Share on facebook
Share on twitter
Share on linkedin
transactional data for customer journey

Is Transactional Data the Secret to Accurate Customer Journeys?

Consider this. 

You can have $150 now or $180 in one month; which would you prefer?

If you opted for the former, you’re not alone. Most people take an immediate payout over a delayed greater reward. Ignoring any immediate cash flow requirements, passing up a 20% return on investment is an objectively poor choice. It’s irrational. Yet, people forgo the extra $30 time and time again. Why they do so is down to a quirk of psychology called ‘present bias.’

Present bias, also called exponential discounting, is the “inclination to prefer a smaller present reward to a larger later reward,” explained Anujit Chakraborty, Assistant Professor of Economics at UC Davis. It’s why people consistently under save for retirement, consume demerit foods, and delay important decisions.

Such unconscious errors in thinking and behavior (collectively called ‘cognitive biases’) are myriad. We habitually seek out and interpret new information to confirm our preexisting beliefs and options. That’s confirmation bias. We perceive past events as more predictable than they were. That’s hindsight bias. We value the first piece of information we acquire more than others. That’s anchoring bias. The list goes on and on.

Previously, economists assumed consumers acted rationally, seeking to maximize their personal utility or satisfaction. Research into cognitive biases dismantled this notion, proving that people are predictably irrational. The revelations sent ripples far beyond economics.

User experience was particularly hard hit. Researchers could no longer trust self-reported customer data as infallible or truth-seeking. Customers may request immediate mediocre features over powerful long-term ones (present bias), overestimate their reliance on the functionality they most recently used (recency bias), or go along with a focus group decision even though they disagree (bandwagon bias). That realization was destabilizing.

According to a survey from McKinsey, 93% of customer experience leaders use a survey-based metric like Customer Satisfaction Score or Customer Effort Score as their “primary means of measuring CX performance.” But only a fraction of those (15%) report feeling fully satisfied with how their company is measuring CX. Worse, just 6% believe their measurements enable both strategic and tactical decision-making.

Faced with dubious self-reported data, customer experience professionals need a complementary data source. Information and insights that could corroborate and clarify existing data, remove bias, and deliver a firmer foundation for customer journey design. Many have discovered that source in transactional and operational platforms. 

While self-reported surveys and questionnaires are subject to bias and partiality, transactional data delivers a crystal clear snapshot of reality. The difference lies in atomicity. The latter data source contains a multitude of tangible, spatial qualities. You can map transactional data to components like location, duration, interaction range, and value. When feedback is supported with transactional data, it corroborates the explanation. 

For example, customer surveys report whether people say they like a product or service. But their transactional data tells the unadulterated version: Were they satisfied enough to put their money where their mouth is?

transactional data for customer journey

Introducing transactional data to customer journey research is challenging. Customer feedback is clear, concise, and easily interpretable. Transactional data, however, is value-neutral. It not only requires businesses to step outside of typical feedback channels but also demands extra time to analyze and interpret the information.

But for those who can broaden the scope and surface area of data collection and analysis, the results are immense. Organizations can cut through bias and partiality to deliver customer journeys that reflect fact, not fiction.

Step 1: Clarify success criteria

“If you can’t measure it, you can’t improve it.”

Peter Drucker’s management mantra is often repeated—and for good reason. It warns people away from experimentation for experimentation’s sake and forces them to take an analytical approach to management. In customer experience, evaluation and measurement are essential.

“Without a quantified link to value and a sound business case, such [customer experience] efforts often can’t show early gains, build momentum among functional executives, and earn a seat at the strategy table,” wrote Joel Maynes, associate principal, and Alex Rawson, principal at McKinsey. “They stall before they ever really get going.”

To create an ironclad link between action and value, analysts at Sense Corp recommend focusing on three groups of metrics:

  • Customer-centric metrics: Session duration, first call resolution, average speed to answer, on-time appointments, etc.
  • Business-centric metrics: Revenue, productivity, profit, etc.
  • Segmented metrics: Customer, product, geography-segmented NPS, CSAT, CES, etc.

But this is only one yardstick. Large organizations often opt for ready-made evaluation tools like a Balanced Scorecard (BSC). It’s a sort of management dashboard displaying the most important metrics across financial, operational, innovation, and customer perspectives. On the other hand, smaller entities are more likely to define success more narrowly—an improvement to NPS, session time, or similar measurements.

Although these tools explain where to look for metrics and indicators, they do not explain how to set targets or thresholds. The reason is clear: Success is relative.

Brittany Davis
Brittany Davis, Head of Data at Narrator

“People often ask whether benchmarks are good or bad,” says Brittany Davis, former IBM and WeWork data scientist, turned Head of Data at Narrator. “Those quick solutions don’t really work because you always want to consider the nuance. The important thing is to track impact over time.”

Although the recommended time frame depends on data variance and volume, the underlying message is the same: Don’t base decisions on a single point in time. Instead, study effects over an extended period. 

Davis recommends researchers search for stability and consistency across three time periods: an improvement for three weeks, a regression for three months, etc. 

