Optmyzr Blog

How to look beyond ROAS for optimization: PPC Town Hall 17

Ashwin Balakrishnan

Senior Marketing Manager



Many marketers who live and breathe ad spend value ROAS as the holy grail of advertising. For many, it’s a way of calculating and formulating next steps to achieve high(er) revenue. While this metric does forecast quite well, one question that arises is how useful it is for long-term growth. 

With the use of retargeting as a way of fulfilling ROAS targets, PPC professionals might not consider incrementality as a way to get good results. Some sharp minds, however, have turned to customer lifetime value for better optimization with maximum profits.

So this week on episode 17 of PPC Town Hall, we spoke to our panelists who are obsessed with driving better results for PPC campaigns:

  • Andreas Reiffen, Founder and CEO at Crealytics
  • Frederik Boysen, Founder and CEO at

As always, you can view this week’s episode as well as previous editions of PPC Town Hall right here.

Here are 5 insights on how to look beyond ROAS to optimize PPC.

1. Understand the problem with ROAS

Andreas: In large retail organizations, Finance is responsible for both budgeting and setting performance targets. While finance has profitability KPIs and new customer acquisition goals in mind, they translate them into ROAS targets. Often they don’t understand profitability or CLV isn’t a random byproduct of some advertising campaign, but rather something which you can explicitly optimize for.

Once ROAS targets are set, the easiest way to achieve them is to sell products with low margins and high return rates…usually to their existing customers. However, at the end of the budgeting cycle, they usually find that even though they have achieved their ROAS targets, profitability is down.

In the subsequent budgeting cycle, the ROAS target is tightened based upon the weak profitability. This vicious cycle can only be broken by setting other targets than RAS.

2. Consider profit on ad spend

Frederik: You need to calculate the real gross profit on every single order, followed by doing the attribution and looking at customer LTV (lifetime value). You need to work on the profit on the first attribution.

We use POAS (profit on ad spend), which is richer in ad spend before sending out the data to the channels. We calculate all our orders, profits on them, cost price, shipping cost, and payment fees. Then we send these to the platform so that one can make transparent decisions.

3. Put the right information in the system

Andreas: There’s a fundamental difference between how we once did things and what we’re doing now. The levers once used to excel in digital marketing have changed dramatically. In the early days we optimized keywords, ad copies, and landing pages; things that weren’t fully automated.

But AI has made all of these activities redundant. To differentiate from the competition today you need to ensure you optimize for the right targets and that you can accurately measure the value of each single click…and feed your algorithm with first-party data. Putting the right information back into the system is key to optimizing beyond ROAS.

In order to activate your data you need to first assess the exact order margin, then deduct the expected returns. As a second step you need to know whether an order was done by a new customer or an existing one. If it was a new customer who purchased, future purchases are to be expected, so you add a (residual customer lifetime value on top of the first-order-margin.

Ultimately, you need to slice this entire value if there were several clicks involved. The end goal should be to attain the value of each and every click. This is a prerequisite for bidding systems to work for your specific business.

4. To retarget or not?

Andreas: Whoever is curious to see the impact of data activation should do one experiment. First off? Let tROAS run. Then, in one instance, you provide the first-click data. In the other, last-click data. Analyze the retargeting share of both settings and you’ll see only a small retargeting share through the first click; as all the credits get allocated to it.

What happens if you test for incrementality? You’ll see that the more likely a user is to buy, the better the results will look, albeit with lower incremental impact. Nobody today can answer this question: whether to bid up – or down – on your retargeting activity.

I believe that attribution systems have completely failed. They assume that advertising must be responsible for the sale…and so it allocates credits based on different parameters. The only thing of value is to run isolated incremental audience-based models. It’s here where you can truly find out the scope of impact.

5. Go beyond attribution

Frederik: We have a dashboard where we take away the attribution fully. We actually look at gross profits, usage on ad spend, and gross profit after ad spend by keeping their ratios the same. This way you can actually see whether your gross profit will increase or decrease if you don’t change the ratios. After this, you can try to allocate with some attribution. This way you can look at the financials rather than just the attribution.


While ROAS might have been the guiding light of the past, one can’t forget that what are essentially Google’s metrics might not accurately reflect your client’s or company’s goals.

The only way to sustain a ROAS-driven system is to layer different rules on product categories. Even geolocations seem to have an effect on ROAS targets; you might find, for example, that the customer acquisition rate is higher in New York than in San Francisco.

It’s hard to automate some things fully as you might not have the same targets across different locations and products. The only way to fix that right now is to move towards a conversion tracking and attribution system that takes smaller things into account, like locations and incrementality.