In a recent post, ‘Maximizing ROI: The Wrong Game’ , the Rimm-Kaufman Group’s George Michie wrote that when choosing between spending to a given budget limit or hitting a particular ROI target, he has “always believed that the ROI target should take precedence.” But which value of ROI should you target?
Many marketers misguidedly wish for their ROI to simply be ‘as high as possible’. ClickEquations’ software has an algorithm called “Maximize ROI” and Google’s webpage for its Conversion Optimizer  is subtitled “Maximize your advertising ROI“. That mantra has appeared in press releases and blog posts by Efficient Frontier , Marin Software , and even The Search Agency  and The Rimm-Kaufman Group .
So, if maximizing your ROI is “the wrong game”, then what’s the right one? Well, our goal as economic agents is not to maximize ROI, it’s to maximize profit accrued per day. As Mr. Michie reminds us: “you don’t put percentages in the bank”. So, we need to find the value of the ROI where our profit is maximized. That value, not ‘as high as possible’, should be our target ROI.
Google defines ROI  as ‘(Revenue – Cost) / Cost’, where Cost includes both the Advertising Cost (AdCost) and the cost of goods sold (COGS). And Mr. Michie mentioned that, when COGS is zero, “profit is maximized when the ratio of margin generated to advertising spend is 2 to 1.”
From these two facts, we can determine the ROI corresponding to the point of maximum profitability.
Which ROI should I target?
Alan Rimm-Kaufman showed in this blog post  that (when variable costs are zero and Revenue increases as the square root of AdCost) the profit per day is maximized when AdCost / Revenue = (1 – cogs%) / 2, where ‘cogs%’ is COGS/Revenue. Therefore, we can find the AdCost of maximum profitability:
By substituting this equation for AdCost into the definition of ROI, we find:
Pick whichever you like, the results are the same. If you know ‘Revenue’ and ‘COGS’ for a given keyword (or if you know cogs%) plug them into the appropriate equation above to find your profit-maximizing ROI.
Don’t Take My Word For It: How to check profit-maximizing ROI with Google’s Bid Simulator
Google has a feature called the ‘Bid Simulator’ which, for keywords that have gotten at least about 20 clicks in the past week, provides estimates of the number of impressions, number of clicks and AdCost that word could have gotten over the past 7 days for several different possible bids.
If the Bid Simulator is active for one of your words, you’ll see a little icon in the AdWords interface next to that word’s ‘Max. CPC’. Click on the icon and copy the estimates from it into a spreadsheet. You can add columns to estimate the word’s weekly Revenue by multiplying the estimated Clicks by your record of the word’s recent historical Revenue_Per_Click. Similarly, you can estimate the word’s weekly COGS by multiplying the Bid Simulator’s estimate of Clicks by your record of the word’s recent historical COGS_per_Click. Take the Revenue column and subtract both the AdCost and the estimated COGS to estimate the Profit per week. And use ‘Profit / (AdCost + COGS)’ to find the ROI.
You can see from the example spreadsheet above, calculated for an actual keyword which The Search Agency manages, that Google’s Bid Simulator estimates that the Profit is maximized for this keyword ($1,382.20 per week) when the bid (aka, ‘Max. CPC’) is $8.42 and the ROI is 37%.
Actually, we can find from the recent Revenue_per_Click and COGS_per_Click for this word, along with the equations I presented above, that the profit for this word is maximized when the ROI is about 35%, making the optimal bid a little higher than $8.42. (The reason for the discrepency is that Google’s Bid Simulator simply didn’t give the actual profit-maximizing Max. CPC as a row in their table.)
You can try this for yourself with your own account and this Microsoft Excel 2007 spreadsheet . As long as variable costs are minimal and Revenue generally increases as the square root of AdCost, the target ROI given by the equations above will be the value at which the estimated Profit is maximized. Afterward, you can just use the equations directly without having to take estimates from Google’s Bid Simulator – which can be very helpful for words where the Bid Simulator is not active, or for keywords on other search engines besides Google.
The short lesson is: for any given keyword at any given time, there is one profit-maximizing ROI. To aim higher or lower than that value simply misses the mark.