The subject of broad match gone wild is a popular one in PPC, and has been since the dawn of search query reports. Search marketers frequently lament the irrelevant and sometimes downright puzzling queries which triggered their ads. In fact, better search query matching was one of my 2007 PPC wishes that still hasn’t come true.
A few weeks ago, I was doing routine search query report reviews for one of our international clients. We use broad match on their branded terms to cast as wide a net as possible, and we use extensive negative keywords to control the wildness. Anyway, I pulled a SQR for our Germany campaigns, and then pulled one for the US. Again, the task was all typical – but the report results were anything but.
We’ve created ad groups by match type for control and new search query mining, using the SQRs for not only negatives, but new positive keywords to add to our account. For both Germany and the US, I looked at just 2 keywords this time: the broad match and phrase match of the client’s brand. I noticed that reviewing Germany’s report took a lot less time than reviewing the US report. This came as a surprise, since our branded campaigns are set to “all languages” and I had to pore over German-language keywords in the SQR as a non-German speaker (Google Translate is my best friend for this). So I decided to compare the two reports.
What I discovered stunned me.
Allow me to illustrate with a few visuals.
Look at the total number of search queries: the US has nearly 3 times as many as Germany. Remember, this is on the same 2 keywords! That’s the stat that got me started on this in the first place. I find it hard to believe that people in the US are 3 times more creative than people in Germany when it comes to searching for the client’s brand (or searching for anything, for that matter).
This goes a long way towards explaining why our US CPCs are so much higher than other countries for this client. I know that the PPC market in general is more saturated here than elsewhere. If nothing else, there are more US-based advertisers. And our population is 3 times bigger than Germany’s (82 million for Germany vs. 311 million for the US), so I might accept the notion that if every person in each country conducts one unique search related to these 2 keywords, we’d see 3 times as many SQs in the US as in Germany. I think it’s a stretch, but it’s at least plausible.
But let’s look at search query distribution across match types. Remember, we’ve segmented our ad groups by match type, so there are no exact matches. What’s left in the SQR is broad, phrase, and session based broad.
A couple of visuals will make this easier. Let’s look at Germany first.
Half of the queries were broad matched, and the rest were pretty evenly distributed between phrase match and session-based broad match. I’m not thrilled about the high percentage of session-based broad matches, but that’s another post. Still, the fact that over 1 in 5 matches were phrase match isn’t too bad.
Now let’s look at the US.
Are you as speechless as I am? Fully 84% of the matches in the US were broad match (and remember folks, there were 860 of them, compared with 193 in Germany). There were virtually no session-based broad matches, so at least we have that going for us. But only 15% phrase matches, vs. 22% in Germany? Why, Google, why?
And here’s the kicker – you know this is coming – the US SQR is loaded with totally irrelevant queries.
Methinks something is rotten in the state of Denmark. (It’s close to Germany, right?)
Have you seen similar behavior in your international campaigns? Are we Americans really that much more creative in our searches? Or is Google showing their patriotism by fleecing us? Share your thoughts in the comments!
Melissa,
Very interesting. What is the breakdown of impressions? I’m sure that the US had FAR greater impressions. Is it possible that the elevated number of broad matched queries is merely a by-product of a much larger sample size (queries/impressions)?
No matter the cause – this is further proof that segmentation is key. If you are targeting PPC internationally, segment by country/region so you can report on them independently!!! Thanks for this post. ; )
Great question, John, and one I didn’t consider in my analysis. (That’s what’s great about sharing!) Anyway, the data shows the opposite of what you hypothesized: Germany had twice as many impressions as the US! Most of them were very close matches to the brand.
I’ll have to do another analysis and write a follow-up. I might need to do some more tweaking to the campaigns.
Either way, you’re absolutely right – segmentation is key! Thanks for your thoughtful comment!
Mind = blown. I’m left scratching my head…
Hey Mel,
A great puzzle of a post! I’ve reread it several times and, without looking at the keywords and search queries, can’t add much else:( Maybe Google’s keyword matching is more controlled in other languages because they are erring on the side of caution??? Maybe Germans are higher educated and more concise in their searches? Maybe if your brand keywords are in English, AdWords broad and phrase matching is more controlled for other languages? Maybe the same applies for German characters if they have any that are a non-US & UK based alphabet… I know Spanish is based off of a similar alphabet than our English, but have others or variations like: á é í ó ú ü ñ. Yep: Ä/ä, Ö/ö, Ü/ü. Good old Wikipedia: http://en.wikipedia.org/wiki/German_alphabet
However, this is another great reason as to ALMOST-NEVER use AdWords “standard” Expanded Broad Match.
I still anchor 99%+ of my broad keywords fully or at least dual: http://www.webranking.com/blog/pay-per-click-keyword-match-types-exact-phrase-broad-and-modified
Hey James – you’re right about the language thing. Ads & keywords in English-only probably do generate fewer search queries. And you’re also right about almost-never using regular broad match. I don’t either – except on branded terms, and even then only if the client wants to cast a wide net. Good food for thought!
I was thinking it may be different devices, where mobile searches may rely on more broad match, but don’t know enough about Europe or the vertical to bet the farm on this.
