Not too long after Google published their 2012 Q1 earnings announcement showing a decrease in CPCs for the second-straight quarter, we started noticing the following paragraph at the top of AdWords Help pages:

“In mid-May, we’ll be improving our exact match and phrase match options to allow your ads to show for search terms that are close variants, which include misspellings, singular and plural forms, stemmings, acronyms, abbreviations, and accents.”

We can’t prove that Google made the change to protect their paid search revenues in the long term, but our antennae went up. And, while we didn’t find any conclusive proof that Google’s trying to drive CPCs up, we audited client data and found some interesting things afoot.

What Google changed

The power of campaigns optimized with every possible variation of highly-targeted phrase- and exact-match keywords is hard to overstate. By avoiding broad match keywords, advertisers avoid wasting money on less-useful clicks. Assigning bids and destination URLs to a variety of misspellings and variations gives advertisers complete control over reach, budget, and user experience. Advertisers have long known that search volumes and CPCs between singular and plural versions of keywords can vary widely, and that buying a list of reasonable misspellings is a cheap way to beat out the voice of less-aware competitors.

With the change in matching options, Google says it is evening the playing field, allowing every advertiser to show up for the misspellings and variations deemed relevant to a search query. This can be a powerful new feature for less-involved advertisers. Organizations or individuals with smaller budgets and less time to spend reviewing AdWords data and making optimizations will undoubtedly enjoy this new convenience.

But the change isn’t good for everyone. Advertisers with more resources and larger budgets may actually lose some control. Those of us digging through AdWords data on a regular basis will suddenly lack the ability to truly optimize individual bids, destination URLs, and quality scores.

Changing keyword match options

In typical style, Google went beyond “allowing your ads” to show for variants. Rather, they automatically opted advertisers in to the new system. Of course, advertisers can opt out at any time. It’s a simple process if you know where to look.

The new keyword matching options are changed at the campaign level, so you’ll have to change your settings for each campaign individually. To do this go to your campaign settings, find “Keyword matching options” under Advanced settings, and change it to “Do not include variants.”

New AdWords keyword matching options

New AdWords keyword matching options

Generating a report for the new keyword matches

Google also gives advertisers the opportunity to see how the new matchtypes are affecting their campaigns. Simply pull up a search term report and create a filter similar to the one below.

Creating a filter to see new keyword matches

Creating a filter to see new keyword matches

Advertisers should then get something like this:

new adwords keyword matches report

Reviewing Google's new keyword matches

Getting the whole story

While the report Google gives advertisers is nice, it may not tell the whole story. The best way to identify how your account may have changed is to:

  1. Generate a keyword report with as much historical data as possible
  2. Identify keywords with significantly higher impressions and/or costs on or around mid-May
  3. Control for seasonal or other variations by comparing year-over-year numbers along with quarter-over-quarter or month-over-month
  4. Control for changes to ad groups settings

We decided to apply this process to a client we’ve had for several years. The company provides energy to residential and commercial properties.

We compared impression, click, cost, and conversion data for keywords with historical data going back a year or more. This client sees seasonal fluctuations in search volumes. To reduce the likelihood that seasonal variations would cloud the data, we compared the first half of 2011 to the first half of 2012. However, for the purposes of this post, we show only three points in time- 5/11, 1/12, and 5/12. Traditionally, January sees higher volumes and costs, as compared to May.

But when we began to analyze keyword information, we saw some surprising changes.

changes to impressions - possibly due to new adwords keyword matches

Impressions over time - 5/11, 1/12, 5/12, and yearly change

cost changes over time - possibly due to google adwords keyword match changes

Costs over time - 5/11, 1/12, 5/12, and yearly change

We saw impressions and costs skyrocket in May of 2012– a completely unexpected result. Although not shown here, we noticed that conversions for the keywords in the screenshot increased by only one. So it looks like the client’s ads appear more often and cost more money than in the past– without any additional sales.

Of course, we could not instantly attribute these changes to the new matchypes. The client industry is different, due in part to a very mild winter. And we’ve been expanding client reach through display ads, as well as some betas, including Google’s search companion.

So we dug even deeper, controlling for these changes by examining keywords within ad groups. We still saw some significant, unexplainable increases. Here’s an example.

keyword impression changes according to ad group - probably due to google adwords changes

Keyword impressions over time - 5/11, 1/12, and 5/12

Now that we could see changes that could truly be attributed to Google’s new system, rather than account and industry changes, we selected keywords and reviewed them in more detail. We examined impressions for several months. Here’s an example of what we saw.

changes to impressions over time for a single keyword - probably due to adwords keyword match changaes

Impressions for several months - final column is 5/12


Having restricted our analysis to ad groups that had not been changed significant for some time, we did not expect to see this dramatic increase in May 2012. But we saw it for several keywords in various ad groups. We made no changes to the keywords, ad groups, or campaigns that should have affected impressions in this way.

Looking at the data in this way, it appears that Google’s new report may fall short of reality.

Moving forward

To conclusively prove that the new matchtypes are causing these increases, tests will have to be run over the coming weeks. We will remove some campaigns from the new matchtype setup. We will have to sort through the keyword data again, digging until we find what we need.

Advertisers that fail to run these tests over the coming weeks, or those that simply take Google’s new report at face value, will likely experience significant challenges to their budgets in the coming months, without really understanding why.