Earlier this month, Travelocity’s chief marketing officer Jeffrey Glueck generated a bit of a stir in the search marketing world after giving a speech saying his company found only 4 percent of non-branded search terms brought in bookings. Problem is, Glueck now says those figures weren’t reported correctly. As it turns out, nearly 25 percent of Travelocity’s conversions come from non-branded terms. But the big "assist" that non-branded terms are sometimes said to give to branded terms wasn’t happening for Travelocity. It might happen for others, but Glueck’s main challenge is for marketers to better measure assists and know for themselves.
"I think search marketing is a fantastic channel. My plea is for companies to get more precise about measuring it, to bid smarter. Because if they don’t, they can lose their shirts," he said in a comment left about his speech at Search Engine Land, where people have been discussing it. Glueck also clarified more about the study, which I’ll dive into below.
Glueck had spoken about Travelocity’s study at an IAB conference earlier this month. Why a Strong Brand Is a Search Marketer’s Best Asset from AdAge was the original coverage from that, reporting:
[Glueck] tracked Travelocity’s visitors coming from paid search and found that 65% of people only interacted with paid search once, via a single keyword. He found 27% interacted with paid search multiple times — but just searched repeatedly on the same term. Only 8% searched multiple times with different terms.
The final findings? Only 2% of paid-search conversions fall into the category where the searcher originally clicked on a nonbranded term only to click and convert on a branded term at a later date. Travelocity now credits nonbrand phrases as being responsible for only 4% of each booking attributed to branded search.
"For us, we can take a loss on nonbranded terms like ‘Hawaii vacation’ – but not much of a loss," Mr. Glueck said. Ultimately, he said, it is a "profound mistake by all of us to think we’ve figured out how to measure ROI on search. We’re in stage one."
There were several inaccuracies and mis-quotes in the original AdAge article, and AdAge has been kind enough to issue a correction.
The AdAge correction clarifies:
This story incorrectly implied that nonbranded keywords convert only 4% of searches. It should have stated nonbranded keywords are responsible for only 4% of the profits from branded paid keywords. Of the bookings on Travelocity attributed to search, about 80% are attributed to branded terms. Of that 80%, 4% should be credited to earlier clicks on nonbrand terms. Mr. Glueck also cited a Nielsen NetRatings study, and it should be noted the study, published in November 2005, specifically addressed travel-related searches in Google and Yahoo. The study found 47% were composed of 100 terms, and most of those were brand terms.
To spell it out further, Travelocity is saying that 20 percent of its bookings come from non-branded terms. Of the other 80 percent, where Travelocity buys a branded term, 4 percent of those get an "assist" or a click that converts after someone made an earlier visit via a non-branded term. For example, someone might have done a generic travel search, then later searched for a specific travel brand and made a booking. The branded search gets the credit for the sale, but a non-branded term helped.
So non-branded terms are important. On their own, generating 20 percent of conversions is significant. Add in the assists, and you get to 24 percent of conversions happening from non-branded terms. But still, I think many assume non-branded terms generate more than a 4 percent assist rate when it comes to helping branded terms to convert.
Indeed, search marketers often talk about a "search funnel" that happens. Someone searches for "new york" when thinking about traveling there. Then they get more specific — "new york hotels" — when they decide to stay. Then they get brand specific, "new york hilton," when they’ve found a specific place they want to say. The searches go from wide (the top of the funnel) to narrow (the spout of the funnel).
Q&A: Travelocity’s Chief Marketing Officer Explains Why Brand Terms Are Better from AdAge is a follow-up article with Glueck that gets into this more, done with Glueck after his speech. Is the funnel real?
We then studied whether people are starting on a nonbrand term like Hawaii vacation deals and, through multiple searches on a search engine, narrowing down to their favorite brands through multiple paid clicks and then, in some magical way, always going to the brand name and clicking on the paid link before buying. That is the funnel theory. If that were the majority of purchases, then the portfolio theory would make sense. We’d be OK to lose money on Hawaii vacation deals and make money on branded search terms.
This January we got the first detailed web analytics study from Travelocity.com about multiple clicks. Last-click measurement is very popular, and it’s what has for the last few years allowed search engines to say they’re the most measurable form of media in history. Travelocity’s point was that you should do the research on your own brand and nonbrand click behavior and don’t let brand profits fool you into overbidding on nonbrand terms. … Every advertiser sets their own threshold based on their conversion and profit level. We weren’t saying search wasn’t measurable. We’re saying advertisers should get more sophisticated and accurate about how they measure it.
