Digital marketing can find and use a “key performance indicator” (KPI) to measure each step in the sales journey, including detours, diversions and pit stops. But using every conceivable measure to get a better sense of the customer is no better than pouring every conceivable topping over ice cream. It makes a big mess and will make you sick if you try to eat it.
Online, there are endless articles with endless lists. Do you need 37 KPIs? How about 43? Not enough? Too many? Perhaps the eight “must have” measurements are fine?
So which ones do you pick? Just as one size does not fit all, one particular set of KPIs will not give you the best picture of how well your marketing effort is going online.
Know yourself before you look for KPI
A company that is eager to use metrics must know its business. If not, KPI just becomes a bunch of letters and the data becomes smoke and nonsense.
“I started out in marketing long enough ago when there wasn’t so much data available,” mused ex-consultant Jen Calise, CMO and co-founder of Fishbat, a NY-based agency. The clients were C-level executives who were very bottom-line oriented, a trait that is missing from many digital marketing backgrounds. “Forget about access to data for now.” Calise said. What is the business goal?
KPIs can show traffic, likes, and engagement. “All that stuff looks successful on paper, but it has zero value for what the company is trying to accomplish,” she said. “Forget about data and go back to the business fundamentals.”
Some “inherit KPIs from a predecessor or pick them up at a conference,” noted Jeff Allen, senior director of product marketing at Adobe Analytics. “Moneyball taught us that metrics don’t matter until you understand what is contributing to that understanding.”
“Metrics lead to that…we tend to measure all kinds of different things that are not necessarily the right thing to measure,” Allen said.
Know what you can see…and can’t
“A vertical approach is a good place to start,” said Robert Schwartz, global leader for agency services at IBM iX. “Different sectors have different inputs and the KPIs should be adjusted accordingly,” he explained. “One does not apply one set of KPIs and one industry view across the board,” he added.
If the marketing campaign involves consumer goods, one could drive awareness through coupons. If it were software/SaaS, the focus would be more on what is driving toward the transaction.
Still, when one drives a campaign across channels — and devices — it is very difficult to attribute actions to users, Schwartz said. And if something cannot be seen, it cannot be measured by a KPI.
Adobe Analytics can measure across mobile, web, apps, recognize all devices, understand searches, e-mail and set-top boxes. Yet one can take all these metrics and still overload. And it is easy to miss the gaps, like measuring web traffic and not doing the same with mobile, Allen explained. “The way to understand the KPIs is to understand the journey itself,” then look for the KPIs. In fact, the KPI’s have to “nest” so that it can be rolled up into a greater sum, he explained.
At the Hudson Integrated Web Agency, Saddlebrook, NJ, measuring the value of different brands will vary across industry lines. But the cost problem is the same: “Know the margin you get on the sale and find the fraction you can live with as a marketing cost, divided by the number of people you reach,” explained Stephen Luke, lead search strategist. “You always want as many visitors as possible”, he said, and conversions are very easy to correlate. A good CRM system can track where customers come from, thus avoiding misattributions.
Know what counts
“In order to get ROI, you need to provide lifetime customer value,” said Fishbat’s Calise. That means investing in a client relationship once the prospect is converted into a sale, all the while knowing what the costs are. Once a business knows what its goal is, then pull all the data that reveals consumer behavior, Calise stressed. “The most valuable KPI is goal completion.”
“The inverted pyramid is the best way to look at it,” Calise continued. Top line is the brand impression, followed by the percentage of people taking an action (clicking to the next page), followed by a qualifying step (now a prospect), followed by sales conversion. That last part is when a program usually falls apart, as there may be no linkage between marketing and sales to close the loop, she noted.
Placing attribution is in itself a major KPI, but it can be misplaced, Hudson’s Luke cautioned. Again, having a large number of visitors to a web site can be a good thing, but the opportunity for a conversion can slip away if web traffic goes to the wrong landing page.
Segmentation is another KPI, which can be restated as sales by state, as a heat map, showing the most transactions or grading transactions by their value, Luke noted. “For clients, it’s easier to segment as much as we can,” he said.
Using IBM as an example of a client experience, “we had to do a thorough analysis of where our margins come from,” Schwartz said. The Blue Mix development platform received a lot of online interest, but the process was anything but a single transaction. It was more digital, more competitive, and required more constant interaction than a traditional business, he said. IBM was selling a service, he stressed. That is not a one-shot transaction, like buying a book on Amazon.
Know what you see
“When we first started the agency, we lost our minds with technology,” Calise recalled. A lot of raw data was generated, but that wasn’t enough. One had to think analytically, and that means looking for the story in the data. For example, if traffic to a web site drops by 90 percent after the price of the product is shown, you probably have a price problem, Calise noted.
“I don’t think there are metrics that are trivial if they are gathered with integrity,” Calise said. It’s easy to create a “smokescreen of success” without doing a good job. “What good is racking up 150,000 “likes” for a product on social media if they were purchased from a click farm in Bangladesh?” she asked.
Adobe’s Allen was more succinct. “People have a narrative to tell and then search the data to support it,” he said. Let the data tell the story.
