From Ogilvy to Adtech: Programmatic CRM with AppCoins
TL;DR: The AppCoins protocol proposes a simplified mobile advertising model where the players are the developers (advertisers), users and app stores. In this post we introduce a protocol extension concept that extends AppCoins capabilities to Programmatic CRM.
David Ogilvy was the most influential advertising man of the 20th century. He was the original Mad Man, born in England in 1911. After dropping out of Oxford, he moved to Paris and found work in the kitchen of the Hotel Majestic. He only made his way into advertising when he was 38, but by 1948 he had already started his own advertising agency. He would eventually become known as the “King of Madison Avenue”, having some of the biggest companies in the US as his clients.
He achieved all this by basing his decisions on a strict marketing research process, rooted in reality and facts. He also had a very strict and scientific management process for the creatives.
This was inspired by his previous experience at the Hotel Majestic kitchen. Most agencies up until that moment avoided this rational management perspective.
As a man of vision he continued to shape advertising and marketing practice. In 1983 he published the seminal book “Ogilvy on Advertising”, where he predicted 13 future changes in advertising. For a long time most of his predictions seemed to have fallen flat.
Amongst these famous predictions two of them came to happen recently in the digital marketing space:
1. Direct-response merged with advertising.
This is evident in today’s merging of advertising and marketing technology (the merging of Adtech and Martech). Direct-response, also known as direct marketing, gave birth to CRM during the 90s. With CRM came the first CRM suites (hailed as the first Martech systems). But these two worlds of Adtech and Martech remained separate until recently. Today you can use your customer contacts from CRM systems to send advertising campaigns across media and channels. This has been referred to as “Programmatic CRM”. These are no longer two different activities.
2. Scientific methods became the norm with the rise of marketing analytics.
Ogilvy was one of the pioneers of advertising research and fact based management in advertising. This included mainly marketing research processes prior to advertising campaign development, but also ad copy testing. At the same time, direct-response analysts were also pioneering other types of marketing analytics. Together these two worlds introduced many tools that are standard nowadays such as A/B testing and predictive modelling. One of the main reasons why this has become the norm is that data is readily and easily available. Autonomic systems estimate optimal ad creative performance before deploying the best campaign.
With the convergence of advertising and CRM we are witnessing the rise of automation systems for one-to-one marketing. This is marketing where the decade’s long dream of marketing holy grail becomes a reality: large scale personalisation.
While this has generally brought increased efficiency in web channels, mobile is still a hard nut to crack. There are no universal open protocols for the different mobile platforms (the main ones being Android and iOS). Even Android, which should be an open platform, has a number of limitations imposed by Google. That explains why mobile advertising is plagued by a long chain of Adtech intermediaries.
Google already has the “Customer Match” system that allows a similar type of targeting across the Google Display Network. This includes app install campaigns in Google Play and AdMob. Apple has a similar system that allows targeting users through on search results across Apple devices (iPad and iPhone).
And that brings us to AppCoins. AppCoins caters for financial transactions on the blockchain which include mobile advertising. Hence it is a Fintech system as well as a novel type of mobile Martech/Adtech that can do new things across platforms.
AppCoins introduces the Proof-of-Attention (PoA) system which provides a solution to mobile advertising problems. Its an open protocol that provides a way of confirming mobile advertising attribution based on a new Cost-per-Attention (CPAt) model. With this we can confirm that the user has paid 2 minute attention to a mobile ad as is described on AppCoins whitepaper.
But how does the AppCoins Protocol support Programmatic CRM?
While this is not directly addressed in the protocol definition, we argue that it will work based on the current specification described in the whitepaper.
Let’s see why:
- When a bid is placed we can map the developer’s app from its portfolio with a set of users that match the filters.
- The filters can include a variety of options. The developer could request that the filter excluded all users that didn’t match a certain user hash
This user hash must be a shared user identifier between the developer and the app store (more on that below). Then the usual PoA system could be applied to that filter. Once the PoA is confirmed by the oracle (app store) the developer has confirmation that the specific user has paid attention to the targeted content.
