October 8, 2021
The Ideal QA Process in Digital Advertising
If you’re a business that focuses on Cost per Acquisition, you may know what it’s like to feel limited to tCPA bidding. However, the only limitation you face is how creative you’re willing to get with your auto-bidding set-up. As the consumer journey continues to lengthen along more and more touchpoints, bidding proactively towards customer lifetime value may be just the thing to level up your approach to today’s automated bidding landscape.
At Metric Theory, an increasing number of CPA-driven clients are testing into Target ROAS bidding. How is revenue tracked, you ask? Technically, it isn’t. We insert placebo-style static valuation into the conversion pixel in order to trick the algorithm into believing we are capturing revenue and thus rendering us eligible for tROAS testing. We determine a dollar amount we feel is within reasonable range of LTV, and then alter our conversion pixel and to assign every incoming conversion this dollar amount.
A crucial benefit to running a value-based bid strategy is the ability to incorporate multiple valuations across a variety of conversion types. While the final transaction remains top priority, valuing conversion actions that precede the final purchase will allow you to better spread your bids across a more holistic understanding of the customer journey. Pivoting toward this approach has allowed us to better scale our mid-funnel initiatives such as Non-Brand ads across Search and Shopping to inform the algorithm with more full-funnel data on which to base our bids.
Imagine you’re an ecommerce subscription brand looking to generate new sign ups. More likely than not, users are engaging with your business multiple times prior to committing to a monthly charge. For example, you may know from your current data that 1 of every 5 users who complete your on-site Quiz end up converting later as subscribers. The set-up below illustrates how you can assign value to micro-conversions while working backwards from your LTV:
|Value per Subscription (LTV)||Quiz Completion to Subscription Rate||Working Ratio for Determining Value||Value of a Quiz Completion|
|$150||20%||5:1||$150 / 5 = $30|
For businesses who haven’t yet quite nailed down an exact customer LTV, keep in mind that the ratio is more important than the accuracy of the value. If I’m bidding $5 for a micro conversion and $20 (1:5) for my main conversion, this set-up is fundamentally identical to $20 vs. $100 (1:5). The key here is understanding that it’s not about the number of dollars, it’s about the comparative valuation. All you’re trying to do is tell the algorithm the degree to which the final conversion event in the customer journey is the most valuable.
Continuing to draw from the example above, the next step in our process would be to estimate our ROAS based on correlation to CPA. To do this, you would first set up some custom columns multiplying your Subscription conversion action x 150 and Quiz Completion action to $30. Your total revenue would look like the below:
|Quiz Completions||Subscriptions||Quiz Completion “Revenue”||Subscription “Revenue”||Total “Revenue”|
|14||3||14 x $30 = $420||3 x $150 = $450||$870|
Once you know your total placebo revenue, you can use this to calculate what your ROAS would net out to be. In the case below, you may decide to launch your tROAS strategy with a 290% as your initial ROAS target, knowing that this correlates to a $100 CPA.
|CPA Target||Ad Cost||“ROAS”||Actual Cost per Subscription|
|$100||$300||$870/$300 = $2.90||$300 / 3 = $100|
At Metric Theory, we took the outline above and applied it to a CPA-driven client who wanted to scale up conversion volume while maintaining the strongest possible CPA. For this setup we incorporated both Leads and Transactions into our total revenue valuation. Below is the year- over-year data showing that scaling spend on tCPA led to conversion & efficiency loss while growing the program with tROAS led to 1.5x stronger volume at a stronger overall CPA.
|Bid Strategy||CPA YoY||Spend YoY||Conversions YoY|
Remember, you aren’t actually driving toward a different KPI or changing your reporting in any real way – you’re simply tricking the algorithm to view the points along your customer journey the same way you do. That being said, given that you are not bidding off actual business data but rather an assumption, here are a few tips when determining your valuation for a conversion action:
When it comes to auto-bidding, there’s no finish line. Bidding algorithms continue to improve year after year which means you should be testing continually and rigorously while avoiding opting for the status quo. Ultimately when it comes down to it, you’re only as limited as your creativity will take you.