Online merchants tend to take the easy way out when assigning a sale to a given marketing channel, such as a search engine, a Web site or an ad. The sale is attributed to the last exposure a customer viewed or clicked.
But this doesn't take into account the value of exposures to a multiplicity of banner ads, Web site sponsorships, search results and e-mail campaigns. As a result, marketers tend to neglect the impressions that bring customers through the conversion funnel — the steps that lead to the sale, download, sampling or whatever the desired action might be.
To offer insight into the cumulative effect of online exposures, analytics and marketing agency Avenue A/Razorfish issued the research white paper “Actionable Analytics.” The study's media analytics section provides clues to determining a correct media mix.
In one example, Avenue A/Razorfish used data from a series of online campaigns run by a travel and hospitality company. Cookie data revealed that a mix of pure brand impressions and direct response initiatives was most effective in generating conversions.
Why use both? Consumers exposed only to a series of brand advertisements showed a greater propensity to buy based on the number of impressions they saw — up to a point. After eight brand impressions, if they didn't view a DR spot, consumers' likelihood of making a purchase fell off, perhaps due to ad fatigue.
However, consumers who saw several brand advertisements and one DR spot were nearly three times as likely to make a purchase.
The travel firm observed a similar pattern when prospects saw a series of DR ads with a brand plug tossed in. The DR ads were more effective in generating sales, but their effectiveness often doubled with one brand-based ad exposure.
The “sweet spot” was between two and five brand impressions, combined with six to eight DR efforts. Consumers exposed to this mix indexed much higher for sales than any other combination.
As the agency points out, the index score and sweet spot varies from one campaign to the next. Marketers must perform their own analysis rather than simply choose a rule of thumb from the list presented in “Web Marketers, Note….”
The initial question remains: How does a marketer avoid attributing a desired result entirely to the last impression made on a customer?
To answer this, Avenue A/Razorfish analyzed six months' worth of cookie-level data from another of its clients. It tracked impressions ranging from online display media, sponsorships and paid search campaigns, and observed clicks and actions taken, including purchases and behavior on the target Web sites.
By using regression modeling techniques to measure the influence of different efforts and various combinations of impressions, the company attributed the value of the last three exposures before a consumer took the action (see pie chart).
Simply allocating all sales to the last exposure would have attributed 41% of the sales to the wrong medium. In the case of a mixed-channel campaign, this would have dramatically underrepresented the value of, say, banner ads or search engine efforts, if the last exposure before a customer's purchase was an e-mailing. A media planner observing this likely would have disproportionately shifted dollars away from search to e-mail.
Of course, the path to conversion within a group of potential customers isn't as simple as the pie chart indicates. “Sample Media Path Leading to Conversion” demonstrates how a larger sample of individuals might wind up at a single conversion point, such as using a search query for a specific brand (“Chevrolet Corvette,” for example, instead of “car”) before making a purchase.
To understand what Avenue A/Razorfish saw, read “Sample Media Path…” from right to left. Start with the conversion medium — the branded search the consumer did before taking the desired action.
By working backward, marketers can determine which of a variety of channels — broad-reaching network sites that triggered a recollection of the marketer's offerings, a search for a specific product, viewing a Web site related to the product — were most effective in leading the consumer through to conversion.
Ideally, marketers would track media patterns that lead to all the channels that result in sales. The example shown represents only how consumers might end up conducting a branded search before making a purchase.
Yes, branded search ultimately led to a sale. But by stepping back, analytics provided clues to how the customer came to conduct that search.
Additionally, a properly designed dashboard can let a marketer view groups of customers by, for instance, sales margin rather than conversion alone. The path taken by someone who makes a high-gross-margin purchase might not be the same as that of most customers. Marketers intending to focus on higher-margin customers will choose the media paths appropriate for that target group.
NL
For more on ROI, subscribe to the MarketingROI newsletter by Richard H. Levey at http://subscribe.chiefmarketer.com/subscribe.cfm.
Web Marketers, Note…
Avenue A/Razorfish offers the following cautions about these analytic efforts:
- Don't use a single campaign to achieve both direct response and brand aims.
- Choose metrics carefully based on the goal — sales, customer acquisition, maximizing gross margin, a product trial, or another business objective.
- If necessary, create a custom media dashboard that links desired outcomes with promotions in a given medium and financial data. Don't settle for metrics that are close to, but not specifically, the ones that will be most useful.
| Exposures | ||||
|---|---|---|---|---|
| Fourth to last | Third to last | Second to last | Last | Conversion |
| Niche sites 50 | Direct sites 20 | Direct sites 30 | Direct sites 0 | |
| Network sites 30 | Network sites 30 | Network sites 50 | Network sites 0 | |
| Search 20 | Unbranded search 50 | Search 20 | Branded search 100 | |
| Hypothetical numerical values for these channels indicate how a prospect will arrive at the next step in the conversion process. | ||||
