User acquisition can be a minefield, across a myriad of ad networks and traffic sources, middlemen and scammers, making it incredibly hard to reach the valuable users you seek. It is challenging to operate your marketing budget with the necessary attention and granular optimizations without mobilizing hundred-strong UA teams.
One increasingly popular option to enable this is automation, which is being successfully leveraged by a growing number of UA teams across the globe to power ad campaigns through the analysis & interpretation of big data by an intelligent machine. Or a not-so-dumb one at that.
What is Automation?
We have all been acquainted to automation for a number of years, in one form or another. Take your email junk filter for example: it works day and night looking for clues suggesting that a message may be unsolicited or malicious. It learns and refines itself over time, based on your behavior and that of other users.
In broad terms, automation is the systematic use of algorithms for a machine to interpret live inputs and autonomously respond with adaptive behavior, within a feedback loop where the outcome of each action becomes the input for further analyses and corrections.
While some level of automation may be actionable through a rule-based approach (whereby it is a human that defines punctual conditions upon which certain actions must be triggered), the real value of automation indeed emerges when artificial intelligence is employed to supply self-learning capabilities, allowing the system to discover patterns in historical and live data and autonomously identify a suitable course of action. AI’s most distinctive feature as opposed to rule-based systems is the ability to learn and adapt.
Whether it’s about blocking malicious emails or choosing the next move in a chess game, AI is meant to treat individual situations in relation to the ever-changing big picture. Analyzing the ‘here & now’ within the context of the transforming environment and other actors’ behavior within: like in nature, any prediction on live phenomena needs to consider evolutionary trends rather than limiting to the current static state.
The Challenges of User Acquisition
Digital marketing, within which mobile marketing has been claiming an increasingly dominant spot, is an ideal playground for this paradigm. With millions of data signals being generated at multiple levels (per advertising network, location, campaign, publisher, app genre, device type, etc.) at any given second, and multiple ways of attributing meaning to such signals. Whether the purpose is to track ad spend, install volume, users’ lifetime value (LTV), percentage of paying users (PPU), return on ad spend (ROAS) or any other metric that is relevant to a mobile business’ success. Data needs to be collected, consumed, and interpreted with different lenses, across all the relevant levels, to identify sensible corrective actions.
Your very performance targets need to be set and continually adjusted for each of your apps depending on intrinsic and extrinsic factors — such as the app’s life cycle, its monetization model, seasonality, target geographies, bidding methods, ad creatives, and a lot more.
Once you have figured all of that out, you may finally enact your desired corrections and adjustments across your multiple advertising channels, which naturally come with a broad palette of campaign management workflows and dashboards, and proprietary internal algorithms requiring specific optimization strategies.
Across such an intricate mass of intertwined variables, an extensive data-driven approach, multivariate analyses, and continuous adjustments across a number of dimensions become necessary to successfully address what we consider the three main challenges of user acquisition: exploration, growth, and profitability.
While there is no magic wand to do it all for you, automation technology does provide ways to enforce a solid and consistent framework where data is processed systematically at all levels and sensible corrections are made responsively across the board. Machine learning supplies the human-like thinking magic that makes those continuous tweaks sensible and adaptive to unforeseeable combinations of contextual factors — evaluating every ‘unit’ (a campaign, publisher, creative, or geography) based on its performance’s evolution over time (and related uptrends or downtrends) rather than today’s static snapshot.
Humans can do fairly well at managing mobile marketing’s complexity, as professional teams across the globe have demonstrated over the past decade, yet that may not be the most efficient use of their time. What if a machine could do the heavy lifting of this continuous analysis→correction loop, across all levels and dimensions, letting humans in charge of high-level steering through a system of levers and triggers? What if a machine could ensure that learnings are preserved and combined over time across the organization, rather than getting lost when a campaign manager leaves the team? What if that same machine could enact thousands of actions per second across millions of ‘peculiarities’ (unique instances of a specific publisher from a specific genre in a specific geo on a specific network with a specific ad creative…)?
Inside the Black Box
Offloading hard work to a machine sounds inviting, but the idea of handing that scarcely achieved, fragile balance of Spend and ROAS to a robot may be daunting, too. It may take time (and data) to trust that programmatic algorithms are effectively working at your service to make the most of your marketing dollars.
