Case 1: Driving better CPA without sacrificing lead volume.
Thinking about spending more on Google Ads but your not sure what the return will yield? Budget Optimize has been designed by Google Ads veterans to answer this question. Using Machine Learning techniques we build a model to predict how changes in investment will impact performance. Below are two case studies how businesses are using Budget Optimize to tweak their performance.
Driving better CPA without sacrificing lead volume.
A leading player in the accounting space was generating 100’s of leads per month. They wanted to see if they could bring down their Cost Per Acquisition (CPA) without sacrificing lead volume. On average over the past 9 months (November 2018-April 2019) the business had the following performance:
We wanted to understand what would happen to CPA if we increased spend by $1200 in the following Month (May 2019) Using Budget Optimize we generated a model of past performance over this 9 month period. The model builds a regression line to map the cost & conversions for each day of the period.
With our model we can then map cost to CPA and we are able to forecast this extra $1200 spend in the month ($39 / day). Our model predicted an increase in close to 90 leads in the month and an actual CPA drop to just under $93. The table below compares our predictions to actual results:
The predictions rows shows the conversions and subsequent ROI that Budget Optimize predicted for the month of May. The ‘Actual’ row are the actual results for May we recorded after May elapsed. We were pleasantly surprised at how close predictions made were to the actual numbers. This shows that the regression model was able to predict extremely closely what would happen when we increased our budget for the month by $1200
Case 2: Predicting impact on ROI at higher spends.
Predicting impact on ROI at higher spends
A startup in the financial services industry generates the vast majority of sales via their website. They were looking to understand how ROI will be impacted with an increased investment of $1,000 / month on Google Ads. In order to predict change in ROI with this increased investment we modelled cost vs revenue over a period of 5 months. The first steps was to split the campaigns into branded and non-branded categories. The business was already maximising their brand investment and therefore we knew growth would come via non-branded campaigns. We wanted to understand the change in ROI as we increased spend on non-branded campaigns only. The existing performance over 5 months was:
Budget Optimize then allows us to pick out only non-branded campaigns from the left hand menu, which we can visualise performance on. In the below graph we map Cost vs ROI for non-branded campaigns which allows us to visualise the diminishing returns. We can then find our tolerance for ROI and ensure we don’t spend over the mark.
We wanted to predict what would happen if we increased spend by $1,000 / month ($32/day) The following table shows the results in the following month. We have predictions for:
- Brand: we assume the same as previous months & no growth
- Non-brand: Budget Optimize models the change in revenue with the expected $32 increase
- Total: This is the sum of brand & non-brand. This is the final prediction.
- Actual: These are the actual results after the month elapsed
We see that Budget Optimize predicted a total revenue figure of $6,321, which was close to the actual amount of revenue of $6,314. The predicted ROI subsequently was also very close. Overall we can say that when increasing spend by $1,000 for the month, the revenue decreased by about $175 / day and subsequently ROI dropped from 4.81 to 4.54. This was closely predicted by Budget Optimize. It’s clear from this that increasing budget on non-branded campaigns in this case was not an ideal outcome, it was predicted that less revenue and a lower ROI would be generated.
Budget Optimize largely makes its predictions on one key feature and that is spend. Spend is in our opinion the most important feature in predicting future performance. But it is not the only one. Other factors like seasonality or supporting awareness campaigns (like TV) or change in strategy etc.. could all affect performance. You can account for some of this in Budget Optimize. For example, if spend increase is going to video campaigns, you can exclude video from your analysis. Or you can set the historical dates to take into account the previous season only year on year. It’s also important to consider which campaigns the increasing spend is allocated to. The above case studies consider account wide increases. But you might just want to focus on 1 campaign for more accurate analysis.
Why is this important?
If you’ve read up to here and your not convinced yet then I have a few more things to say. As a business how do you currently predict or evaluate whether you should increase investment in Google Ads? We don’t believe Budget Optimize will always be super accurate because there are many factors still unaccounted for. But, it is important to make informed decisions based on data. With Budget Optimize you can run advanced regression models and make these predictions in seconds. There is no tool out there (until now) that can provide you a reliable model of what is actually going to happen to your campaigns as you adjust your campaign budgets. In some of my discussions with Google, they often provided me predictions simply based on existing conversion rates! They didn’t account for diminishing returns or campaign types. This modelling, while not perfect, is much closer to the reality. Digital advertising is dynamic. You should be pushing boundaries and looking for ways to constantly find performance increases. Even if they are incremental. Budget Optimize allows you to understand where you can take your account and the potential that lies ahead!