5 5 CATEGORIZACIÓN PARA EL ANÁLISIS DEL ESTUDIO Para poder analizar los resultados que se pretenden obtener con esta investigación se ha
6. EXPOSICIÓN DE RESULTADOS.
6.3. ÁMBITO SOCIO SANITARIO Subirats y Otros/ as(2004) declaran que:
Finding an attribution model—a system that accurately attributes the contribution made by each and every one of a company’s
Target Audience: Metric: Executives Conversions B A
marketing tactics to leads and to revenue—is the holy grail for data-driven marketers. Some marketers are very close to realizing a model that works, but most marketers are using primitive forms of attribution at best, which typically place too much value on lower- funnel, last-click tactics and result in many companies allocating too much of their budgets on the wrong efforts.
Richard Roberts, senior vice president of sales and marketing at B2B digital marketing agency BusinessOnline, defines attribution this way:“allocating proportional credit to all marketing commu- nications across all channels that ultimately lead to the desired customer action.”
There are three basic attribution models.
Thefirst islast-click attribution. This model gives 100 percent of
the credit for a conversion or a sale to the last marketing element the prospect interacted with. Oftentimes that marketing element is a lower-funnel tactic, such as e-mail or paid search. The deficiency in this model is that it gives too much credit to the lower funnel without recognizing the contribution made by branding efforts that got the prospect in the funnel in the first place or nurturing and education efforts that moved the prospect deeper into the funnel. Or, to return to the baseball analogy, this would be like giving a grand-slam home run hitter credit for all four runs even though the three previous batters needed to get on base to score when the fourth batter knocked it out of the park.
Roberts doesn’tfind last-click attribution to provide an accurate picture of customer behavior, because it ignores much of the buyer’s journey and other reasons. “It’s the most common approach, but we at BusinessOnline don’t even really categorize this as attribution, in fact,” he said. “We find it can do more harm than good by overvaluing certain types of interactions.”
The second type of attribution model is rules-based attribution.
With this model, a marketer assigns a certain value to particular tactics based on predetermined rules. For a prospect that inter- acted with three different tactics before becoming a customer, a marketer might assign each tactic equal credit. If that prospect
interacted with display, e-mail, and search, each of the tactics would be credited with driving one-third of that customer’s reve- nue. While this approach is more sophisticated than last-click attribution, rules-based attribution often mirrors a marketer’s own prejudices about what tactics work rather than reflecting the actual influence of various tactics.
“This kind of rules-based approach doesn’t really have the analytical rigor that we’d like and may actually not be much of a better approach than the first one—the single touch kind of measurement,” Roberts said.
The third and most accurate attribution model is algorithmic attribution.“The third approach assigns values to each interaction
based on statistical regression or probabilistic models,” Roberts said. “These data-driven models typically provide the most accu- rate picture of the customer journey.”
In part, this method examines how a marketing tactic contrib- utes to conversions, leads, and revenue by examining data from ad platforms, search engine optimization (SEO) tools, web analytics, paid search tools, marketing automation software, customer rela- tionship management (CRM) systems, social measurement tools, and even offline events and tactics (such as TV, print, or radio ads). Because most of these tools are digital, this performance can be examined at scale and essentially in real time.
This kind of attribution requires some grunt work, some statis- tical know-how, and an openness to data. Roberts said some marketers aren’t always comfortable with the conclusions this kind of attribution model reaches.“Sometimes marketers actually are uncomfortable with knowing the truth—with having to give up their data to a model they don’t understand and not being able to add their gut feeling input on decisions,” he said. “When I say uncomfortable with the truth, I mean sometimes we’ll have results come out of these stat-driven models, and our clients just won’t like it. They’ll say,‘Well, geez, I just feel like, you know, such and such has been my workhorse and you’re suggesting that it’s a complete waste of money. I just don’t believe that.’”The truth—in
marketing as in life—can often be a hard pill to swallow, and this can be a significant reason why some organizations don’t drive hard toward data-driven models.
But Roberts is a believer. “It gives you a powerful data-backed methodology for looking back on anonymous visitor activity and then linking it to known leads or sales,” he said.
The first step in algorithmic attribution is identifying anony- mous visitors to a corporate website. These visitors can be identi-
fied with a cookie or other tracking technology with the goal of tracking their path to purchase.
“Once those visitors who were anonymous early in their journey become known, when they register to download something or they interact with your sales team, it gives you a full picture and ultimately the ability to measure which activities, campaigns, or even content assets are best leading to sales,”he said.“It allows you to optimize programs based on revenue rather than front-end metrics, like visits or leads.”
To build this statistical attribution model, a company would look back on its marketing data and correlate all of the activities that drove sales. The company looks back for a time period equivalent to its typical buyer’s journey—as long as 18 months or two years for some companies. Based on the data, the company assigns higher values to the marketing tactics that are meaning- ful—the ones that drove conversions or revenue or both.
This kind of attribution is not theory. It is in practice right now. At DocuSign, Inc., which provides digital transaction manage- ment services, the marketing team is using a fairly sophisticated attribution model. The company generates about 130,000 leads per quarter. DocuSign has built a system that scores leads, ties them to a particular campaign or tactic, monitors conversion rates, and connects the leads to revenue. Ultimately, this approach enables DocuSign to prioritize specific leads, so that salespeople are following up with only those most likely to make a purchase.
“It’s really important that we are measuring the leads that come in effectively, so we can prioritize whom to talk to, and so we can
make it a lot more efficient for our sales team,”Meagen Eisenberg, vice president of customer marketing at DocuSign, said.
She explained the philosophy behind DocuSign’s approach:
“This is a closed-loop model. We score it. We tie it to a campaign. If it’s converting, we see that in the results, and we’ll adjust the score accordingly. It’s all about continual improvement and the learnings that come with that. Lastly, what really makes marketers today is their ability to measure and show the attribution and spend effectively.”
The system also analyzes leads based on their business demo- graphics and their level of engagement.“We’re taking that knowl- edge and putting it back into our scoring systems. So we iterate on our lead scoring model at least once a quarter,” Eisenberg said.
DocuSign acts quickly to kill underperforming campaigns.“If I see a program or spend not working, I’m going to either stop it midcampaign or reallocate or both,” Eisenberg said. “Then I am going to double down on the ones that are working.”
Eisenberg is extremely confident in the capability of data-driven marketing and solid attribution models to prove that online marketing spending is working. “I feel pretty confident in the metrics, and the tracking, and the technology we have today to prove what spend is working and what spend is not when it comes to online marketing and am excited to see the improvements that technology will bring,” she said.