profiles by liking a variety of different organizations, specifying what schools they attended, and
their relationships.
When a user engages with almost any application, the application collects personal data about that
user and packages users into groups which then offered to advertisers for targeting. Long ago, when
this was, at its root, a mechanical matching process which coordinated user preferences with
advertisers.
In recent years, there has been increased public concerns about disinformation on the internet.
These concerns arose alongside widening discourse about user privacy. As additional security
measures, platforms took steps to protect user privacy which had the byproduct of limiting what
information was potentially available for advertisers.
Increasingly, when platforms collect information about users, this information is encrypted before it is
shared with the programs that perform the matches that determine what users will see. This
modality lacks the transparency of previous paradigms and has given rise to automated audience
finding which, is believed by many marketers to be the audience collection method which will replace
previous tools.
Going forward, more and more advertising delivery will be managed exclusively by Artificial
Intelligence. A new advertiser will need to depend more on their site text, their social media pages,
and site user lists, to inform systems about who their ads will deliver to. The risk to advertisers is that
platform’s automated delivery systems will curate bad customer lists which will result in ineffective
ads. We have an ongoing concern that, particularly in the arena of political advertising, AI advertising
modalities will reinforce echo chambers, and continue to make it difficult for people to reach
audiences that they have not been exposed to.
For big spending advertisers, bad targeting is likely not as detrimental. Big advertisers in sectors like
insurance, automotive, and pharmaceuticals want as many eyes as possible and are likely not seeking
extremely targeted audiences. For niche targeting, the reduction of advertiser controls could result in
a limited access to the marketplace, and an effective reduction in their market size. Platforms need to
take steps that continue to allow advertisers to reach new audiences. Unfortunately, so far, AI
audience building is profitable and valuable for important advertisers. For that reason, we do not
expect a return to the old paradigm in the near future.