I have received a number of questions regarding different ways to bid. All in all, I feel there are a number of schools of thought that revolve around the same central tenets. The short answer is that you’ll want to set and adjust bids according to your goals (ROAS? CPA? Revenue? Gross Profit?), but there are best practices for doing that no matter what goals you pursue.

We’ll break this into two steps: 1) considerations you must take before setting your initial bids; and 2) setting filters to make bid adjustments once the data starts coming in.

Here we go!

Before you bid, remember to

1. Prioritize your goals. Are you most invested in improving ROAS, CPA, Rank, Revenue, Gross Profit?  You are going to have to identify and create priorities if you have multiple goals. The goal is going to set the line between good and bad spend.

2. Split out the outliers. Are there portions of your account that behave differently in conversion rate, AOV, or Avg. CPC? These need to be broken out. It doesn’t make sense to have a $3,000 dollar computer have the same CPA as a $50 iPod.

3. Adjust according to time frame. Look at your average time it takes to convert.  You can find this figure within AdWords under tools and analysis => conversion funnel => time lag. If the average is short (1 to 2 days), you can have a shorter look-back period to make your decisions.

4. Factor in latency. If you are using day-of-click tracking, you know that latent conversions can be a big factor. In the example below we can see that 16 percent of all conversions occur after 12 days. If we were looking at two weeks of data, we should adjust the ROAS goal to be ≈15% lower (or higher if using CPA targets).

One thing to be careful of is that more expensive products generally have a longer click-to-conversion period. This means that average order value is correlated with time lag, and from a revenue perspective you might need to adjust by more than 15 percent.

Click to enlarge the image below:

5. Factor variables into your budget. In a perfect world we would be able to invest as much as we like based solely on keyword performance; however, this is not the case. If resources are an issue, we are going to have to move money towards the most efficient terms. In situations where campaigns are being paused due to daily budget caps, it is almost always better to have lower bids and increase margins.

6. Calculate click threshold. We don’t want to bid down KWs that have not attained enough data for us to make a judgment. I usually look at the conversion rate of the category and back out into a lowest-clicks threshold at which, assuming an even distribution, I would expect a conversion. For example, an expected conversion rate of 2% would need 50 clicks.

7. Use a different bidding strategy for the long tail. Go through your account and find the KWs that chronically do not achieve click thresholds. Use the AdWords label feature to categorize these as long tail.  We are going to use a different strategy to bid these down.

Filter your results and adjust accordingly

OK, now we’re ready to create the filters: bad spenders, winners, losers, long tail keywords, and sleepers. Before we set these and make adjustments, though, a reminder: as in the graph below, CTR (red) can be drastically affected by even small changes in CPC.

Lesson learned? Onward.

First up are the bad spenders. If your account is overspending, these keywords will most likely be a culprit. These keywords are defined by their high spend and lack of conversions. To address these keywords, we need to complete the following steps:

1. Set your date range to the last 30 days

2. Create a filter with cost > your worse-case scenario CPA.

3. Select all the keywords you have determined have bad spend. Click Edit -> Change Bids -> Decrease by – > 20% (click to enlarge image below).

bid adjustments adwords

The fun part is bidding up the winners. Here we’re looking at the keywords that beat goals and have some room for their average rank to grow. Remember that you can save these filters to use again. Since we do not care about click threshold here, this also applies to long tail keywords. In the below example, we’re increasing max CPC bids for the winners:

winning keyword bids

The losers are keywords that have failed to meet their goals. These keywords are close to meeting their goals, but came up just short (and we’re bidding them down only slightly, as shown below). The filters we apply for the losers are fairly similar to the winners, but we use a click threshold to qualify the change.

losing keywords bid decrease

Repeat this process for the long tail, but decrease the percent change in the bids and increase the time frame to collect more data. Keep in mind that because of latency, your target goals will probably have to change.

Last but not least, I like to wake up the sleepers – active KWs that have not received any impressions. This is to ensure that there are no KWs that are not being touched due to lack of data. This does not have to be a very frequent thing, but it ensures all keywords are in play, and none slip through the cracks.

Stay tuned for my next post, on bidding automation!

Leave a Comment

Jay Stampfl
-Jay Stampfl, who joined 3Q Digital in January 2013, originally started working in performance display in 2007 with adBrite. In 2010, he continued his digital marketing career with the French-owned search agency eSearchVision, where he managed some of the most recognizable brands in America. Jay has a background in Evolutionary Biology research and received a degree in Biological Anthropology from UC Berkeley.