But how does the app store identify the user?
The user identification must be done through the usual mechanisms. In the Android ecosystem this is usually done through the device’s Google Advertising ID (GAID). On the Apple ecosystem it is the Apple Identifier for Advertising (IDFA).
This is a code that ensures that you can target the device while keeping personal details hidden. This GAID or IDFA can be used also as the shared user hash between developer and mobile app stores. Through this system the app store is able to target that specific user with a sponsored app. Then they can confirm the conversion attribution through the PoA procedure.
Lets imagine that a marketer managing a CRM program for a mobile app developer has a customer database Γ and wants to run a cross-selling campaign on this database. Since we are in a programmatic CRM environment we can define a campaign as a set of bids.
The company has a portfolio of apps that are being distributed for both Android and iOS. Basically he wants to present a complementary app from the portfolio to each user. Different users are using different apps, so its interesting to do individual personalised recommendations. That means one bid per customer.
To achieve that the marketer has defined a set of rules that maps each customer from Γ to a vector of apps Α of the developer’s portfolio that they believe the users will install.
This vector of apps A can be defined manually or automatically through a machine learning model. The machine learning model learns app offer success from a testing campaign that includes a variable about offer acceptance and the features of each user. Thus we can consider this to be a supervised learning problem. The resulting model f maps rows of Γ to A:
Once the individual app targeting is set, the marketer can start the campaign. As long as the marketer has the GAID or the IDFA on the CRM database Γ he can now do direct marketing campaigns on any AppCoins compliant mobile app store. All he has to do is publish a bid for each row of Γ on the AppCoins advertising ledger. Note that different customers have different customer lifetime values (CLTV). A very simple way of calculating CLTV is by calculating the net present value of the average revenue (per time unit) of each customer on a given time period. The value of the bid should reflect that, such that:
Right now marketers can use a marketing suite such as Adobe Marketing Cloud, that has access to customer databases through direct SQL integration or a microservices approach (usually REST APIs). These suites also allow marketers to manage advertising campaign through several networks of publishers that can be integrated (such as AdWords) targeted to individual customers on the database.
In the mobile marketing context several things can be centrally controlled through these kinds of suites, including push messaging and other services. These require universal protocols to be integrated. AppCoins provides just that.
If AppCoins is one of those integrations we can run campaigns covering compliant app stores advertising placements targeted for individual users amongst other channels such as web and social. All it needs is a filter condition with an app identifier and a user identifier.
The user identifier we have seen already (GAID and/or IDFA). The app identifier is usually the Android application id or the Apple app id.
The machine learning model that we described already does the mapping assuming the GAID/IDFA is in our database. In practice its fairly easy to obtain these using the Google Play Services API or iAd Publisher API in Android and Apple environments respectively. AppCoins even allows you to include the definition of a specific device model, in case you want to target that user on a specific device (and app stores can match the user with different devices through login data).
This means the marketer may target that specific user on his VR device, or his Smart TV but not his Tablet or Smartphone. This is a complete customer journey targeting solution for mobile. The PoA system allows the marketer to confirm each individual conversion across platforms and channels.
With AppCoins, transaction confirmation can be done using the Raiden’s REST API, since thats the infrastructure that AppCoins will use for all its transactions offchain. It also ensures privacy of advertising campaign bid details and performance metrics.
Up until now, no universal protocol existed for cross app store advertising. AppCoins is currently the only solution for mobile advertising that supports cross platform programmatic CRM.
While Martech consolidation may not be happening any time soon, by creating open protocols for advertising we are getting one step closer to having universal systems integration across platforms. Instead of a large vendor suite we are heading towards several separate components of a digital marketing stack, that are beginning to speak common languages with each other.
AppCoins supports this universality, just as David Ogilvy envisioned 40 years ago.
Originally published at medium.com on November 10, 2017.
