To celebrate the new and shiny PPC section, I wanted to share here a post I wrote almost 10 years ago on my blog, later to be modernized for another forum back in 2021, and now to be polished again for Office Outlaw.
The reason this is the third time I'm writing it is that this is the most important thing that you should look at when you run campaigns for clients or for yourself, and 90% of the campaign managers have no idea what it is, let alone how to calculate or use it.
Not accounting for or knowing what Marginal CPA is is the reason successful campaigns fail to scale up. Knowing how to look at it prepares you to plan for scaling properly and maximize profits from a campaign rather than just getting more conversions.
First, let’s look at the difference between ROI (or ROAS) and maximum profit.
Max Profit vs. Max ROI (ROAS)
Most campaign managers think that higher ROI guarantees higher profit. This is as wrong as assuming that higher CTR guarantees more clicks. It just isn’t.
ROI and Profit are two different terms that relate to the results of a campaign (return, profits, conversions, revenue - you name it) but that’s where the similarity between them ends.
ROI measures your Return On Investment, and in other words: how much you received for a $ spent. Meaning, it doesn’t take your actual ad spend (or scale) into account.
Profit, on the other hand, takes both your ad spend and ROI into account. Your profit (a simplified version of gross profit, for that matter) is the result of your income minus your spend.
As long as the ROI is positive, the higher the ad spent, the higher the profit (or, if you prefer, the louder the “cha-ching”).
Calculating ROI
People calculate ROI in one of two ways that differ from one another by their lowest minimum value:
- (total revenue / total ad spend)
This is the simplest ROI calculation (often referred to as ROAS). It shows you how much you generated from every $ spent. If you spent $100 and generated $200, your ROI would be 2 (or 200%). If you didn’t generate anything, it would be 0, and the break-even point is 1 (or 100%).
In this article, I’ll be using this method of calculating ROI.
- {(income – ad spend) / ad spent}
This is the overcomplicated method that shows how much you *profited* from every $ spent. I assume that at some point, a bored accountant came up with it to justify going to accounting school or something.
This is, however, the “proper” way to calculate ROI.
If you spent $100 and generated $200, your ROI would be 1, or 100%. If you didn’t generate anything, it would be -100%.
This is the source for the terms negative or positive ROI - in this calculation, the break-even point is 0.
Calculating Profits
Net vs Gross Profits
Generally,
campaigns are optimized toward gross profits, and not net profit.
The reason is that the gross profit measures the contribution of an operation toward the bottom line and doesn’t take costs or income that are not directly related to that operation into consideration. For example, rent and internet are two things you shouldn’t think of when you’re running a campaign.
You can’t work without internet or an office, but their cost doesn’t make a campaign more or less profitable.
On the other hand, if you’re in e-commerce, the cost of an item directly affects the profitability of a campaign (the price *you* paid for it, not the amount you sold it for).
So when we’re calculating profits from a campaign, we look at the {revenue - media cost - COGS (cost of goods sold)}
Defuq is COGS?
The cost of the item you sold, shipping (if you covered it), sales tax, labor that is directly related to the sold item, etc..
The point of the thread: Marginal CPA
Let’s assume we have a campaign that consistently brings 50 conversions a day at a CPA of around $10. We are at a point where we can’t increase the scale just by raising the budget..
In other words, we know that by paying $10 or less per conversion, we can acquire no more than 50 conversions per day (daily spend of $500).
Each sale is worth $20, so our ROI is around 200%, meaning that you’re doubling your money (each dollar spent generates $2 of revenue)
If you increase your bid, you’ll have more conversions, but at a higher CPA.
We want to scale up (yay). The client called, he needs us to 2x the number of daily sales starting tomorrow. We decided to increase the bids by 60%.
Let’s look at the numbers:
Thanks to increasing the bid, we’re now getting 100 conversions per day, at a CPA of $16, 60% more than before, and are spending $1,600.
Each sale is still worth $20, so you’re happy, the client’s happy, and the campaign is still profitable.
Or is it?
Let’s compare the stats before and after increasing the bids:
Before increasing the bids, we spent $500 a day for 50 conversions at $10 each
After increasing the bid we spent $1600 a day for 100 conversions at $16 each.
The campaign is still profitable, given that the value of each conversion remained $20,
but did scaling up really help us?
Marginal cost helps us find the cost of the last conversion (or batch of conversions) by comparing the differences between before and after altering the bid.
{(Total cost after increasing the bid - Total original cost) / (Conversions after increasing the bid - No. of original conversions)} = (1600-500)/(100-50) = 1100/50 = 22
So, even though it looks like we only paid $16 for each conversion,
we actually paid $22 for every conversion after the original 50 we had before raising the bids.
Meaning, by scaling up, we lost money.
Let’s look at it in another way:
| Metric | Original | After raising the bids | Difference |
| Cost | 500 | 1600 | 1100 |
| Conversions | 50 | 100 | 50 |
| CPA | $10 | $16 | |
| Revenue | $1000 | $2000 | 1000 |
| Profit | $500 | $400 | $-100 |
So how can you tell when to stop raising your bids?
As sad as it is, the answer depends on the demand/supply curve and your campaign setup. That being said, you can (and should) test your bids methodically.
To do that, we need to establish a baseline, which is the current state of your account. If the conversion value vary, use an average, and if the CPA vary, try to find a consistent 7 day average.
Anyway, try to be as consistent as you can, and raise the bids every week at the same time, by the same %.
For example, raise your bids in a specific campaign by 7% every Monday at 8 am.
Just bear in mind that sometimes there's no point in increasing your bids. In Google Ads, you can look at your impression share and your absolute top impressions share to get a better idea where you stand.
If you have most of the available impressions, try lowering them and see when you start to lose traffic and revenue. Your max profit point could require you to scale down, you've got to test things out!
Anyway, assuming you are still looking to scale up, calculate the CPA for the extra traffic and make sure that you’re still positive. If you are still profitable, keep increasing the bid until your marginal CPA is breaking even with your income per action. When you reach this point, start decreasing your bid slowly (slower than the raising rate). The moment you become positive again, you have reached the maximum profit point.
It’s not as easy as it sounds, because your campaigns do not run in a vacuum, and every move your competitors make changes your performance.
That said, knowing your marginal CPA and level of profitability gives you clear boundaries, helps you set budgets for tests, and impresses your client when you casually mention it in a conversation.
Things to pay attention to:
This can be done at almost every level, but depending on your current scale, I’d advise you to stick to a level where you have at least 30-40 conversions every timeframe (week/month)
Know the differences between days. Some campaigns are more active on weekends and some during the week. Increasing or lowering the bids is a long process. To know exactly what every bid change has done, I usually wait a week between changes.
Your campaign is not static and it changes every single moment. After finding the critical point, keep experimenting with different bids.
This post was initially written for Google Ads, but the same principle can (and should) be applied in other networks as well. Not just programmatic advertising, but even in cases where you buy leads from an affiliate network, etc.
The real “moral of the story” is not to get excited about the volume of conversions while forgetting the bottom line. Marginal CPA is just a way to simplify this process and aim you toward your most profitable point on the graph.
Bronze Star
