Leveraging Cohort Maturity to Improve ROAS by 40% | WeDiscover Case Study

Alex Davey, the Article Author - WeDiscover - Paid Search Marketing Agency London
Alex Davey
16 May 2022 - 10 min read
Facebook Icon - WeDiscover - Paid Search Marketing Agency London
Twitter Icon - WeDiscover - Paid Search Marketing Agency London
LinkedIn Icon - WeDiscover - Paid Search Marketing Agency London
Email Icon - WeDiscover - Paid Search Marketing Agency London

Introduction

WeDiscover and Adelante Shoe Co. partnered on Paid Search in mid-2021. Adelante Shoe Co. are based in the USA and create made-to-order shoes that are fully customizable while focusing on their ethical and social impact through initiatives such as ensuring their craftsmen earn more than double the local average wage.

Due to the custom nature of their product, purchase journeys are typically longer than other ecommerce brands, leading to a significant conversion lag in Google Ads

This in turn causes complications with accurate and timely optimisations and reporting.

Challenge

Conversion lag is a term used frequently in the world of paid media. 

For a lot of consumers, purchasing a product or service is not a decision made in a matter of minutes but rather over days or weeks

Conversion lag is used to describe the time between the click and the attributed conversion. This could be minutes, days, weeks or even years. 

Conversions in Google Ads will continue to be attributed to the day of click in Google Ads for up to 30 days as standard, though can be extended or shortened as required, after the click. 

While it is great we are seeing the impact of paid media on those longer conversion journeys, it leads to complexities in optimising and reporting to the most recent performance data

Even with attribution analysis and non-last click modelling becoming widely adopted within Google Ads, an issue that continues to persist is the impact of long conversion lags leading to incomplete data and therefore low confidence decision making. 

After partnering with Adelante Shoe Co., we quickly noticed that conversion lag was going to be a complexity for Google Ads optimisation

One of the main benefits of digital media over traditional advertising is the access to fast and reliable reporting. 

If we report on weekly Google Ads performance on a Monday, then a portion of our conversions will yet to have occurred and we would be likely to underreport our performance by valuing complete spend data against incomplete conversion data

Google themselves are aware of this issue have introduced some basic Conversion Lag reporting to inform advertisers when a conversion delay may be affecting reporting:


For Adelante Shoe Co.,  viewing data for the last 30 days shows we are only seeing 90% of expected conversions meaning our cost to conversion ratios (CPA/ROAS/ERS) will be under-reporting. 

Google’s recommendation is to either “check back later” or select an earlier date range. 

The clear disadvantage here is that we want to be making decisions and reporting on the most accurate and recent conversion data, rather than relying on performance from almost 2 weeks ago.

While Google’s insight is useful to illustrate the issue, it isn’t very actionable and often raises more questions than it answers. 

Simply not reporting on performance for 12 days after the click would not suffice. We also wanted to avoid reporting anecdotally by guessing at how performance would look once the conversion lag had reduced.


We therefore wanted to create a data-driven approach to accurate reporting in spite of the conversion lag experienced on the account. This would allow us to make the best optimisation decisions within the account with the most reliable data.


The Solution

Data Analysis & Conversion Maturity

Since Google Ads would automatically attribute conversions to the day of click, we needed to create a solution that would sit outside the Google Ads ecosystem and allow us to run analysis. 

WeDiscover decided to explore cohort analysis based on Adelante’s conversion data to better understand the path to maturity seen within the account.


The first step was to create a daily snapshot report to export data so we could analyse data at a cohort level. Using Google’s Big Query, Sheets and Data Studio allowed us to automate, analyse and visualise this data:

This export will therefore present the maturity of conversions over time. Once significant data has accumulated we can plot the volume of conversions against cohort age.

The above chart is visualising all cohorts we recorded during our initial analysis and then how conversions develop. 

Conversions recorded on day of click, regardless of when they occurred from a calendar perspective, will be recorded on Cohort Age 0. Cohort Age 1 would be the day after click and so on. 

It is clear that the vast majority of conversions occur within the first week of acquisition and that there are notable spikes in days 0-3. 


Another way of viewing this data is as a maturity curve, where we see the cohort age and a % of total conversions:

This representation allows us to start extrapolating conclusions from the data

We have also ensured that we include confidence intervals so any conclusions and recommendations based on this data are statistically sound. 


From this data we can confidently conclude that:

The data here is noisy until day 2, where our future estimations become more reliable. 

Using this data we can recommend we only review performance after 2 full days have elapsed, much more palatable than the 12 day delay we were seeing in Google Ads. 


We can also use this data to extrapolate predicted mature conversion metrics within 2 days of a click, for example:

Visualisation & Utilisation

Now we have the data available, we needed to create a method of utilising this data in our Paid Search optimisation and reporting. 

Using the above conclusions we are able to use ongoing daily snapshots and historical data to create reports using multiple conversion metrics:

Reported Conversions - conversions metrics reported like for platform data. This will likely underreport recent days due to the conversion lag. 

Estimated Conversions (Avg) - using the average conversion lags to forecast the fully mature cohort’s conversions.

Estimated Conversions (Lower) - the conservatve estimate of mature conversions.

Estimated Conversions (Upper) - the upper estimate of mature conversions.

These data points, and associated data such as CPA, Revenue and ROAS, were then visualised using Data Studio, producing a report for both the account team and the marketing team at Adelante Shoe Co. to have an accurate prediction of recent performance.

Performance can now easily be visualised as “Reported” and “Estimated” and we can easily switch between these views through a simple drop down menu in Data Studio, as shown in the below GIF:


The Results

Primarily this solution allows confident and timely optimisation and budgeting decisions.

Generally, we work towards a ROAS target with Adelante Shoe Co. and therefore having an accurate representation of this metric is of paramount importance.

Since the implementation of the report month on month revenue has increased 30% with ROAS increasing 40%. This report has become the primary source of optimisation with the team making decisions on forecasted efficiency metrics. 

This solution has allowed us to scale performance by avoiding making short-sighted decisions based on incomplete metrics.

For example, a natural reaction to seeing ROAS 20% under target for the previous 7 days would be to decrease budgets and bids to reduce spend. However, with our solution, we could see that the estimated ROAS is actually 5% over target meaning that we may want to be more aggressive. 


Increasing confidence in paid media spend has been instrumental in the development of the Adelante paid search strategy,  as Nick Fioretti of Adelante Shoe Co. explains: 

“Accurate measurement and attribution is so crucial for a young company like Adelante. Our primary focus is efficient growth that can lead us to profitability. This can be very difficult to achieve when you have incomplete metrics/data leading your decision making. WeDiscover’s solution has allowed us to get a complete grasp on Google Ads attribution by fully understanding the conversion lag time and therefore understanding the full impact of our PPC efforts. The confidence levels in our short-term and long-term decisions have been reinforced by the performance we have achieved since implementing this solution.” - Nick Fioretti, Associate Director - Performance Marketing, Adelante Shoe Co. 


This solution is also scalable to a range of clients with conversion lag affecting almost all advertisers on Google Ads in varying forms of significance. 

If you have any questions regarding this solution or case study, please feel free to contact us.

Case Study
Facebook Icon - WeDiscover - Paid Search Marketing Agency London
Twitter Icon - WeDiscover - Paid Search Marketing Agency London
LinkedIn Icon - WeDiscover - Paid Search Marketing Agency London
Email Icon - WeDiscover - Paid Search Marketing Agency London

Enjoying reading?

Sign up to our newsletter and be the first to know when we release new articles.