New Rules of Advertising

In 2012 I started saying to people that the (advertising) cookie will die. In 2022 it’s happening and it’s the single biggest thing to disrupt advertising in two decades.

Mikko
4 min readMar 9, 2021
Hilma of Klint

He is a kind man, he is always taking care of others. Without years of his generous mentorship, I think I still would not understand the way modern online advertising works. I debated with him the “death of the cookie” for the second and last time in 2012.

I said, “the cookie will die”. He then replied…

“Without the cookie, the whole industry will stop functioning. None of the things we are doing now would make any sense and we would no longer be able to do our jobs”.

Why is it such a big deal if the cookie goes away?

The way the programmatic advertising eco-system works is that there are a lot of different systems serving different functions. There are thousands of companies operating such systems. These companies, for the most part, depend on the cookie directly or indirectly.

For many companies, their business entirely depends on the cookie.

For the great majority of companies, their business notably depends on the cookie.

The cookie is the single most ubiquitous thing in modern online advertising.

If there is a total block for cookie-based tracking right now, with the exception of the major platforms — such as Google and Facebook — most of the eco-system would cease to function, and virtually all of it would be adversely affected.

These effects would be most visible through an instantaneous collapse in advertising effectiveness across all campaigns that are somehow dependent on the technology and services of the affected companies.

What is the current status of cookie-era advertising?

The cookie is going to die and the era will end.

Devices are already blocking the cookie.

Browsers will soon all block it.

Everything that the cookie makes possible for the programmatic advertising eco-system, will become hard or impossible.

Cookie was an exploit. It provided an expedient means for advertising systems to communicate with each other.

This exploit will now be patched by Apple, Google, and others.

Any company that depends on the cookie directly or indirectly, can no longer depend on it.

Any value such a company delivers to their customers today — depending on the cookie directly or indirectly — can no longer deliver it, or it will become hard to deliver it.

Within two to three years the cookie is gone.

With it, thousands of companies will be gone.

What happens next with people-based targeting?

As a direct response to the cookie’s demise, there are four different tracks of development, three corresponding to a method of targeting people, and one entirely avoiding it. These four are in addition to various walled-garden approaches, which are unaffected by the cookie’s end-of-life. Separate from these approaches are contextual and other means of targeting that are independent of the person (user).

The four are:

  • Circumventions
  • Cohorts
  • Zero-Knowledge
  • Zero Data

These four are all-pervasive; every media buy that leads to a delivered impression will fall into one of these categories of targeting.

The four targeting approaches can be evaluated against set criteria:

  • How much it will eat working media?
  • How much it will consume energy (and cause greenhouse gas)?
  • How well will it be supported by browsers and devices?
  • How precisely it allows to target individuals?
  • How can it be trusted by advertisers and people?

Circumventions are various approaches that focus on individuals, where means other than cookies are used to create profiles of individuals and households. Such ids can then be used for example on a limited number of publishers based on login information.

  • Cost similar to cookie-based approaches
  • Will consume a significant amount of energy
  • Highly unlikely to receive support from browsers and devices
  • Targeting precision will vary depending on the approach
  • Trust depends on propriety actors (e.g. advertising technology companies)

Cohorts are methods focused on grouping people. Cohort groups are generic enough or large enough, to provide guaranteed anonymity. Individuals are assigned cohort-id which can then be used by an advertising system to target ads to all the people with that cohort-id.

  • Cost similar to cookie-based approaches
  • Will consume negligible energy
  • Highly likely to receive support from browsers and devices
  • Targeting precision will be low to moderate
  • Trust depends on open standards

Zero-Knowledge methods are focused on using crypto-economics and blockchain technology to provide guaranteed anonymity while allowing any level of individual targeting. Zero-knowledge approaches work on any platform where both the platform and the user of the platform have approved the use of their data for targeting.

  • Cost similar to cookie-based approaches
  • Will depend on notable energy consumption
  • A moderate chance for receiving support from browsers and devices
  • Unparalleled targeting precision
  • Trustless (i.e. complete confidence without 3rd-parties or standards)

Zero Data methods are focused on entirely avoiding, by the means of system design and engineering, all forms of data associated with people. The methods entirely avoid any concern or risk with respect to user data and instead focus on other vectors of data.

  • Zero cost
  • Zero energy cost
  • Guaranteed support from browsers and devices
  • No targeting precision (targeting is prohibited by design)
  • Guaranteed trust*

*Assuming the relevant parts of the code are maintained as open-source.

Word of advice

Follow these simple principles…

  • Stay away from Circumvention
  • Cohorts are easy to invest in but will deliver limited value
  • Zero Data is a safe-haven in the short-term and a hedge for risk in the long
  • Zero-Knowledge will take time to scale but will deliver the most value

Eventually, an advertiser should split their media investment so that Cohorts, Zero-Knowledge, and Zero Data investments are each present and not less than 10% of the sum of the three.

Thanks for your time! If you have a few seconds more, hit that clap button and share in your socials.

--

--

Mikko

Worked with machine intelligence for 15 years, and built the interwebs for 25. Nothing here is my own.