Keep the BRAVE up: Challenges the new Basic Attention Token ecosystem needs to address

The Brave browser not only allows you to own the personal information you share with the world, but through the Basic Attention Token, it proposes a new mechanism for brands to purchase ad space and reward publishers. In a silly intro post here (link), I detailed the few mechanisms that define Brave in a world where consumers are worried about tracking and malicious 3rd party code.

This post will go further into detail, specifically outlining the challenges Brave and BATs will have in adoption by advertisers and, conversely, challenges advertisers and brands will experience if the browser reaches a tipping point. Knowledge of the white paper is necessary (link here), as it is the key reference point to any of my thinking around Brave, even if I mostly go off in tangents not addressed by said paper.

My experience in the advertising world is that of ad-tech, from Data Science to Attribution. I work closely with a variety of marketing and brand stakeholders across verticals, and I believe the only way to thoroughly review any innovation is through the eyes of the end users. So here we go, framing a few challenges through their eyes in the best way I can:

Marketing Manager: What is the ROI of my investment in a Brave ad?

Reminder of how BAT works from the point of view of an advertiser: let’s assume that an advertiser has secured ad space (more about auctions below). The price you pay in BAT for that space will rise with the attention provided, as calculated through a specific Basic Attention Metric, within the browser. As an advertiser, you are charged a specific BAT amount.

Then, how do you calculate the ROI of this investment? Well, let’s assume you agree that the mechanism itself guarantees that this ad will cost exactly what it should cost proportional to the attention won. However, it doesn’t answer the question whether the attention you are paying is actually driving any sales, or any other KPI. This is what we use 3rd party tracking for — e.g., a cookie is dropped when an impression is served, we then link up that cookie id to any purchase history on a brand’s site over a period of time (usually 30 days).

Brave destroys the ability to validate the attention we purchase. As a marketer, I would be much less interested in the world where Brave existed because of the uncertainty in the tangible, non-attention based impact of my advertising.

If Brave was the browser of choice for everyone, this uncertainty would probably lead a marketer to invest less as a whole in Brave, and push the spender further to closed garden solutions such as google search and Facebook. Overall, high adoption of Brave would nudge marketers to revisit analytical tools which have been left in the dust by Attribution and Machine Learning algorithms, the ones relying solely on tracking data. These tools are Media Mix Modelling and other Econometric services which mine for statistical relationships between investments and hard KPIs, and while are susceptible to high degree of error, they do not require 3rd party tracking.

Agency planner: How do I bid for a placement?

How does programmatic buying work today? There is near real time auction between advertisers for on demand audiences on a publishers site. How do you know your maximum bid, your reservation price for an impression on a publisher’s content? For one, you would like to calculate it through expected ROI.

As we have just shown, with the Basic Attention metrics, you can’t really do that. Nor could Brave really commission a study and get to the bottom of it for all advertisers. One unit of measured attention will be different for advertiser A than advertiser B in its impact down the funnel. So how do you decide your bid? Without validating BAT metrics, you can’t.

ROI isn’t only metric that advertiser use to set their bidding. The certainty of premium audiences is another. Identifying a premium audience may shift the reservation of a marketer way north. We know that under Brave, the advertiser won’t have access to your personal info as they see fit to collect (ignoring browser signals at the moment). You can’t validate a campaign for 20–30-year-old female tech workers because you can’t cross reference their cookies with audience aggregators and calculate the % of the total impressions which actually are 20–30-year-old female tech workers.

I general, the auction system is a black box. I have only a corollary set of questions for the Brave team knowing that some local machine learning is supposed to happen within the browser: how do you decide which ad is shown between two ads? Will Brave run competitive ads on a specific placement, rotating until enough sample size is determined to know which ad draws the highest BATs for a specific browser? If an ad earns so much BAT, will a campaign manager be allowed to limit maximum spend, when attention is too high based on budget? We need a bit more than 2 models and Coase’s theorem (as per the white paper) to make this come alive.

Direct Response Marketers and Agency Planners: Why wouldn’t we just cookie bomb certain publisher’s audiences?

The black box at the moment is the suggestion that the Brave browser will do some machine learning on your own data to determine which ads you should be seeing, although again, not entirely sure how that syncs with any auction system.

Let’s leave it at that… and do some speculation.

Well, as a smart marketer (of course, meaning very unscrupulous!) what would I do? I would spend quite a lot of money dominating a single user’s device, and his algorithm input. That is, I would make sure that the only information that the browser had, was my brand’s information. Then the browser wouldn’t be able to predict how any other brand would perform on the device, rendering their ads impossible to be placed on a user’s browser.

This is a principle in machine learning: self-fulfilling prophecies. The best ad is a Nike one because the algorithm is only make Nike predictions.

Data Scientist and Measurement Analysts: What browser signals will we have available and why?

A part of the aforementioned black box is the further idea that there will be certain browser signals in Brave that advertiser can see, and utilise for their ads.

It’s slightly counter-intuitive (as I flagged earlier…) to the policy that your Brave browser will make you the owner of your data. However, I don’t want to make any assumptions as to how this will work — it could yet be anonymised.

More questions arise: There definitely needs to be some basic information communicated to advertisers about the viewers. However, what these will be should inevitably also vary from advertiser to advertiser, as each advertiser will want different info for their campaign: how many times did a visitor buy over last few weeks? What is the location and gender of the visitor? How engaged are they with content in general against the average user? Finding a common set of variables, or a common schema that advertisers or the Browser itself can train algorithms on, well that’s simply integral for any of this to work.

Again, without much information here, my suggestion here would be for Brave to create a secondary market place for user information: visitors are asked to share information and are paid in BATs, then ads adjust according to that information. Giving up privacy is rewarded with something more than just better ads, and yet ads might be made better.

Thanks for sticking with these tangents. The Brave Browser and Basic Attention Token are two bits that I spend quite a lot of time thinking about. The benefits for one end user, the browser, are very clear. I use Brave, but currently there are basically no Brave ads on the sites that I regularly visit.

Ultimately, if all consumers dropped Safari and Chrome for Brave, advertisers would have to adopt it. It would be me a mess, as much of the auctioning, insight and evaluation and general infrastructure still hasn’t been developed, nor has the white paper made me confident that any of these functions are going to made redundant.

Further clarity, a white paper 2.0, is needed to help bridge this exciting piece of tech with advertising end users.