Today’s marketing ecosystem is both complex and fluid. In order to effectively manage ad spending, advertisers need experienced media managers and accurate data. However, one often underutilized, valuable metric is return on advertising spend (ROAS). The formula for calculating ROAS is simple enough but the practical application of this data is the missing piece for many companies.
ROAS is the one thing every marketer needs to determine the effectiveness of your marketing dollars and should be at the top of your data-driven strategy. It gives every marketing team a common metric to measure, compare and optimize results.
What Works Best?
There are a few ways to calculate ROAS, and it is important to understand the nuances so you can determine what works best for your business model. They all work, but it is essential that the whole team use the same calculation for an apples to apples comparison.
The most common expression of ROAS is as a percentage on the return, using the formula spend/revenue times 100. So if you spent $20k and received $200k in revenue, your ROAS would be 1000%.
An alternative ROAS calculation is as a dollar amount. The calculation is revenue minus spend divided by spend, so with my above example, the 1000% ROAS would be $9.00 ($200k-$20k/$20k=$9). This means every $1 spent generated $9 in revenues.
There is also a calculation for the expression of ROAS in terms of cost of revenue. Using the data above, the ROAS would be $20k/$200k=10%. When considering this data across channels, it is easy to see what percentage of revenue is going towards advertising. The rest of this article uses this calculation as examples.
Once you have settled on a prefered ROAS calculation, you need to incorporate your business or marketing channel key performance indicators (KPIs). These can vary by brand, channel, or time period. Some of the most common KPIs for retailers are conversion rate, average order value, new vs. returning customers, discounts, and margin.
The first consideration is measuring both the gross and net ROAS. This means you need to factor in any returns and cancellations that will change your ROAS. In our example, if we had a 15% return rate our ROAS is now 11.76% resulting in poorer results than first calculated. It is important to measure both gross and net ROAS to see the effects of returns by channel.
Another valuable KPI to consider is new versus returning customers. You may be comfortable paying more for certain media to acquire a high percentage of new-to-file customers. What is your threshold for new vs returning? How do you value the loyalty/repurchase rate of customers?
The next data point to consider is the level of discounting applied per channel. If certain channels or marketing campaigns have large discounting in them, it is good to look at sales both before and after discount so you can understand how your ROAS changes due to discounts. You should consider the discounts as a cost of media. If you spend 10% on media and give the customer a 10% discount, you should view this as a 20% cost of sale instead of 10%.
Finally, analyze the ROAS by incorporating margins. This is probably the most difficult yet valuable analysis. When you look at cost of sales by channel and overlay product margins, you get a new and unique perspective that will help you better optimize your ROAS. The broader your margins, the more valuable this metric will be. There may be keywords, channels or specific affiliates that drive higher or lower margin products which will affect the ROAS results.
You can also use ROAS to determine your diminishing point of return for a given channel. You can’t keep throwing money at a marketing channel and expect to get more out of it. At some point, spending more will not get you more. Factor in all of your KPIs so you can more accurately determine your pain threshold on when to stop spending more for a given channel.
Once you have your KPIs incorporated into your ROAS analysis, you can then move on to click path analysis and attribution modeling. The first step to click path analysis is pulling in all of your media costs into a centralized reporting platform to see the performance of all media and how consumers interact across various marketing channels and the total costs related to each conversion.
This is where BIG DATA provides tremendous value. Collating all the important metrics mentioned above into a single source allows you to measure and optimize ROAS – giving you a competitive advantage.
Calculating ROAS by the most common paths to conversion gives you insights into the main ways customers typically touch your marketing efforts. To make things easier, try cutting off your analysis at 90% of your total revenues and spend so you aren’t including too many outliers.
If you’ve gotten this far, then you are ready to create a custom attribution model. There are no shortcuts to perfecting your analysis. You need to understand the underlying data above to develop an accurate attribution model that makes sense for your business or marketing channel. You’ll need at least 90 days of data to have enough to be statistically valid for building a model. In the meantime, focus on gathering the data mentioned above. If you later want a data point you neglected to include, you’ll have to wait another 90 days before you get to a statistically valid data set again. So it makes sense to get everything included at the beginning.
Building an attribution model must include your business KPIs in order for it to provide value. By understanding the click path data along with the costs and KPIs, you are then able to develop and test attribution models. You will need to determine the value of each touch point depending on where it is in the click path. So if paid search is last click, for example, you will change the value of that contribution in the click path for your attribution model.
The key is taking all of these steps in sequence and not trying to jump ahead or create shortcuts. As you put these pieces together, you learn more about the data and have a better understanding of their effects. Each step builds on the previous and is essential in the development of your ROAS calculation and attribution model.