You’ve heard the famous quote about marketing budgets before: “Half the money I spend on advertising is wasted. The trouble is I don’t know which half.” This quote is attributed to John Wanamaker, a famous retailer and marketing pioneer from the 1800s.
The good news is that Wanamaker believed half his marketing budget actually worked. The bad news is that the problems with marketing attribution persist even today. After all, when your CEO asks for proof that the money spent on marketing is getting results, what can you say?
With traditional marketing media, accurately assessing and quantifying success was never easy. How much business could be attributed back to campaigns that ran in TV, print, radio or direct mail? Any answer to that question was largely based on guesswork, and the truth is, no one ever really knew with any degree of certainty.
But with limited budgets and a customer journey that now spans across multiple online and offline channels, it’s more important than ever that CMOs figure out how to allocate their marketing dollars efficiently and for maximum impact.
Which is why Korey Thurber, Chief Data Analytics Officer at Harte Hanks, says having a robust attribution model must be a top priority for every financial marketer today.
“If you don’t have clear visibility into what each component of your marketing spend is contributing to the bottom line, then you have a real attribution problem,” Thurber says.
Thanks largely to the rise of digital channels, the longstanding and elusive dream of marketing measurement and attribution is finally becoming a reality. Today, there isn’t really an excuse. Financial marketers should be able to tell their CEOs and CFOs how much their marketing investment contributes to the bottom line, and do so with a confidence.
In fact, with the right attribution model, Thurber says that today’s marketers should be able to tackle all the following issue with reasonable accuracy:
1) What was the incremental benefit driven by the marketing spend?
2) How much does each channel contribute relative to its overall cost?
3) Am I optimizing my marketing spend?
4) Do I see meaningful differences in channel usage (and therefore marketing attribution) for various customer segments?
“It’s all about giving credit where credit is due,” Thurber explains. “Once you have that piece of the puzzle figured out, you can make very informed, very strategic decisions. All the other pieces of the puzzle fall together. You can determine precisely how to allocate your marketing spend to get the most ‘juice for the squeeze’.”
Most financial institutions are still using “last touch attribution,” where the behavior being measured (like a purchase) is attributed to the last marketing channel the consumer saw or “touched.” And this is a big mistake, according to Thurber.
Thurber says it’s a significant problem because it yields a biased and inaccurate view. For example, let’s say the financial institution has several marketing silos — a digital team, email team, a direct marketing team, etc. If the direct marketing team sends out a campaign and Jane Doe opens an account online, then the direct mail team thinks its campaign should get credit for Jane’s purchase. However, Jane also received an email marketing message, and Jane ultimately converted online, so both the email and digital teams will try to claim credit for the sale as well. When each team reports its “success,” one sale can end up looking like three.
“Now you are counting the same purchase multiple times without digging into what actually happened,” Thurber points out. “It doesn’t tackle the fundamental question: What is really driving Jane’s decision to buy? Was it direct mail, email or digital only? Or did other factors come into play?”
Despite the obvious flaws with such a crude attribution model, Thurber says it is commonly deployed by many financial institutions. Why? Because everyone wants to make their department look good, so manager don’t feel much of an incentive to analyze sales more honestly and accurately. Unfortunately, fudging with the numbers in this way results in flawed justifications for spending marketing dollars in the wrong places and/or at the wrong level of investment.
According to Thurber, artificial intelligence has quickly become a driving force in this attribution space. Today, data scientists can write code that tells an attribution algorithm how to improve itself. Utilizing the principles of machine learning, the algorithm improves automatically and in real time as new data is collected.
“A good example of where this comes into play is in the digital space,” Thurber explains. “Beyond just saying can you attribute to direct mail or email, you can go deeper. You can now attribute to specific components served in that display ad — from the timing and placement of the message, to whether it was triggered by a search engine or a Facebook newsfeed. Through machine learning marketers can then discover what resonates best, with not just that one individual making a purchase but also others that share similar psychographic and demographic profiles and make real-time changes and adjustments.”
Picking the Right Attribution Model for the Modern World
The most significant impact of any marketing attribution model is how it informs investments in each channels, and points towards specific tactical optimizations that improve the performance of individual channels and media buys — both online and offline.
“At the end of the day, the attribution solution you pick should be all about getting the most bang out of your marketing bucks,” quips Thurber.
He says the right attribution solution can keep the marketing spend flat year over year while increasing the return on investment 20% simply by reallocating the existing marketing budget across various forms of media. But like any significant initiative, Thurber says this a true marketing attribution solution must be driven from the top for it to be effective, with consensus and buy-in across the C-suite.
Determining which attribution solution is the most suitable depends on the questions marketers are trying to answer, the types of media to be used, the overall marketing spend and the data available, says Thurber. Here are four steps Thurber says you can take to get started:
1. Perform an audit and build a plan. Audit current skill sets and technology infrastructure to show what needs to be different. Put a detailed plan together that details how you will get where you need to go, using a phased approach that defines clear milestones and success metrics along the way.
2. Emphasize the need for change, and use math to prove the point. For example, make the point that “by investing 2% of the overall marketing budget in developing a more accurate attribution model, we can tell you what the remaining 98% is actually doing.” But don’t stop there. Further explain that by making the investment, “we will identify at least 15% of ineffective spend, which can then be reallocated.” Find good case studies from outside the organization that provide proof that there is a benefit to your bank or credit union.
3. Make it clear why it matters to the end customer. Clearly show how a customer-centric focus — enabled by a robust attribution methodology — leads to happier, more loyal and profitable customer relationships. Again, find case studies to help prove the point.
4. Be tenacious to ensure the transformation sticks. Accurately tracking and quantifying the impact of marketing isn’t easy — it will impact many people throughout the organization, including frontline staff — but it’s imperative. You’ll need to implement training for employees so they understand what’s expected and how it impacts them. And don’t forget to create an onboarding program for new employees. Outline your vision, set clear goals, and implement metrics that can be used to provide incentives for employees. It doesn’t hurt to remind people why you’re doing this: that data-driven decisions are better decisions.
You can learn more about identifying and implementing the right marketing attribution model at The Financial Brand Forum 2017, where Korey Thurber will show you the different attribution models used today, which ones work, which ones don’t, and how to get your attribution solution off the ground.