The good news is, you don’t need to take a blind leap of faith. Borrowing words from Michael E. Berger, automating is about delegating rather than abdicating. A reliable automation platform provides the transparency needed to build confidence by visualizing what happens under the hood and tie any auto-correction back to the context that raised the need for it. That’s where reporting comes into play: without aiming to replace MMPs or in-house BI systems, which are best suited for in-depth analytics, your automation platform needs to provide streamlined reporting about your key spend and performance metrics and about the automation engine’s own behavior, in a way that enables fast but well informed human steering. A reliable automation platform doesn’t require you to close your eyes and trust in a black-box magical AI, it rather puts automation powers in your hands to inform your desired strategy, have it implemented around the clock and monitor its success in real-time.
The perfect automation solution provides visibility of where your dollars are going, and why. While in the background, it relieves human resources from repetitive, focus-demanding, and immensely time-consuming analyses to identify punctual issues and respond at speed.
Beyond that, since user acquisition presents different challenges for different apps at different stages of their life cycle on different channels, effective automation comes with a great deal of customization. So while we all accept that there is no blueprint for the ultimate optimization strategy to deliver monetization and engagement, your automation partner needs to leave enough room for you to formulate the punctual challenges you are facing, set your goals, and selectively activate the automation features that you feel the need for.
Predictions? Yes, And…
Nowadays when hearing about UA Automation, most people think LTV Prediction: how to forecast the future value of a ‘piece of traffic’ (a user? a publisher? a campaign?) based on early performance indicators, so as to steer the media buy strategy as early as possible and achieve long term profitability?
That’s indeed an important part of automation: your Day-3 or Day-7 ROAS goal should always be chosen as an actionable early proxy for future profitability — which requires you to walk backwards from the predicted LTV (or pLTV for short). And while LTV is often plotted as a logarithmic curve, that may not always be the case — depending on your app’s actual ARPU progression given its monetization model (ad revenue? in-app purchase? subscription?), the pricing model you are buying with (CPI? CPE? CPM?), the specific creative format you are using (banner? video? playable? hybrid?). Predicting future LTV based on early clues becomes a science that once more needs to encompass a variety of dimensions to produce one magic number. A science that machines are well suited to master.
To put things in perspective, however, predicting LTV is only a piece of the automation puzzle, a rather small one at that. Whilst working with a specialized AI partner can help perfect forecasting accuracy, more and more app publishers have become increasingly good at formulating accurate estimates in house, through proprietary data and deep domain know-how. Using those estimates within a more holistic framework that systematically collects and analyzes real-time data from a broad range of sources and formats, processes it iteratively, and intelligently deploys your budget 24/7 across channels according to velocity, marginality optimization, and stop-loss mechanisms… that is a different job entirely. It is the job of a UA Automation System.
To be of actual help, your pLTV needs to be orchestrated with several other components of the user acquisition science — like budget pacing, eCPM deconstruction, risk management, outlier normalization, cross channel budget allocation, creative A/B testing, campaign monitoring and alerting, and traffic exploration algorithms to optimize those traffic sources (normally the majority of them) that simply haven’t generated enough data to legitimate a reliable LTV forecast. A UA Automation Platform is meant to do all that, while allowing you to customize your desired strategy according to punctual needs.
As the marketing industry evolves, and everyone races to refine processes and maximize marketing returns, UA Automation is not only a helpful enhancement, it becomes increasingly necessary to maintain a competitive edge. If you don’t have some sort of automation pipeline in place, leveraging AI to suggest (or directly take) actions for you across the board, important bits of information will likely be missed rather than used to unlock new ‘pockets of value’, inform future strategies or pivot existing campaigns.
It’s not about whether AI can support your marketing efforts or not, it’s about how: while the human component remains (thankfully) indispensable, effective AI-based automation needs to provide those very humans with levers and triggers to work in symbiosis with the algorithms, instructing their own strategy and maintaining tactical control, while accessing an arsenal of scientific weapons for all of their diverse growth missions.
Bubbleye’s Kraken is the world’s first and most comprehensive UA Automation platform for mobile marketers. It features holistic real-time reporting, multiple algorithms to address specific pain points (LTV prediction, budget allocation, bid adjustments, traffic exploration, custom alerting, fraud blocking, A/B testing, and a lot more), and a console to fully customize your desired balance between automation and hands-on control.
If you are interested in harnessing the power of AI for your mobile marketing and propel your apps’ growth, don’t hesitate to book a free demo.