tips and tricks
troubleshooting campaigns with the leaky pipe
thesis
By deploying a leaky funnel analysis when troubleshooting campaign performance, you can ensure you – or your paid digital executors – are focusing efforts to find the cause of problems and delivering specific actionable solutions.
focusing analysis by visualizing the pipe
In this context, the pipe is a synonym for the user journey. It is the journey that a user takes from the time they click on your link to the conversion event that is your primary objective.
What’s important to understand about this pipeline is that different metrics are available to be optimized at different stages: metrics that are earlier in the pipeline typically influence those that are later in the pipeline. For example, a key metric that is often optimized towards as a business objective is Cost per Acquisition (CPA). CPA going up may be a cause of concern, but rising CPAs never implicates a specific action or set of actions because CPA has several inputs that can be influential. This can be illustrated by the different ways in which CPA can be calculated.
Total Spend / Total Conversions = Cost Per Acquisition/Conversion
This is the main way that CPA is calculated, but this method gives the least information about its other inputs. It cannot be used to triage campaign performance.
Cost Per Click (CPC) * Clicks * Conversion Rate (CVR) = Cost Per AcquisitionThis is another way to calculate CPA and is far more useful for identifying potential causes of campaign performance fluctuations. Performing analysis with this model allows us to see which metrics are causing performance shifts.
Within this, we can see how leading variables like CPC and Conversion Rate influence the lagging variable which is CPA in this case.
- If the CPC goes up, then CPA will go up even if your conversion rate stays the same.
- Similarly, if your conversion rate goes down, your CPA will go up even if your CPC remains the same.
This may seem trivial or simplistic, but it is crucially important to methodically troubleshooting performance. An executor that doesn’t break lagging metrics down into the leading metrics that influence them may not have an understanding of these how these metrics relate to one another and which they have control over. After all, we can’t just “make the ROAS go up,” we have to make tweaks that positively influence ROAS, which is where the leading metrics come into play.
Following are some examples of action items that I would consider depending on which variable moved and how it moved.
If CPC changes
- Pull reports on auction insights to determine if competitors could be driving up auction costs.
- Pull a search term analysis to see if search terms skewed towards higher/lower cost queries
- Pull demographics reports to see if gender, income, or age demographics may have changed period over period. (different demographics have different average costs)
- Troubleshoot ad quality & ad relevance, which may impact what CPCs search engines allow you to pay. Consider revising:
- Ad Copy
- Landing Pages
if CVR changes
- analyze landing page metrics that affect the user experience and thus conversion rates
- Bounce Rate (which is a lagging indicator in a different sense)
- Load Speeds
- Layout changes
- Pull page paths and behavioral reports. Perhaps users are engaged but they’re exiting your conversion path in search of more or specific information.
- Pull demographics reports to see if gender, income, or age demographics may have changed period over period. (Different demographics may engage or convert differently.)
- Consider AB testing landing pages to determine if landing page qualities could improve CVR.
As you can see, the metrics to look into and fix are fairly different. The fact that the action items are so different is exactly what exemplifies the need for an analytical method that points executors in the proper direction for isolating problems and finding opportunity.
Actionable takeaways
- When tracking changes in campaign performance, don’t focus on KPIs like CPA, ROI/ROAS, etc. Focus on the leading metrics that influence the down funnel metrics that determine performance.
- Develop hypothesis that drive testing strategy. Don’t just pull every lever at once. Develop a set strategy for achieving the end objective and test different tactics in sequence.