How Algorithms Can Solve The Attribution Problem

Mapendo Team
May 27, 2022
How Algorithms Can Solve The Attribution Problem

We live surrounded by algorithms, and as the world, and internet, have grown exponentially in the past few years, these types of machine learning processes have become common for organizations of all kinds. But, what is an algorithm? An algorithm is a procedure, or formula, for solving a problem, based on conducting a sequence of specified actions. In simpler words, you can think of an algorithm as a recipe for preparing a meal. The algorithm is the specific instructions to follow in order to cook that meal. Algorithms are widely used throughout all areas of IT (information technology). A search engine algorithm, for example, takes search strings of keywords and operators as input, searches its associated database for relevant web pages, and returns results.

Why algorithms are more useful than ever in the new iOS 14.5 scenario

Algorithms are starting to be used in performance marketing. They are gaining value because they are able to optimize advertising campaigns from a budget point of view. This is particularly significant after the last Apple iOS update which has had a great impact on advertising strategies. The iOS 14.5 update is now giving Apple users the option to opt-out of data sharing, for example, every time they download an app. Some reports have suggested that global consent opt-in rates could sit as low as 5%. While others claim that the situation is significantly better, with the e-commerce opt-in rate hovering at an average of 17%.

In any case, the game has changed. Advertisers used to rely on data sharing to gather data and derive ad targeting decisions for app-install campaigns, and more. Privacy is a key feature of this Apple update, therefore the optimization of the ROI of user acquisition campaigns, through precise user profiling, has become even harder. Thus, with the implementation of SKAdNetwork, the attribution model has shifted.

The deterministic attribution has given way to the probabilistic one. With this model, the data you can collect to predict your LTV users are limited and less fragmentary. They might include a user’s IP address, operating system, mobile hardware, web browser. It must also be said that the data gathered, such as the user’s IP address, change much more frequently than a device identifier, which means that probabilistic attribution has a shorter attribution lookback period than the deterministic one. Thus, all these changes impose a rethink of performing mobile advertising, and to face the new challenges of paid mobile user acquisition.

The impact of algorithms on mobile advertising

Algorithms take a complex set of data and use it to find patterns, or weigh evidence, on an objective scale, ultimately producing a suggested action. As we suggested above, without a right attribution, marketers may incorrectly assume a channel or campaign was ineffective when it actually worked great, or they may give too much credit to a touchpoint that had little to no effect on the end purchase decision/on the final conversion

An innovative way to use algorithms is in mobile programmatic advertising through DSPs platforms. By connecting data and creatives, marketers can tie ads into larger networks that provide more holistic insights, with actionable patterns, and obtain a greater ROAS. With programmatic, marketers can expand the reach of their messages without spending too much money, improving their ROI and also the campaign value, in terms of quality of traffic and leads.

Especially on Facebook or Google, scaling is harder, as volumes get bigger, ad costs get way way higher, while with programmatic the cost stays affordable. In a nutshell, algorithms can indicate the strategy for programmatic media buys, selecting the audiences with the highest value and determining the best time and place to display an ad. The algorithm, during real-time bidding, limits purchases to those impressions that have a higher probability to generate post-install events, and thus increase the conversion rate.

The consequences

Algorithms help inform marketing decisions, but can also help determine which features get prioritized in the product roadmap. Marketers will gain a new level of customer insight that will enable them to use their data proactively instead of reactively. Rather than asking machines to measure clicks, swipes, likes, etc… they should be asking the “tough” questions, such as “Why did customers engage with my products and services in the past?” and “Where will I find my customers and prospects in the future?”. When they do, the algorithms will crunch millions of data points to identify patterns that artificial intelligence will eventually forecast to more accurate and predictable outcomes.

This will have a tremendous impact on the media ecosystem and the environments where customers are bought and sold. Firstly, transparency will be table stakes to buy and sell advertising. That’s because the machines will require a clear, straightforward view of all inputs and outputs. With a better understanding of how the money is spent and its impact on their cash registers, marketers will drive media sellers and their agencies to quantify their value beyond just media and technical metrics. Marketing outcomes will matter more than ever. Secondly, marketers will require more and higher quality data services and integrations from their media partners to mutually justify their advertising dollars. Lastly, marketers will need to build and grow their data analytics and science teams to measure and synthesize the return on questions about the cost to acquire and retain customers. By asking the right questions, machine-powered, predictable customer journeys will allow marketers to look past technical metrics and focus on business objectives.


Here at Mapendo, Jenga, our AI technology, which leverages a proprietary algorithm, based on reinforcement learning, collects tens of thousands of data related to a given topic, finds patterns, and it manages to predict the possible outcome of a marketing campaign and finds the audience that is most likely to convert for a type of ad.

Our algorithm has been trained to optimize the traffic according to the client’s KPIs, maximize user retention and generate post install actions. Advertisers need to leverage technology to find meaningful insights, predict outcomes and maximise the efficiency of their investment, by choosing the right channels and budget.