Keyword Research and Targeting Without Exact Match

Whatever the motives behind Google's recent removal of exact-match keyword targeting from AdWords, the resulting uncertainty makes keyword research that much more difficult. In today's Whiteboard Friday, Rand talks about the implications of the change, and offers tips for the most effective research going forward.

For reference, here's a still of this week's whiteboard!

Video Transcription

Howdy Moz fans, and welcome to another edition of Whiteboard Friday. This week we're chatting about keyword research and the challenge that's being presented with the loss of exact match bidding capabilities inside of Google's AdWords platform.

AdWords has sort of become a keyword research and opportunity tool of choice for SEO and, of course, PPC folks for a decade now. We've always had some optionality around how we choose keywords inside of AdWords.

Say I was selling groceries online. Maybe I'm selling Asian groceries online and, specifically, fish sauce, and I want to do some modifications to which terms and phrases I bid on. So I could use things like these brackets to say exact match only, bid on keywords that are precisely fish sauce, no modifiers, no changes, not fishes sauce, not fish sauces, not Vietnamese fish sauce. I just want the word fish sauce. Or I could go with a partial phrase match, meaning no modifications to this part of the phrase, but yes if it's Vietnamese fish sauce or fish sauce recipes, that's fine. Or I could go fish sauce broad match and then let Google sort of extrapolate out and add all sorts of things on there.

Now, as of September of 2014, Google AdWords is making a change to their policy. All campaigns and keywords that you employ inside campaigns must use close variance. Essentially, they're removing the exact match and saying, "Hey, we don't think this power tool is useful, and that control is going to be lost to folks."

There are two ways to look at this. One is Google took down their plaque on the wall that said "Do no evil" and put up a plaque that said "Be kind of evil when it makes us more money." That is one perspective.

As many folks have pointed out, including Larry Kim from WordStream, many, many campaigns, in fact a vast majority of campaigns that are integrated with WordStream he noted, don't even actually use exact match in this format. So maybe they're not losing all that much, and Google is just saying, "Hey, this is a very tight feature, and we're worried about how small businesses and people who are bidding might be employing it. Not all the users who are using it are power users. People are getting confused. So we're taking away that functionality."

My guess is the truth is probably somewhere in between. This will almost certainly lead to a considerable amount of more revenue for Google, because a lot more people will be bidding on terms and phrases that perhaps they should be bidding on and really want and perhaps they didn't intend to bid on and don't particularly want.

In any case, it loses some of that fine control. That's very frustrating for PPC folks, but it can also be frustrating for us SEO folks. Now, we honestly don't know. We don't have data. It'll be pretty interesting to see whether in September this changes.

If you go to Google's Keyword Planner today inside of AdWords -- which is free by the way, you just need to sign in with a Google account -- you can do a search term like "fish sauce" and it'll return a bunch of things. I did a search for fish sauce, and it returned for me things like fish sauce, average monthly searches 22,200, competition low. This is not competition for SEO, by the way. You can get that from something like Moz's Keyword Difficulty Score. This is competition in AdWords itself -- how many people are bidding, how aggressively they're bidding, that sort of thing.

Then, it suggests other things like Thai fish sauce, fish sauce substitute, vegan fish sauce -- I don't think that's going to work -- sauces for fish. Sauces for fish? Are you kidding me? I understand that technically has the words sauce and fish in it, but that has an entirely different meaning. It's sort of odd that they're showing that to me. Then, they give me the search volume for all these and this kind of thing.

What we don't know is whether these are exact, partial, phrase match, broad match. My guess is they're broad match, whether they include those close variance or don't include them, the number.

It's been kind of tough. It'll be very interesting to see if, when this shift happens in September, a lot of these numbers change dramatically, and we're seeing like oh, yeah, Google was showing me more specific exact data previously for these terms, and now they're showing broader numbers for each of these, or whether that's already the case today. I suspect it's actually already the case today, and it's been a while, a couple of years, since Google actually offered truer, closer to reality numbers around what these are.

I think these numbers probably include a lot of close variance and potentially even some broad case matches. For example, fish sauce, this 22,000 number might actually include sauces for fish right in there. This makes keyword research really tough, really hard.

For us in SEO, the lost ability makes it a lot more difficult. The bidding situation in AdWords makes it a lot more difficult to determine keyword performance, and keyword performance is something that's critical to us. That tells us when someone searches for "Thai fish sauce," they're much more likely to buy from us, or when they search for a particular brand of fish sauce they're much more likely to buy from us, versus the broad phrase
"fish sauce."