About 4% of the bookings that occurred on brand terms should actually be credited to an earlier search on a generic term like Hawaii vacation. That 4% was less than the 12% that other companies had reported in a 360i study last year. Every company will have a different number, but the point is, we didn’t see any evidence that the funnel theory was the majority of bookings. Most people who bought with us only clicked one paid search ad in the 45 days before booking. About 76% of bookings from search were based on a brand search. … Thanks to that 4% funnel, we could lose a couple dollars on each booking from nonbrand phrases and still feel good about our spend.
All of these figures apply to Travelocity in a given period of time. Other marketers should do their own research in the same way. You have to calculate an accurate "assist percentage" for your business. Every business is different. … Small companies without a strong brand will probably gain more exposure through generic searches. But they still have to do the math and decide if they’re getting profitable business or not.
Overall, interesting findings, and it’s hard to disagree that anyone should take a close look at how their own site is converting. Some may find the funnel effect is stronger; others might find they are more like Travelocity and want to take a harder look at their purchases.
As I said, Glueck’s comments sparked some reaction on Search Engine Land. Here’s a close-up on his replies to some questions raised (see our original article to view all the comments on the story):
1) I never said that buying non-brand search terms was a waste of time, or that it’s impossible to measure. Far from it. At Travelocity we spend half our marketing budget in online and search. My point was that search marketers should get more precise about calculating "assist" percentages, and more accurate in their ROI methods. As the comment from SearchQuant agrees above, I was simply saying that assist percentages only modified our ROI by a few percentage points– same as SearchQuant calculates.
2) We calculated that 4% of brand bookings should be attributed to earlier nonbrand clicks. Based on better "assist" measurement, we think non-brand terms drive about 24% of our search bookings, rather than the 20% indicated by last click measurement. Every SEM campaign will be different, so you have to calculate your own figures. I was very clear everyone should do their own research, depending on their own unique situation. AdAge misunderstood the 4% assist and jumped to the conclusion that 96% of bookings were from brand terms, but the math is closer to 76%.
2) The Nielsen NetRatings study I cited was for the TRAVEL category, not the whole internet. It indicated 47% of all searches came from 100 terms, the majority of which were brands.
3) On the comment by Mr.Greitzer in regards to natural SEO, I would simply say that thanks to click tracking, buyers who come back to a site thanks to SEO links would still be attributed to the original paid CPC click (within a long lookback window), so the problem is not in natural clicks being missed, but rather that last click tracking has problems with multiple paid clicks.
4) As to the comment by DanielR that my October 2006 speech at Shop.org was a change in tune, that’s an easy misunderstanding as well. In that speech, I cited the authors of a 360i published study which looked at a large number of large etailers (not including Travelocity) and noted that 12% of their brand profits should be attributed to earlier nonbrand clicks. I invited the audience to do their own studies on their own circumstances. Travelocity upgraded its web analytics to VisualSciences and we finally in 2007 were able to do our own study on clickstream for the first time, and that’s where the 4% figure emerged. Every company will be different.
Regardless of whether the assist is 4% or 12%, that’s a long way from the funnel/clickstream theory that the majority of generic term searchers come back via brand terms to buy. That was my key point, and it remained unchanged. There is some lift/assist, absolutely, just not a huge one.
5) As to the comments that focusing on landing page optimization and conversion is the key to making paid search work, I couldn’t agree more. I think we’re a strong site– or we wouldn’t sell over $10B of travel a year–but we absolutely can get better. And that will help our non-brand profits.
6) As to the comment above that offline marketing is completely unmeasurable, I respectfully disagree. Those who heard my speech know that I emphasized multivariate regression models across all channels. My point is that nearly ALL marketing is measurable over the long term, and that unified models will actually measure MORE ACCURATELY than click tracking in some sense, because they get at the inter-relationships between TV and Radio and search.
In sum, I think search marketing is a fantastic channel. My plea is for companies to get more precise about measuring it, to bid smarter. Because if they don’t, they can lose their shirts. And I suppose, I have a second plea: To read my comments in the proper context.