July 11, 2013
I’ve been creating a lot of campaigns lately between getting new clients, standard keywords research for existing clients, and existing clients getting new products or running sales on seasonal products. With a tight deadline on these campaign builds, I’ve found myself wondering why I should even bother with phrase match keywords.
See, I build my campaigns by match type. I like to have my proven winners as exact matches with as unlimited a budget as I can get them. That way I know the keywords that work for my account aren’t being limited by budget. Then, I separate other match types into their own campaigns. Depending on the client and the campaign, that could be phrase, modified broad, and/or broad match.
For example: I’m building campaigns for one client that are in related fields, but aren’t exactly what my client’s site provides. That’s dicey territory. So, I really didn’t fancy broad match on these keywords. I knew I wanted exact matches of keywords I found that were sure to work, as they’d been converting in the account already. I also wanted a campaign to do some keyword research for me and impress for versions of my targeted keywords I hadn’t thought of, or just hadn’t converted in the account in the past. To do this, I decided on modified broad match. I set my bids and budgets lower on my modified broad match keywords, since I didn’t want them to get too competitive for search terms that aren’t exactly what my client’s site is about.
Now that I had my exact match winners and a nice modified broad campaign to perform keyword research and impress for keywords not exactly matching my exact matches, I figured I was all set. But then I thought, what about phrase match?!
I thought about this for a long time. I really wanted to be finished with each campaign build-out as soon as possible; was phrase match even worth building out?
As part of this consideration, I thought about how the match types work. Let’s say, for example, I’m selling bicycle horns. I’d, of course, want to bid on [bicycle horns]. But people might search for cheap bicycle horns, horns that fit on bikes, etc. A phrase match “bicycle horn” makes sense to capture cheap bicycle horns, but what about horns that fit on bikes? A modified broad match version of +bicycle +horns would capture that. For that matter, it would also capture cheap bicycle horns. So, why do we need the phrase match middleman?
An argument could be made that broad match captures everything, so why bother with any other match type, but we all know that broad match captures a lot of crazy search queries. I was once bidding on industrial steel table as a broad match term, and it triggered an ad for the search term patio furniture. facepalm Thanks, Google!
Since I’m a data-driven PPC nerd, I had to set out and analyze data to see how phrase match really compares to modified broad match. So, I picked a couple of client campaigns and ran some pivot tables.
You can see in this account, modified broad brings in more conversions than phrase match and brings in more clicks and impressions. However, phrase match has a better cost/conv., CTR, conv. rate, and conv./100,000 impressions.
I decided to pull out one keyword, exactly the same, in phrase and modified broad match to compare stats.
I can’t show you the exact keyword, but trust me when I say they were the exact same keyword with different match types. It’s also of note that it was a one-word keyword. As an example, consider the keywords +puppies vs. “puppies”. I would think these should function almost exactly the same, right? Looking over the stats, we can see the modified broad match version vastly out-performed the phrase match version.
Looking deeper, I found these two keywords were housed in the same campaign (read: competing for the same budget), but they were in different ad groups. However, the phrase match was not added as a phrase match negative to the broad ad group, so they were still at equal chances for impressing ads for the same search terms.
I started thinking about how these keywords are triggering search terms, too. How does Google decide to impress a search term against a modified broad match or a phrase match keyword if it matches both? So, I ran another pivot table!
The match type column in a search term report refers to how the search terms were matched to the keyword. So, an exact match of a modified broad match keyword means that it matched the keyword exactly. Then, this is where it gets super fun: close variants! That means a search term doesn’t even have to exactly fit your phrase match keyword, and it can still match, which even further blurs the line between phrase and modified broad match keywords. You can, of course, change this setting to NOT include close variants, but it looks like most of the account managers here at Hanapin use them.
Another issue is that this account isn’t clean in terms of data. We’re looking at keywords that aren’t present in all match types, no embedded negatives to prevent cross-pollution; it’s kind of a mess in there.
So, I decided to pick another account that looked way cleaner. Here’s the first pivot table ran again for this cleaner account:
This account actually ONLY has modified broad, exact, and phrase. We can see modified broad has more conversions and a better CTR. However, phrase match has a slightly better CPA, conv. rate, and CPC.
I dug deeper into this account, and we DO have embedded negatives. Oh, it’s so wonderful. Therefore, we know all the modified broad matches are NOT capturing the phrase match versions of the same keyword. So, let’s look at the search term pivot table for this account.
In this data, we can see the top three conversion rates belong to modified broad keywords being matched by exact, phrase, and phrase (close variant). Yup, close variants in this account, too! We also see the most conversion volume from modified broad, and the two lowest CPAs are from modified broad keywords, too, but at low conversion volumes.
At the end of all this analysis and thinking about these two match types, I’ve come to hold the opinion that phrase match can get a little bit cheaper leads for you when used correctly. So, if cost/conversion and/or budgets are SUPER important metrics for you, I would say for sure take the time to build-out nicely embedded negative phrase match campaigns. However, if you’re going for speed and you can spare a little on the CPA side of things without restricting budget, my opinion is to leave the phase and take the modified broad. As long as you get a good set of negative keywords and continuous search term reports to weed out new poor intent search terms, you’ll find you’re getting way more bang for your buck.