That's pretty frustrating, because we often use AdWords data, PPC data to say, "Hey, this is a super valuable keyword. SEO team, let's go get this search term and try and rank for it organically, because when we rank for it in paid search, we get a lot of ROI from that." That's going to make it harder, absolutely.

Potentially, it means more noise in these keyword research numbers. That noise could come from the inclusion of more close variance in the data. We'll see how that happens. That potentially muddies the research and prioritization process for us. It might be the case that this is already happening though.

What Can We Do?

There are a few things we can do. We're not powerless. We do have some ability to influence this. First off, any time you're doing keyword research I now suggest that you just can't rely on AdWords alone. It's not good enough. You've got to be using at the very least something like Google Suggest. I love the tool SEMrush. I love keywordtool.io. I think those are both excellent.

1) Google Suggest

Google Suggest, for example, when I start typing "fish sauce," knowing that I'm in Seattle if I am geo located, it'll show me some things. It did show me fish sauce Seattle, fish sauce Portland. Portland, I think, was actually higher than Seattle. I guess we're looking for fish sauce from Portland more so. It showed me fish sauce chicken wings, which is particularly popular around here and delicious. It showed me fish sauce uses, nutrition, fish sauce versus oyster sauce.

These were not things that I got on my suggested list. Granted, I didn't go through all 800 or so suggested keywords, but a few of these were very different from what I saw over here. I think AdWords tends to be very focused on commercial intent terms, things that they know people are trying to buy or do some sort of commercial activity around. So it is valuable for advertisers.

A lot of this is more informational searches, which is huge for content marketers, huge for bloggers, big for anyone who's doing SEO to try and attract awareness, brand attention, links to their site, those kinds of things. So you can't ignore these keywords.

The other thing that's very nice is if you do these, you can do them geo modified or non-geo modified, and if you do them, they tend to be in popularity order. That means I know that "fish sauce chicken wings" is probably a more popular search term in Seattle right now than "fish sauce uses." Also fascinating useful information. I can rank some of that stuff against the numbers that I'm seeing over here and try and compare and contrast.

It's not always perfect, by the way. Sometimes they over geo modify, or people in your search area are searching a little differently from how the rest of the world is searching, whatever the case may be. There are a lot of temporal factors going on here. So if all of a sudden there's a fish sauce food truck that opens up in Seattle, that might get super popular in the search terms even though it's not very popular anywhere else.

2) Google Analytics/Adwords

A second thing you can do is follow up directly inside of your Google Analytics or AdWords to see which specific, unique exact terms sent traffic and how that performed. Unlike organic search, where Google's taken away 95%, 97% of all keyword referral data, that referral data does still exist in GA and in AdWords. It doesn't appear that this change will mean that Google will take sauces for fish and report it as fish sauce in your campaign. It looks like they'll still be reporting the actual keyword that sent traffic, and so you can infer from that this prioritization importance process.

3) Bing/Yahoo! Referrals

Number three, you can actually use Bing and Yahoo or any search engine that is still reporting referral data. Approximately 5% of Google's keyword data is still being reported. You can use those referrals to help infer relative quantities and relative performance on a per keyword basis at least for your most important keywords. For stuff in the long tail and the chunky middle, it's going to be harder, maybe even impossible in the long tail. But at the head of the demand curve at least you can say, "Yes, Thai fish sauce doesn't perform quite as well as Vietnamese fish sauce for us. It turns out Vietnamese fish sauce really gets us the great quality traffic that we're looking for. We'll focus on that one first."

4) Broaden Your Keyword Targeting

I think because of all of this keyword data removal, just in general we have to almost become more like Google Hummingbird, the update, around how we do keyword and intent matching, a little less towards the exact phrase, exact match keyword targeting, and a little more towards the intent of the searcher and all of their potential interests and intent around that. We need to serve a wider set of potential search visitors with the actual content on our pages.

That's going to be a challenge too. But basically we can say, "Hey, how can we group this stuff into content for SEO that's going to make for a meaningful, useful searcher experience and potentially has that ability to rank for all of these different combinations of terms that are closely aligned in intent?" That's kind of where we're going broadly with search, keywords, and keyword research and targeting.

All right, everyone, I apologize that Google keeps taking more and more useful and functional data and power tools away from us. I wish there were more that I could do to stop them from doing that, but it's not my place. Hopefully, this will help out your processes.

I'm sure there'll be some great comments and suggestions in the comments around other things people are doing and can do. We should all get ready for this change. I hope we'll see you again next week for another edition of Whiteboard Friday. Take care.

Video transcription by Speechpad.com