Hello crypto buddies,
On the BATProject subreddit, there was an engaging exchange between a Fortune 500 marketer and the BAT Team. So I want to share it here you can have a better understanding of how BAT works. Disclosure: I am long BAT tokens.
Here we go!
Questioner/Marketer:
So I'm a digital marketer at a consumer-facing F500 company (as much as I'm willing to share). Not on the team that buys programmatic ads but have a good relationship with them. I've been reading about BAT and thinking about how our company could use it (as an advertiser), and had a few questions, mostly around targeting and attribution:
Am I correct that all targeting is done locally in the browser? So could I, as an advertiser, only buy an ad/place a bid for ad space for users who 1) haven't visited my site before [new customers, and truly new since Brave handles x-device], and 2) have an intent to purchase a product in my industry [determined via machine learning algo running in browser?]? If so, that's awesome and goes a long way towards solving certain problems with programmatic. As an aside, how does a local ML model even work? What if you want it to be updated,retrained, etc? Would that be done through a browser update?
How would I design attribution around BAT? Of course if a user clicks on the ad and go to our site we can cookie them and get it that way. But most people don't click, yet are still influenced by ads. One thought would be to fingerprint the Brave browser itself, but that's a bit beyond my knowledge once you get past the user agent string. So how could you prove that the ads on Brave were having a conversion lift at the user level? If that could be solved, and "close the loop", so to speak, millions if not tens of millions of ad dollars would be unlocked just from our company alone.
BAT Team:
Hi, DM me your email and we can talk through potentials if there's interest - Will provide short form here:
Here's the flow: Advertiser becomes verified, similar to how a publisher or creator verifies (with wallet) with the platform.
Server side: Ad inventory flighted into an inventory catalog (manifest) of edge cached URLs of available inventory.
Catalog pushed to devices (lang & region)
The catalog takes the role of a local ad server on the device.
Local machine learning + intent signals + browser corpus progressively train on the user's attention as they browse (this data does not leave the device).
At the opportune time, the platform activates a local auction layer and bid-matches from available inventory catalog options that fit the attention-experience context.
Winning ad presented to the user in a native push notification (ads on separate rail from content). Native push contains a call to action, and is only presented when viewable. (Attention baseline)
If the user engages with the notification, a new tab displays which either displays the landing page or a rich media creative (html5-embeddable). (Engagement baseline)
User converts on opportunity (transactional | lead-gen | etc.)
Analytics will be measured using Anonize2 zero-knowledge proof protocols (allowing for user anonymity). Analytics will be presented in aggregate in dashboards within the advertiser UI (APIs will be available).
Think about it as being able to verify with high confidence (in public) that the actions took place on a regional heatmap.
We can prove that conversions take place for a brand, without having to persistently track users or leak data. We don't have to have cloud based profiling when we can match against the most accurate data set from the local level (client side).
Users will be freq capped to >10 ads per day (fewer ads, but quality over quantity).
Some will often claim that shotgunning ads repeatedly leads to engagement. This is likely true in the existing ad model. We are using this opportunity to rework the flow into one that has a higher level of attention and likelihood for engagement. We bypass banner blindness.
In our user private ads, users will receive 70% of the revenue (in tokens) for their attention. The token replaces the need for excessive and often discrepant 3rd party measurements (with all the headaches and waste). Exchanges are direct and verifiable.
No fingerprinting or user profiling. Check out our BAT Mercury technical preview blog on.
Click here to see the full discussion.
Original post: https://www.reddit.com/r/BATProject/comments/7t2iz2/digital_marketer_at_f500_company_here_a_few/
Thoughts? Questions? Please leave it in the comments.
The Non-Greedy Crypto Investor (
Disclosure: I am not a financial advisor nor am I advising you to do anything. Please do your own research. These activities involve risks, including the loss of all your money. Tread carefully at your own risk.