August 18, 2014
Last week AdWords announced it would be removing the option to select whether we include or exclude close variants in our campaign settings. The justification is that those typing in search queries aren’t perfect spellers. The intent behind a misspelled and correctly spelled keyword is the same. This is the same logic Google uses to tell us Tablet and Desktop are the same thing (they aren’t). Why are we ignoring the fact that a user who can’t spell might not be as valuable a lead for say, Harvard Business School as one who can? Why are we being forced to assume someone looking for multiple products is as valuable as someone looking for just one?
What Close Variants Mean
Essentially it means that someone searching for “laptops” or “lpatop” will trigger the (increasingly incorrectly named) exact match for [laptop].
Why isn’t it a big deal?
Most advertisers on AdWords already have campaigns set to include close variants.
Why is it a big deal?
- Google is continuing to erode options that power users of AdWords value
- A small but significant number of advertisers have bad results with close variants
Investigating the data
With the collective outrage of PPC professionals everywhere in mind, I set out to answer 3 important questions:
- Do close variants make a difference to my accounts?
- How much is my CPA affected by turning on close variants?
- Are the net gains in conversions worth it?
Note: I’m only attempting to answer this for my own accounts. I appreciate that every account is different which is why I’d love for you to comment with your own stats.
How to get data for your own account:
Thanks to the magic of Excel pivot tables it’s easy to find how close variants are doing in your account. First, navigate to your keywords tab and select “Search Terms > All” from the “Details” menu.
From here download your report making sure to add columns for any metrics you will need – clicks, cost, conversions etc. I like to look back over 30 days, but if you don’t have much data go as far back as you need to.
Now you’re going to want to create a pivot table. Set “match type” as your row label and clicks, impressions, cost and conversions to your values. Remember to add calculated fields for CTR, CPC, CPA and ROAS.
Let’s look at the data from some of my accounts…
Account #1: Lead Generation
In my first account close variants are performing worse than non-close variant matches.
CPA for exact match terms is $15.69 compared with $18.49 on the close variant matches. That’s a dramatic difference (18%) for something that Google is telling us is insignificant.
However, am I getting too worked up here? Take a look at the overall impact on performance metrics:
What this chart shows me is that my overall CPA is only 1.42% higher for having close variants enabled. I’m worse off, but not by that much.
What’s the trade-off? Well close variants brought in 165 extra leads! That’s a 16.8% increase.
What I have to ask myself is “are 165 extra conversions worth a 1.42% increase in my overall CPA?” In this case the answer is definitely yes.
Conclusion: Close variants are valuable for this account.
Account #2: Lead Generation
In stark contrast to the last account, close variants here perform terribly. On both phrase and exact my overall CPA is much higher than without them.
Result: 101 extra conversions for a 22.8% increase in my overall CPA
This account was the second I looked into and I wasn’t expecting results so strongly against close variants. Google forcing a change on this account which will take overall CPA from $91.01 to $111.76 is ridiculous. That could be the difference between continuing with, or bailing on AdWords altogether for some advertisers.
Conclusion: Let me turn them off!
Account #3: eCommerce
What’s interesting here is that the account’s performance was actually better overall with close variants enabled in virtually every metric.
Result: 58 more sales, $6,563.38 more revenue and a 1.43% reduction in CPA!
Conclusion: Close variants work really well here.
Account #4: eCommerce
Result: A 3.31% higher overall CPA to bring in 1.2% more conversions.
Conclusion: Close variants made things slightly worse.
Let’s go back to those 3 questions:
Q: Do close variants make a difference to my accounts?
A: From four accounts: Two improved, one was about even and one was much worse.
Q: How much is my CPA affected by turning on close variants?
A: Worse in three accounts, better in one.
Q: Are the net gains in conversions worth it?
A: Take a look at the chart below showing the total percentage of conversions to come in from close variants vs. non-close variants. It adds weight to Google’s argument that close variants can add a nice boost to traffic and sales in 3 out of 4 accounts.
Digging a little deeper into one of the above accounts shows there are a number of additional considerations when dealing with close variants. For example, what percentage of your traffic comes from mobile devices? Take a look at this table I put together:
Notice how much worse the CPA is for Tablet and Mobile compared with Computers? If anything Google should be giving us additional controls to segment close variants by device, not removing it.
I also pulled the same numbers for Google search only vs. search partners. Unsurprisingly I once again found a large discrepancy between the two.
Conclusion: Google Should Leave Things As They Were!
The thing that frustrates me is that this should be my decision to make. Google has looked at its data in aggregate and decided that close variants increase clicks and conversions for relatively minor changes in CPAs. However ‘aggregate’ won’t cut it when it comes to a 22% higher CPA for an important client.
What’s my solution?
It’s simple. Leave close variants as the default, but let advertisers retain the right to opt out.
Most will continue to use close variants and Google will get the clicks and spend it wants. Other advanced users will retain a valuable tool for account optimization. Unbelievably to Google, some people out there are pretty good at this stuff and can handle the added complexity.
Win – win for everybody, right?
I’d love to get all of you to quickly pull this data for your accounts and comment below so we can start to build up a clearer picture of the impact.
Featured image courtesy of MattysFlicks.