AI is powerful, but quite literally has a “mind” of its own. It will make decisions without consulting you, or even letting you know that something has changed. Machine learning-based bidding algorithms present one of the biggest potential opportunities for data-savvy marketers today and should absolutely be utilized, but they must be carefully managed. Here are 5 essential elements to effective value-based bidding with network AI bidders.
Master your inputs. Figure out what the algorithms are designed to do, and adjust your inputs accordingly.
Practice good data hygiene. Make sure the conversion values you're feeding in are based on sound data. Especially with AI: it's garbage in, garbage out - make sure your data is representative of actual performance, or the AI will be all too happy to spend your money on low-value audiences, based on the signals you're providing.
Develop a good projection model. Unless your business realizes revenue within 24 hours of your users' first interaction with your site, projecting your conversion values will be a critical part of impactful AI-based bidding. Find the balance between accurate values and current data, knowing the networks will weigh the most recent data most heavily, and largely don't care about value generated from clicks even a few weeks ago. As a general rule of thumb, the higher-value a customer is, the longer it takes them to turn into value (imagine the decision process for a customer considering a $2.99 app download compared to a $250K/yr SaaS contract). Especially if you have high-value customers or anything other than an instantaneous sales cycle, get an effectively granular projection model in place before getting into AI-based bidding, and continuously recalibrate.
Set good targets. Even if you provide the networks with perfect data, there will be some expected error due to latency, attribution differences, and user behavior. Make sure you set targets based on the financial outcomes you want, combined with the feedback the network gives you on what it’s understanding from the inputs.
Invest in visibility. The algorithms, competitive landscape, and your own data will continue changing over time. Powerful as they are, AI bidders do not fall into the “set it and forget it” category of tools; because they can change without warning, it’s on us to stay on top of performance. Developing this visibility as the bidders, and our understanding of them, evolve is critical to identifying opportunities and minimizing risk.
Automate your processing. Small variations in process scheduling, for example, can have profound impacts on how effectively AI algorithms work. To best minimize human-influenced data fluctuations, make your interface with the machine as automated as possible.
Stay current with the ad networks. In some cases, we’ve found that we notice symptoms of a change to bidding algorithms before our network partners are aware of a change having occurred. Having partners at the ad networks helps you stay informed and able to adjust strategy rapidly. We've also found our network partners incredibly grateful to receive useful feedback on their products, so they can continue increasing the value of those products for those customers (our clients).
Practice data-driven agile prioritization. Keep up with the rapid changes by maintaining a consistent view on performance. Use those insights to then adjust prioritization every day, so you don’t get caught flat-footed, either missing customer acquisition opportunities or wasting precious budget on low-value audiences.
Working Planet has been utilizing "value-based bidding" since our inception. We have built our business on a system of profit-centric feedback loops and bidding models, with the goal of maximizing contribution margin (post-marketing gross profit) for each of our clients. Now that the ad networks are catching up, and adding in their own vast data networks and machine learning algorithms, value-based bidding is more powerful than ever. Getting ahead of the curve on effective AI bidder utilization will be a competitive necessity that continues to pay dividends with each major evolution of the digital marketing landscape.