The reason we want full and easy access to true ROI numbers for our paid search campaigns is so that we can make smarter decisions about improving those campaigns. The Google Adwords Blog recently wrote about how to calculate ROI using Adwords (see earlier commentary) and included in the final post some recommendations on what to do with this new information.

So what did Google recommend? It starts with some ideas of when your new ROI information is useful:

…look for keywords that have a negative ROI (i.e. less than 100%), but enough clicks that you’d reasonably expect some conversions to have happened. These are keywords for which your advertising costs exceed your profits, so your bids for these keywords may be set too high.

There are two problems here. First, a simple positive ROI isn’t the goal. You should have a target ROI in mind (you’ve got fixed costs to cover and still want a little something called profit left over) so whether your number is 120% or 150% or higher, you should have an idea of the real goal is for your business. This should be obvious, of course, but the article should have mentioned it.

Second, ‘reasonably expect’ is not a great way to decide what is essentially a mathematical problem – most folks just aren’t adept at eyeballing margin-of-error. Instead, do a simple calculation (one over the square root of the sample size, for example) to figure out if you’ve had enough clicks to have a margin-of-error that would allow you to decide anything.

In most cases, you should lower the bids for keywords with ROI less than 100% to the amount in the “Value / Click” column from the report. The “Value / Click” amount reflects how much profit you gain per click, so if you set your maximum CPC to this amount and the performance remains consistent, you will at least break even on these keywords.

If your ROI is < 100% (or your target) then you do want to do something. Is it rewrite the creative? Add negative keywords to your campaign? Change your landing page? Turn ads off on Weds mornings? There’s no way to say. On the other hand, if you lower your bids and don’t try to figure the other stuff out, you’ll save some money and a lot of time, but you may never know the potential revenue and profit of actually fixing whatever is really wrong. And if you do lower it, chances are pretty good (unless you’re starting at an unreasonably high bid price / position) that you’ll just keep having to lower it until you just turn it off.

For keywords that have a positive ROI (i.e. greater than 100%), consider increasing your maximum bid — but not higher than the amount in the “Value / Click” column. By increasing your bid, your ROI will decrease but you may end up making a greater total profit because you’re getting more clicks when your ad moves to a higher average position.

This one would require its own post (or book chapter) to challenge all the wacky assumptions and PPC mythology. Testing higher bids when keywords produce good results makes a lot of sense. It may send your ROI up or down. It may send your total profit up or down. And when your bid gets anywhere near your revenue per click, Google is making all the money and you’re not making any.

I genuinely don’t mean to pick on this particular Google Guy. But paid search is a largely quantitative exercise which is today primarily driven by guesses and rumors. Leading people the other way is an important effort, but this series in large part snatched defeat from the jaws of victory. One step forward, three steps back.