ROAS (Return On Ad Spend) is the true and only king!

In the past user acquisition teams were relying on metrics such as conversion rates (from a visitor / non customer to a buyer) or overall traffic. These days are gone. Also, in hindsight it can be deceptive. For instance, it may be more profitable to focus on identifying an individual huge customer, rather than 1000 small ones, or even 100,000 ‘empty’ users who have visited your digital product/ downloaded the app but have not yet paid a single cent (and therefore won’t pay you back forever).

If someone argues (probably representing a bot company) that traffic will eventually result in generating revenue through the viral effect, we would respond by stating a simple fact: just as a friend refers similar friends, a revenue generator will attract more revenue generators. Conversely, empty installs may lead to more empty installs…

At a glance, on paid marketing (no matter what is the source or which lair) , we would like to make sure on each dollar we will invest:

• It will mature into ROAS positive eventually

• We will get a clear recoup (payback) window

• Conclude this asap so if the 2 paragraphs above wont work we will be able to stop the spend in this route and revise it into something more profitable

• We will be able to drill down as much as possible while maintaining a workable margins of error

We can use historical multipliers as the 1st priority and a power curve model as the 2nd when the historical data is flawed or insufficient, to analyze the data by cohort, including ad source, campaign, ad set, and creative vs influencers to generate the ROAS curves.

Things that impact the model’s accuracy (by lowering the margin of errors)

• The more historical data we have collected on a specific source, the less we will be depended on the the power curve model

• The higher the spend($100k spend a day will ( usually ) bring more traffic than $10k a day)

• The stronger the early signal ( For instance, the time it takes a potential customer to respond to and ad, and generate a revenue stream for us)

• Lower variance (similar ROAS curves between 2 cohorts from the same source)

• In a nutshell the reason to consolidate or group some budgets together is helping with having more clarity and reducing the margins of errors early on.

Let’s look for an example. We have been thinking about a new source for advertising our digital products and we have decided to spend $10K on January 1st this year to see how profitable it might be for us. All of the customers who were coming from this ad’s cohort will be gathered and monitored to see how fast we are regenerating our money back. This source turned out to be very profitable and was able to set our ambitious result of getting our spend back in a 50 days window (It was just a very simple example to surface the principles here. We might deal with a completely different payback windows but all the calculations with remain the same)

Based on that we can generate a ROAS curve setting 100 percent as $10k:

As we can see from the chart above the ROAS curve is not linear but is very similar to a power curve

We can mathematically extract the power coefficients (b and c) here:

ROAS %= b* (Days Since Start) C

Next, we can set the multipliers:

Through this method, we can identify the profitability of each source at an early stage. With more data, we can reduce our reliance on the power curve model and shift the weight of our predictions towards the multipliers. Although it's possible to calculate the margin of errors for each DSS using both methods, it's beyond the scope of this article. By using this approach, we can avoid spending money that we cannot recover and do not have to wait for several weeks before making a decision.

Measuring alternatives to paid marketing

For instance, if we are able to tag the potential customers that were coming from the a specific sponsorship (like influencers) ( by giving them a token for faster adoption or some other benefits like higher cut ) it will normalize it to an apple to apple ROAS comparison

Best measurable commission model with influencers might be based only on revenue share. This way you won’t have the risk of paying them to bring empty traffic to your website and they will see the benefits only after they bring some valuable creators onboard.

Last but not the least, it does not matter which platform you are running your campaigns or how you store your data, we can help you to generate these

Contact us today to learn more about it