Measurability has made search engine marketing. In large part, its success as an online medium was built on two factors: One, everybody searches the Internet, and more and more they’re doing it before making a purchase. And two, the results of using those searches to generate ad placements can be tracked and plotted in almost any way a marketer could wish. Most want to take their scrutiny down to the level of return on investment (ROI), to know how many orders, sign-ups or impressions they’re getting for their search ad spending.
But how do you migrate that measurability advantage into the world of brand advertising, which has its own traditional (and very different) measures for judging ad effectiveness? Those two sets of measures don’t combine very well, any more than tools designed for the metric system can efficiently work in inches. Search marketing has its generally accepted toolkit—clickthroughs, conversions and return on investment (ROI) –with which it has been hugely successful. Brand advertising has another: gross rating points, reach, awareness, brand lift. It’s almost easier and more comfortable to simply keep the two tactics separate.
Interactive marketing agency Performics has set out to solve that reconciliation problem—and they’re getting pretty close, according to Cam Balzer, director of search strategy at the DoubleClick subsidiary.
Speaking at Search Engine Strategies New York in early March, Balzer told an audience that figuring out the proper way to deploy and measure search is important in large part because more people are searching every day: about 60 million Internet-enabled Americans by September 2005, 55% growth over the number who moused to the query box daily in June 2004. “Search is becoming synonymous with being a consumer,” he said. “At Performics we’ve been looking at how to leverage this propensity and use search to extend the brand.”
Research from Yahoo! and others has shown, Balzer said, that search can be used to create consumer awareness of a brand and to get an advertiser’s brand included in a shopper’s “consideration set”, or the list of brands to be chosen from in making a final purchase.
Performics has set out to run some test campaigns about the interaction of brands and search, using a few of its more brand-oriented clients. In an interview after SES, Balzer said that his company’s clients have been asking for a way to bridge the gap between search marketing’s clear-cut ROI metrics and the ways marketers measure the impact of brand ads.
His conclusion: Search marketing can indeed be shown to have a measurable effect on brand-ad campaigns. But finding the right way to integrate search into a branded campaign involves going outside the strict ROI calculations that have been search’s traditional strength.
“To do this, you need to take a new spin on search,” he told his audience. “You can’t think of search as a purely ROI-driven mechanism and only expecting to directly track sales that come from your efforts.” Marketers who use search to help their brands need to take a broader view of search, he said. The metrics that can show search impact on branding are proxies, Balzer said; but then many of the measurements used to monitor branding, such as “brand lift”, are themselves proxies for watching consumers move through a conversion funnel as they narrow their choices.
One test campaign Balzer outlined looked at the costs and benefits of building the visibility of a consumer electronics brand by marketing to high-volume (and relatively high-cost) category keywords—terms as generic for the product categories involved as “electronics”. Most ROI-focused search campaigns find that these generic terms are too costly and produce too few conversions compared to their high clickthrough volume.
Performics launched SEM campaigns for four of the brand’s electronics categories and used 10 to 20 high-volume generic keywords in each product category, bidding on the terms to produce a top ranking for six weeks. (Performics let a fifth product category go without search marketing, to serve as a control group.) The ad copy and the ad landing pages were aimed at brand awareness and features rather than specific offers or calls to action.
The agency tracked return on ad spending both for the generic keywords in the campaigns and for the total campaigns in those categories, including sales conversions that came from clicks on branded keywords—either the vendor name or the product name or model. Sales from search in those categories were then tracked against sales for the same period the previous year.
Performics found that while advertising with the generic product keywords did not pay for itself, the overall campaign seemed to benefit greatly from including generic keywords. Total search sales in the product categories increased by 84% year-to-year, and sales coming through the advertiser’s one-word brand name increased 100%. And although the generic keywords alone produced a return on ad spending (ROAS) of less than 1:1, so that spending $1 on keyword bids produced less than $1 in sales, ROAS for the overall program was much better and approached the 10:1 ratio the advertiser had targeted.
Performics also tracked sales in the five product categories, looking at online sales in the categories in which a branded search term was the last click before the sale. The agency found that sales in the four product categories that had used generic keyword marketing increased an average of 103% over the same period the year before, with sales in one category rising a full 200%. By contrast, sales in the category that did not use generic keywords as part of its search campaign were up 34%.
In other words, Balzer said, people were searching on the generic keywords, and if they saw a search ad associated with those generic terms, adding consumer electronics advertiser to their consideration set. They then kept shopping, did a later search for the client’s brand name, came through and bought the product that they initially saw through the generic keyword.
“We saw very clear lifts on a category basis for the categories in which we were getting this increased visibility,” he said. And those heightened conversion results made the added cost of bidding on those generic keys a feasible strategy for a brand advertiser looking to grow search marketing results.
In another brand/ search investigation, Balzer said Performics investigated search performance at a multichannel apparel retailing client who occasionally tested the same high-volume generic keywords for its categories. By mapping the conversion rate from specific branded keywords to the number of clicks from non-branded generic terms such as “shoes”, Performics discovered that for this client, 10,000 non-brand impressions increased the branded conversion rate by 0.01%– a small move of the needle, but worth the effort in this case to build conversions.
In some specific cases, advertisers are looking to make consumers aware not of the brands they make but of those they sell. That was the case with a mass-retailer Performics client that wanted to make customers think of buying a certain brand from their stores. The retailer bought the first position on both Google and Yahoo! for the term “KitchenAid” for four weeks going into a peak holiday shopping period.
The test garnered half a million impressions and 17,000 clicks for a healthy clickthrough rate of 3.5%.
Performics then wanted to check the linkage for consumers of the “KitchenAid” search term with searches on the retailer’s brand name. The firm did this using the Yahoo! Buzz Index, an internal Yahoo! measure that is usually used to reveal the lift to searches from display or banner ads by showing the number of searches on a given term before and after a campaign.
Yahoo!’s Buzz Index can also measure search overlap—that is, the number of people searching on one term who also search on another term in the same week. In this case, Performics used the index to reveal the percentage of consumers who searched for both “KitchenAid” and the brand names of mass retailers in the same category as their client. They measured this overlap at two points in time, just before and just after the keyword campaign– in effect, measuring the degree of association between the search terms among consumers.
Of nine competitors measured for that week, Performics’ client had the second largest increase (108%) in those linked searches for “KitchenAid” and its brand. “We were actually able to double the connection in the mind of the consumer between the word ‘KitchenAid’ and this client’s brand name,” Balzer said. One interesting thing about this measurement is that it makes irrelevant any seasonal uptick in buying or variations in offline promotions, because the overlap was measured against eight competing retailer brand names. “We got the client a sizable lift that couldn’t be attributed to anything but the KitchenAid search campaign,” Balzer said.
Test campaigns like these from Performics are all aimed at giving marketers some of the tools they need to use search marketing as more than clicks and conversions. “You need to broaden your perception of the value you’re getting from search,” Balzer told SES attendees. “Search to date has built an enormous industry out of the direct-response model, where you track from click to cost, and that gives you an ROI, so you don’t need to look any further.”
“We need to move from thinking about search as just a highly efficient sales channel to viewing it as a highly effective marketing channel for engaging consumers throughout the consideration process.”
But as search gets more competitive, and as marketers try to do more through search, broader metrics than just clickthrough rates and ROI are needed, whether for measuring offline sales, latent sales that occur after a cookie has expired, sales on a home computer after searches done at the office, or in this case, the impact search marketing can contribute to branding campaigns.