Step 2: Synthesize intangible data

Transactional data covers most data generated by everyday business processes, including financial data like sale values, logistical data like shipping statuses, and work-related data like employee productivity. According to Davis, this data drives “90 to 95% of insights.” Even so, she says there’s a tendency to bring in new data sources—some of which are disruptive or unhelpful.

“There’s a tendency, especially among stakeholders, to want to gravitate towards non-transactional data when they come up with hypotheses,” she says. “They might want to know whether someone’s sentiment affects their subscription tenure.”

To add an extra 5 or 10% to her insights, Davis leverages firmographic (information used to categorize organizations) and demographic data (information used to categorize individuals). These data sources are mature and reliable, especially compared to more innovative and untested options. 

This does not discount the role of more experimental data sources.

Contextual or environmental data (sentiment, emotion, perception) add nuance. Davis says she can see narrow functions for these data sets: “You can use sentiment analysis of reviews to add additional features to your data. This person was upset when they submitted this review and this person was happy. That’s helpful.”

Step 3: Combine datasets and operationalize analysis

Perhaps the most significant blocker to data-driven user journey design is access

Data inevitably lives on a mosaic of different systems: product, MAP, CRM, customer success, internal collaboration tools, and so on. Without a unified dataset, analysis is impossible. 

When running a data-driven customer journey transformation at SAS, Adele Sweetwood adopted a simple preliminary process.

  1. Data cleansing: “Correcting nonstandard customer data and removing duplicate records.”
  2. Data profiling: “That enables better understanding of the data by uncovering related data across tables, databases, and applications.”
  3. Entity resolution: “Identifying data from multiple sources and attaching them to a single customer.”

The three-part process allowed Sweetwood, then Senior Vice President of Global Marketing and Shared Services at SAS and now a principal analyst at Forrester, to turn incompatible data sets into “usable data stores.”

With usable data, customer journey researchers can turn their attention to analysis—but that requires somewhere to work.

“Even with the artificial intelligence (AI) and analytics capabilities needed to segment customer data and recommend engaging experiences, most B2B companies lack the specific capabilities needed to unify individual customer profiles with real-time insights on customers’ cross-channel engagements to support the hyper-personalized experiences customers expect today,” wrote Cecilie Burleson, manager of technology consulting, and Joel Wright, senior manager of technology consulting at Ernst & Young.

Organizations [of scale] need some sort of dedicated data warehousing or business intelligence platform, to consolidate disparate data sources. Burleson and Wright introduced Adobe Enterprise Platform to “unify customer data across the organization.” Others used Power BI from Microsoft, Snowflake Data Cloud, or similar products.

Step 4: Unearth winning data combinations

With transactional and supporting data collected, cleansed, and unified, organizations can begin interrogating the information, searching for correlations and connections. There are manifold approaches and strategies, many of which are complex and intricate. But organizations need not delve too deep to return results.

“Data science methods and techniques can be very advanced or very basic,” says Davis. “From my experience, most companies struggle to do the simple stuff well. I always recommend that people start with something simple and get into good data science habits.”

When asked for a real-life example of transactional data in action, Davis is true to her advice. She doesn’t offer a story of regression analysis, predictive modeling, or dimensionality reduction. Instead, she explains a simple strategy that delivered outsized impact.

One of Narrator’s clients, an eCommerce subscription company, had a strong acquisition pipeline but wanted to improve its retention. Specifically, they wanted to better understand how a customer’s first purchase affected their long-term retention.

They collected transactional data and compared retention rates between cohorts: those who bought product A first, those who bought product B first, and so on. Davis computed the difference in lifetime value from her product analysis and discovered some led to stronger retention and higher lifetime value. The insight allowed the company to tweak its customer journeys, prioritizing sticky products for first-time products.

“I like this example because the gains are compounding,” Davis says. “If you can increase retention from first to second order, there’s a compounding effect on lifetime value. We’ve seen something in the ballpark of a five to 10% improvement in lifetime value.”

Use every tool at your disposal

During her data-driven transformation at SAS, Sweetwood surfaced numerous opportunities and points of friction in her user journeys.

She discovered that much of her customer messaging was misdirected or out of sync. For example, all contacts received the same post-deal communications—whether the deal landed or not. Her research also revealed contacts were frequently requesting the wrong content. People involved in a deal for one solution (say analytics solutions) would request information in another (customer intelligence). It was clunky and confusing.

These learnings were possible because Sweetwood broadened the scope and surface area of her data collection and analysis. If she had merely relied on self-reported surveys, it’s unlikely she would have stumbled on the opportunities for improvement. She advises other customer experience leaders to follow suit, using every tool at their disposal.

“Ultimately the customer journey is defined by the data that our customers are willing to share or that we can track effectively,” Sweetwood said. “The intent is to understand preferences and behaviors in order to communicate and service our customers effectively. The journey isn’t stagnant and we are constantly learning.”

Leave a Reply

Your email address will not be published.

Ask an expert

Ask an expert Have a question related to this story? Or a thought you’d like an expert to weigh in on? Write to us.