Predicted Returns: Three metrics to measure paid acquisition performance in SaaS

For performance marketers, paid acquisition is an ever-changing jungle of opportunities and traps; a galaxy of ad networks, formats, channels & keywords, each with their own idiosyncrasies. What makes performance marketing so appealing is that every campaign, click, conversion & bid is trackable and analyzable. Tools like Google Analytics make it easy to comb through, filter, segment & visualize your campaigns, channels, costs & returns. Line your traffic sources by campaign up against your website engagement & conversion metrics and you can see how users behave after they click on each ad.

The problem with pay-per-click in B2B is choosing how to measure ROI. Looking at email generation optimizes for acquiring low-value emails, but the SaaS buyer journey may take weeks or months to convert to actual revenue, which is far too long and far too complex to measure. Performance marketers want to know which spend is yielding results (to double down) and which spend isn’t (to shut it off).

Generating few great leads with a high CPL yields much better results than generating many low-quality leads with a cheap CPL. SaaS marketers need an transaction metric that mimics eCommerce-styles transactions to measure their performance against, and of course we have just the solution.

Measure performance marketing with smarter metrics

The goal of this play is to make sure we are investing our ad dollars into channels that generated qualified pipeline. We want to see which campaigns return a positive ROI in the long-run (though we can cap this definition at, say, 12 months). All we need is MadKudu & Google Analytics (or any analytics solution that allows for segmentation by UTM tags) to get this done.

The play itself is quite basic. As always, at the core of modern marketing operations is leveraging historical customer data to build a model for what your best leads look like and do. We’re going to use MadKudu for this, since it’s easy to operationalize across the entire buyer journey and requires no in-house data scientists. You can read up on about the three stages of lead scoring to start building your own.

Once we’ve embedded MadKudu into our conversion points (lead generation forms, user signups, etc.) – potentially via MadKudu Fastlane – we’re going to send MadKudu data back to a website analytics tool like Google Analytics so that a visitor’s data can now includes its Madkudu score.

Once that’s set up, there are three new ways you can measure your performance:

Cost per Qualified Lead

If your PPC campaigns are pointing to a lead capture form (webinar, eBook, free tool), you can look at which channels & campaigns are bringing in qualified pipeline in an ROI positive way. Specifically, we want to compute the total spend on the campaign divided by the number of good vs. very good leads – some campaigns bring in many leads that don’t convert, while others bring in very good consistently qualified leads. Identify where your generating qualified pipeline efficiently might lead you to cut 50% of your paid ad spend (as it did for Drift).

Predicted Spend vs. Cost Per Lead

This may be more useful for campaigns that are leading to account creation or demo scheduling, where the rest of the buyer journey follows the traditional buyer journey. Here we’re leveraging MadKudu’s ability to map user’s not only to their likelihood to convert but to their predicted spend. This is calculated by looking at how much a lead resembles historical customers in each segment – “do they look like a pro plan?” With predicted spend, we can develop a threshold for cost per lead accordingly: we align our acceptable cost per lead (and therefore bidding threshold) based on the average predicted spend of leads acquired.

By averaging predicted spend across leads who are predicted to convert or not, we’ll be able to quickly optimize for campaigns that acquire any combination of lead values so long as we are not paying on average more than they are worth.

Predicted Value vs. Campaign Cost

The last way we can view performance marketing ROI is by leveraging the predicted value field that MadKudu provides, for example, to Facebook’s Ad Engine. Instead of feeding this data to Facebook to train its bidding engine to bid on our best leads, we can look at all channels of spend through the lens of this metric and answer the question: are the leads we’re acquiring predicted to spend more than what we’re paying for them?

We can look either at average predicted value vs. cost per qualified lead or we can look at the sum total of predicted value vs. the total campaign spend. In both cases, we’re getting a picture of whether the leads we’re acquiring today will be profitable down the line.

Since Predicted Value is designed by looking at predicted spend by plan multiplied by % chance of conversion, we’re accepting a 12 month CAC. We can adjust this either by calculating predicted spend as 6 Months of MRR instead of the plan sticker price (12 months of fully loaded MRR), or by dividing our predicted value by a variable as a function of the CAC threshold we want to set.

The Impact: -50% in Ad Spend

No matter which metric we optimize for – predicted value, predicted spend or cost per qualified lead – we’re arming ourselves with metrics that we can measure in real-time as our campaigns generate leads. Drift integrated MadKudu into Google Analytics and identified campaigns that were generating no qualified leads (although potentially many qualified leads). As a result, overnight they cut 50% of their ad spend and saw a major dip to their website traffic but no meaningful dip in their pipeline creation.

That 50% in recuperated Ad Spend can go into doubling down on existing campaign or creating new campaigns – either way, we’re increasing our performance marketing ROI.