Transactional truth vs. personal truth

What are SSOTs?

Everyone in digital commerce discusses creating a “single source of truth (SSOT)” for customer transactional data. Let’s first define it and then discuss the glaring hole – the elephant in the room.? In e-commerce, there are many choices for sources of truth. Platform-reported numbers are pulled straight from paid media platforms. Analytics are sourced through Adobe Analytics, Google Analytics, and built-in-house platforms. Transactional data is pulled from the e-commerce site, a CDP, third-party niche solutions, and internally developed data stores.? Each source produces different data. It is often difficult to gain alignment on which sources to use for raw, unprocessed, customer data. The task becomes even more difficult when layering on metrics, reporting, or insights. Layer on again that different teams within the same company require different SSOTs (think Finance's Essbase system vs. Marketing's data lake).

SSOTs for transactional data

Let’s start with transactional data, and then advance to the elephant in the room. I wrote a piece for CIO Review (https://www.cioreview.com) detailing the five levels of analytics "truth" for transactional data:

• DATA is the simple raw elements used to describe an event, like which customer bought what items at what price on what date, or what the weather was on a given day in each place.? Generative AI can find public sources of data. It can also be the source of fabricated data which can be assumed to be factual if not properly identified and controlled.
• METRICS are the result of aggregating and using arithmetic operations to arrive at a new "truth,” such as average order value, ROAS, and click-thru rate.? Generative AI can identify useful metrics and even propose new metrics.? It can also provide inaccurate information about them, such as providing an incorrect formula on how a given metric should be calculated.
• VISUALIZATIONS are the presentation of collections of metrics and data, such as a report of performance by campaign or a chart of how ROAS has been trending.? Generative AI can retrieve visualizations and create new ones.? It can suggest visualizations to represent a given set of data.? In the not-too-distant future it will be able to “see” an existing visualization and create a narrative for it.? But it can do all these things incorrectly as well.
• ANALYSIS is manipulation of metrics and data to answer questions, such as what the acquisition drivers are and why the brand is not adding new subscribers.? Generative AI can recommend analyses to be done, perform rudimentary analysis, and even provide possible explanations as to why the numbers are trending as they are.? As of this writing, it underperforms at these types of tasks, but there are efforts underway to improve its capabilities.
• NARRATIVE is the story that combines all the above, identifying the focus and what should be done. The generation of narrative content is a strength of Generative AI.? If prompted with the relevant information about how numbers are trending, it can produce a narrative to explain it, although the quality of the narratives vary.? In the human world, narrative is subject to perspective.? For example, if sales are down, we can expect potentially different perspectives from operations, marketing, finance, product, and other sources.? Generative AI at present provides just another perspective.

Multiple sources of truth can emerge at any of these levels.? Multiple analysts looking at the same data can arrive at different conclusions.? Senior leaders with different experiences and skill sets can look at the same analysis and form contradictory narratives. Any brand will have multiple versions of the truth; there is no easy way (or often even a problematic way) around this. Brands will do well to identify and track their various SOTs, understand their origins and why the SOTs exist, and have agreed ways to translate from one SOT to another. They can also address the elephant in the room.

SOTs for personal truth

Personal truths are a different matter.? Personal truths are person-specific, fluid, temporal, defined in the moment, and subject to interpretation.? This is why transactional SOTs can tell us who purchased what, when, where, and how it happened, but they cannot tell us “why it happened.” Take for example a brand ambassador asking a customer if they want to try a product.? If the answer is no, the reason does not align to a single “truth”. There could be many reasons.? The product might have no relevance to the customer right now, at that moment. The product might not be appealing at any moment.? The customer might not have the time, or might not be in the mood to experiment.? The customer might believe that they won’t like it, and therefore they don’t want to try it.? The customer might think the product is, or will be, too expensive.? A conversation on a certain day can reveal a certain truth, but a followup a month later might reveal a deeper, revised, or conflicting truth.

These personal truths are actually layers of complexity, and all subject to change. If a customer doesn’t want to try a product today, then the objective truth is that, when offered by this particular brand ambassador in this particular context, the customer didn’t want to try it.? We don’t yet know whether the customer will want to try it later.? Capturing the “why” of the customer’s reluctance is key to helping us understand whether to re-offer the product in the future, and which products, if any, to offer instead. To know why a customer is behaving in a certain way, it becomes critical to actually know the customer. The way to know the customer is to create an opportunity to meet, engage, and create a recurring relationship with the customer. As the brand increases trust, empathy, care, and the art of anticipation, the customer will share more intelligence. We’ve created a customer intelligence platform that triangulates 1:1 customer engagement, AI-driven intelligence, and digital commerce integration that will unlock the “why customers do what they do.”??

We unleash insights by learning personal truths. Even knowing that a customer previously refused a product trial allows a brand ambassador to revisit the subject in the future.? A question like, “You weren’t interested in our product XYZ, right?”, allows the brand ambassador to revisit in the future and invites an explanation from the customer. No need to keep bringing up product XYZ with the customer, but it can be premature to entirely abandon discussion of XYZ with that customer until the offer has been discussed more than once. This will change the way brands use SSOT to guess next-best-action, build propensity models, and determine customer lifetime value. With insights from millions of personal truths, we will fundamentally change the way brands create single sources of truth for their data.

At Umego, we use AI to help people build relationships.? But it’s not just about the app and the data capture, it’s about the training for brand ambassadors to understand what the data means in the context of personal truth.? Identifying and learning personal truth is not as simple as doing so with defined layers of transactional truth, but it is richer, deeper, and more useful for building relationships.

-

@Larry Seligman, Chief Data Officer at @Umego.

Follow me for insights on AI, analytics, and customer centricity.?