Bid Management Has a Long Way to Go

I have written about Bid Management many times, and am a big fan of what it can do to help scale a PPC campaign. However, a bad bid management system can trap or limit your campaign.

Portfolio based bid management is a must. Basically, what this is, is a method for dealing with individual keywords that do not present enough data to determine their ROI on an individual basis. The classic cases are long tail keywords, such as one that has a single click, but it resulted in a conversion, or another keyword that has 10 clicks but no conversions.

Making sweeping bid price decisions based on such a small amount of data is fundamentally foolhardy. Smart bid management tools will aggregate data in larger groups. One way to do this is to take all the keywords in an ad group, and then treat them as one statistical entity and calculate their bids based on their performance as a group. If that does not work, the next step is to look at the campaign level.

But this is just the beginning of the issues you face. Here are some others:

  1. You may have keywords that you bid on for branding purposes, so ROI be damned, you are going to bid aggressively on them anyway.
  2. Other keywords may exist that you want to measure on an ROI basis, unless they fall below a certain position, in which case, you are willing to sacrifice ROI.
  3. Some bid management tools don’t allow you to manage to ROAS (or ROI), but force you to work to a CPA goal. Go for all the accuracy you can my friend, and ROAS/ROI is a far superior approach to CPA.
  4. Attribution is another huge problem. If you have multiple clicks made by a user to the site, which visit do you allocate the revenue to? First? Last? Allocated across all clicks? If you figure this out, what do you do when display ads and email campaigns deliver some of the clicks prior to a transaction. This is a problem that analytics and bid management vendors are just beginning to cope with.
  5. I have searched high and low for a bid management application that uses trial and error testing to determine when the ROI goal is being met. I have yet to find one. But if you can’t get this, you really don;t have the control you need when setting ROAS.

This industry is in its infancy, and the needs of sophisticated PPC marketers are met best by some of the better bid management applications out there, such as Marin Software, and Efficient Frontier. But, boy to we have a long way to go.

For web marketers, the key is to pick the tool that most closely meets their biggest needs. Making this determination is far from an easy task. What is needed is an analysis of the marketers online business, the dynamics of its marketplace, and an assessment of which tools best meets its needs.


  1. chris1 says

    This is a great post, Eric! There is a definitely a human element needed in bid management at this point in time. I would also take your premise here one step further and say that the goal should always be total profit rather than ROI. If you focus solely on ROI (for each term), in many cases you might find that you are bidding too low (losing out on higher volume with slightly lower ROI) or too high (losing out on large gaps in the bidding that can account for very high ROI at slightly lower volume). Maybe this is

    Assume in the following example that your break even cost/conversion would be $50. To find the optimal position for a given term, you will need to test. The optimal position, optimal CPA and optimal ROI can and will vary for each term.

    Term #1:

    Position 1 – $6750 spend, 150 sales, $7500 revenue = $750 profit ($45 CPA, 111% ROI)
    Position 2 – $4000 spend, 100 sales, $5000 revenue = $1000 profit ($40 CPA, 125% ROI)
    Position 3 – $2800 spend, 80 sales, $4000 revenue = $1200 profit ($35 CPA, 143% ROI)
    Position 4 – $1500 spend, 50 sales, $2500 revenue = $1000 profit ($30 CPA, 167% ROI)
    Position 5 – $600 spend, 30 sales, $1500 revenue = $900 profit ($20 CPA, 250% ROI)

    Term #2:

    Position 1 – $2800 spend, 80 sales, $4000 revenue = $1200 profit ($35 CPA, 143% ROI)
    Position 2 – $1000 spend, 50 sales, $2500 revenue = $1500 profit ($20 CPA, 250% ROI)
    Position 3 – $600 spend, 35 sales, $1750 revenue = $1150 profit ($17 CPA, 291% ROI)
    Position 4 – $400 spend, 25 sales, $1250 revenue = $850 profit ($16 CPA, 313% ROI)
    Position 5 – $300 spend, 20 sales, $1000 revenue = $700 profit ($15 CPA, 233% ROI)

    In the example above, if you were to base your bids strictly on ROI or a target CPA, it would be impossible to bid optimally for both terms. Instead, what PPC managers should do (and what currently is impossible for bid management software) is to test and find the “sweet spot” for each term. Now this might be impossible to do for an entire account, but you should be doing this for your most important terms and an experienced PPC manager should be able to make an educated guess as to where the sweet spot is based on limited data.

    Looking at the above stats, it would be easy for a PPC manager to think he was doing a great job if his ad was in any of the positions listed above. All of the positions are making money and they all provide either great ROI or lots of conversions. However, what separates a great PPC manager from the rest (and the automated bid management tools) is their ability to find that sweet spot for as many terms as they can, thereby maximizing profit for their campaigns.

  2. chris1 says

    Eric – That article is spot on. So many SEMs are leaving $$ on the table by not tracking and optimizing down to the most granular level.

  3. Oliver33 says

    Interesting! , when you try to analyze what is the best Bid Management you have to understand that Bid Management is only one piece of the puzzle. We are using quality management technology – finding the best keywords, bid management, a/b testing, attribution and tracking is much more than just bid management. I agree that bid management has a long way to go – the way to the quality management ;)..

Leave a Reply

Your email address will not be published. Required fields are marked *