Also could there be any SEO side effects that would take some of the broad searches in Germany (hence the impressions) away from paid, yet in the US rely more on the ads?
Finally, is the brand term a German word that would open it up for non-brand searches there (leading to a better phrase response and poor broad match response)?
PS – Best. Line. Ever.
“Or is Google showing their patriotism by fleecing us?”
Ha Chris, glad you liked that! 🙂 I hadn’t thought about the mobile/device angle. I’ll add that to my follow-up analysis because that’s an excellent point. I’ll also look at how SEO plays in – I was focused on PPC-only in this analysis.
The brand is definitely an English word – but that’s a fantastic point also. Yet another reason to be careful about broad match!
Another thought similar to Chris’: Could the English word be a close misspelling of a German word? Wow. International PPC can get complex, can’t it?
Hi Melissa – I’ve seen this too when we are doing kw expansions or neg expansion for US and other countries (not just DE but lots of EU markets, APAC, Canada, and South America). On US account, we see all kinds of crazy SQs, 100s. On other non-US accounts that we expect to have enough data to do some decent kw expansion, we see maybe 5 new sometimes. We’ve seen this on “American” brand terms so maybe it has something to do with that, not sure.
As a percentage of SQs you are getting a lot of session-based broads in DE, versus those reported in US. Perhaps matching and SQ reporting have been more of a focus for US geo accounts than other countries for Google. They do tend to follow the money, when they sniff out an opportunity. 🙂
Hi Lisa, thanks for your insight. I’m glad to see it’s not just us, anyway! I also suspect that matching & reporting is more robust for US accounts than for other countries. The session-based thing is also curious. Thanks for your feedback!
Have you looked at this by search partners/google only?
Some search partners have a far larger broad match percentage and I also think (but can’t be bothered to verify right now – way past beer o’clock over here) that there is proportionally more SP traffic in the USA
Nope, that’s another segment to add to my follow-up. Thanks Richard!
So my thought on this is much less search focused and much more language/sociologically focused. Chances are people searching in Germany are speaking English as their second language. Having hosted a few German friends in the past, I can tell you that no matter how fluent a person may be, there’s still little nuances in the way that they speak that you can TELL its their second language. Little nuances like mixing words up, using less typical words for things we’d normally do etc etc.
I guess where I’m going with this is that I think they’re entering more basic queries than the US folk (who tend to type absurdly long/weird things into Google). That, and the percentage of English speakers relative to the population (who’d be getting matched to your terms) would be far fewer.
That said, I have a few clients who are running cloned campaigns overseas. I’ll take a look and will report back if my findings are the same!
Hey Melissa,
That’s an interesting phenomenon – thank you for sharing! As the German #ppcchat outpost I can say that this is definitely not common for AdWords over here. We sometimes get new features a little later than the U.S., but overall AdWords is just the same. What Google defines as “close variants” plays a bigger role in German, but this shouldn’t make a difference in this case.
I did a quick SQR analysis for one of our clients and found 20 session based broad matches compared to 4000 normal broad matches. I also found a case similar to your setup, where an American brand was used for a broad match keyword targeting Germany, but it didn’t result in as many session based matches.
As far as the number of queries goes, the fact that the brand is an English word could’ve lead to more irrelevant searches from people who meant something else. Germans probably wouldn’t search it as often in another context, because we would use the German term. Or maybe the brand is better known in the U.S., so that it is searched much more often in general. Pure speculation, of course…
By the way, can you share anything about traffic quality? Do those many session based matches make sense or convert well?
Martin, I was hoping you would chime in here! The session-based query number is definitely something I’ll be digging into, because I don’t think I’ve *ever* seen it that high.
As for traffic quality of the session-based queries, a few things jump out:
– Most of them are one-off (1 click, 1 impression) searches
– The total conversion rate is fine, but few of the one-offs converted (in other words, 95% of conversions came from a couple terms)
– Most of them also include German words, which tells me it was people looking for related info to the client’s brand. That’s not a bad thing. Problem is, I don’t know what most of the German words mean! 🙁
Thanks for your insight, it’s good to know what’s normal in other locations!
Mel,
This reminds me of the fact that Google goes where the money is. It is probably much the same reason why they “clump” all unknown geo-location of users into “US – undefined”. If they threw them into China – Undefined they wouldn’t get half of the revenue from it they can by shoving them on US advertisers.
I know part of the reason for the discrepancy is there is probably more “english” keywords they “could” potentially match to in the US than Germany, so a greater likelihood that they would. But beyond that I am sure much of it has to go back to queries without much advertiser competition where they then use broad match to fill. Remember one of their all important factors they report on in their financials is eCPM. Or the revenue per impression/search they are driving and how that changes. It is also the real reason behind why more previously “free” or “organic” products are shifting to paid. You don’t get revenue off of free!
Thanks for writing the post and taking the time it was awesome.
Great Article and Analysis! I have noticed the same thing in several accounts spread across multiple markets. I separate out all of my non-us campaigns by English and Local language and have seen similar trends across both. I wonder how long before Google adopts the “US Strategy” worldwide?