Measuring campaign performance: Attribution models hit the spot
In the ongoing quest to improve performance, organizations have sought reductions in process cycle times, enhancements in supply chain efficiency, greater quality assurance and many forms of cost savings. Recently, interest in obtaining better measures of advertising campaign performance has shown a noticeable uptick.
“Organizations have identified the low-hanging fruit for productivity, and are looking for other areas of performance improvement,” says Jonathan Kraft, research program manager for business excellence at the American Productivity and Quality Center. “Many of these initiatives are related to consumer behavior because of the formidable power consumers wield.”
Marketers spend billions of dollars each year on advertising to reach consumers. eMarketer estimates that media advertising spend will approach $180 billion this year. Although television ads still account for nearly half of expenditures, digital media advertising is expected to be a close second within a few years, driven mainly by growth in advertising on mobile devices. Despite the heavy investment, though, advertisers have struggled to determine which dollars are actually producing results.
The multichannel customer experience has only made the job more difficult. Consumers move from one channel to another during a ?purchase; typically, they do a significant amount of research online, but ultimately, they often make purchases offline, either by phone or in-store. Consumers may also click on a display ad, read some reviews, comparison shop and then buy online.
“The customer purchase path is not linear,” says Tina Moffett, analyst for the customer insights team at Forrester. “It is very complex. In addition, a tremendous amount of data is now available, but it is difficult to make sense of.”
Some of the more rudimentary attribution models use the “last click” approach to assign credit, but given the multichannel user experience that is typical of the purchasing path, that model is simplistic and does not reflect the consumer’s behavior. Other models divide credit equally among all touchpoints with which the consumer engages. Those models also do not accurately weight the impact of each channel.
Fractional attribution models use sophisticated algorithms to sort out the effect of each form of advertising and assign credit. They are considered the most accurate ones currently available. Those models can account for the impact of ads at different stages of the purchase process, for example, which might begin with a display ad that generates brand awareness, continue with a keyword search and proceed through conversion (purchase).
PC manufacturer Lenovo was using a model in which various channel marketing departments each claimed a percent of credit for sales, but the data was siloed and added up to much more than the actual sales. Wanting a data-driven approach, Lenovo began analyzing sales using Adometry (adometry.com), an attribution analytics and marketing performance management solution. Adometry ingests data from advertisers’ existing systems, such as ad servers and site analytics platforms.
Adometry then analyzes all the incoming data and looks at the paths within a given client’s data set. “We analyze the data by using the full marketing path,” says Natasha Moonka, product marketing manager at Adometry by Google. “Our algorithm allows us to evaluate all the various combinations of marketing events customers encounter to understand how significant each touchpoint was in driving a conversion.”
Moonka adds, “Adometry uses complex algorithms combined with data-driven machine learning to process upward of 6 billion touchpoints per day. In addition to incorporating all channels, we also take into account other advertising variables like frequency, order and placement.”
The output of that data analysis is provided in an interactive user interface through which the Adometry customer can see the results and make future decisions to optimize the effectiveness of their campaigns while increasing marketing return on investment.
“In general, the increase in the return that marketers see on their marketing campaigns is vastly greater than the cost of Adometry’s solution,” Moonka says. “Typically, the cost of the solution accounts for less than five percent of a brand’s total marketing investment, and produces many times more than that in additional revenue.”
Besides providing results in a user interface, Adometry can also export data to enterprise business intelligence (BI) systems and other marketing tools for further analysis. “We have customers who have been using Tableau, for example, and if they need to continue using those systems, we can work with them to enable this,” Moonka explains. “More often, they choose to have their BI data exported to Adometry for analysis, but in heavily regulated industries, their security or organizational infrastructure requirements may mandate keeping some of the data (like revenue or costs) within their enterprise system.”
In Lenovo’s case, the company was able to make tangible improvements in its advertising strategy as a result of the analyses carried out by Adometry. For example, it verified that most of its revenue was coming from multitouch conversion paths, but that direct navigation and search accounted for the largest amount of revenue overall. Lenovo also discovered that a comparison engine it had considered eliminating was a high performer for conversion. In addition, branded display ads had a greater role in moving consumers toward conversion than had been realized, and they are now being used more effectively in online and offline marketing strategies.
Spending money wisely
Launched with a video that went viral, Dollar Shave Club ships razors and other personal care products directly to consumers. Initially focused on digital advertising only, the company eventually tried some offline ads, starting with radio. As its product line expanded, the company also began considering television ads. Given the cost of the new advertising venues, Dollar Shave Club wanted assurance that the dollars would be well spent. On the advice of its television agency, Media Design Group, the company began using Convertro to evaluate the return on its investments.
Convertro’s software originated as an in-house solution for attributing sales to different advertising media, and then evolved into a commercial product. Its data-driven analysis relies on several algorithms. “Our first algorithm determines the weight to give each marketing channel as the conversion process takes place,” says Jeff Zwelling, co-founder and senior VP of Convertro, which was acquired by AOL in May 2014. “We use behavioral information to measure the impact of each channel using a regression model, billions of data points each day.”