Part 4 of the Operationalizing Customer Experience Series
Intellyx BrainBlog by Jason English, for Conviva
In this series, my colleague Jason Bloomberg and I have defined quality of experience’ (QoE), discussed operationalizing telemetry data for QoE, and explored the dynamics of Time-State technology. All of which should help us break out of traditional thinking when optimizing how our critical customer-facing applications perform.
Unfortunately, all of the time-state telemetry data and quality-of-experience optima in the world can still produce suboptimal results, unless we can understand our customer’s intentions and impressions during the moments they are taking action. Subjective indicators about the customer’s state of mind make it very hard to extrapolate useful customer experience metrics.
Given this difficulty, it would be tempting to continue measuring the success of our customer-facing application in terms of concrete results, such as reduced customer churn, increased signups and sales revenue, or the old gold standard of high NPS (net promoter score) from happy customers who answer a quick pop-up survey to say they’d recommend us to a friend.
While useful for measuring company performance, results-oriented measures can be misleading for teams delivering a digital experience. For instance, if economic conditions improve for our customers, or we are lucky enough to carry a hot product everyone wants, more revenue would still give us little insight into what we should specifically improve about our application. We’d still be leaving money on the table.
That’s why successful digital transformation doesn’t focus on the destination—it focuses on improving the customer journey. And the lived experience of that journey consists of multiple ‘customer flows’ which are the cellular unit of measurement for experience-centric operations.
Why are customer flows critical?
Flows are a familiar concept in software development. For instance, they have existed for years in web traffic monitoring, tracking the aggregated visitor click-throughs as they navigate a path through a website.
Observability dashboards may also display back-end flows, for instance, tracking sequences of API calls and responses, database reads and writes, REST transactions, or logs passing between services (i.e. the giant hairball map of system operations.)
Engineers are also looking at RUM (real user monitoring) or synthetic monitoring solutions–which can capture and/or help them replay the traffic patterns that influence application performance. Telemetry such as error rate, network traffic load, API response times, database latency, and so on can be isolated and captured.
Without customer flows, detailed measures are putting the cart before the horse. To collaborate effectively, we want development, operations, product management, sales, and support all monitoring the same flows as common actionable building blocks of customer experience.
What’s in a customer flow anyway?
As a basic unit of customer experience, you might think of each customer flow (or CF) as a subway or train station stop on the customer’s journey where they spend moments or minutes to view information and/or make a decision.
Like previous stations visited on the customer journey, each CF has one or more entry points of other downstream activities that lead into it, and one or more exit points of upstream CF activities they could visit next, until they reach a terminus of success or failure.
Note, you can come up with your own concept of a CF, but I wouldn’t recommend metaphors like ‘lego blocks,’ ‘pipelines’ or ‘links in a chain’ which would be useful for describing data movement and/or integration steps, rather than a unit of customer time and attention on their intended journey to a destination.
Figure 1. Within this Conviva interface, customer flow metrics are presented as “widgets” or cards on a dashboard, each of which breaks down goals for time spent, errors, usage and success rates. Aggregated application performance metrics are shown under the flows.
Are customers able to complete each flow, within an acceptable period of time, and without noticing any problems? Each flow needs to work perfectly to contribute to a great experience.
This requires that both product and engineering teams gain an end-to-end understanding of each customer journey, which can be broken down into critical customer “flows” such as “add user” “view shopping cart” and “check out.”
By elevating flows as experience-centric metrics, we can look at them alongside conventional measurements. Both DevOps and business-side teams will collaborate more effectively by grouping granular system and traffic telemetry data underneath customer flows as a common unit of experience.
Teams can then set a baseline for what is acceptable, and set objectives for improvement toward an ideal, as they improve flows across all interactions between a customer and the digital brand over time.
Solutions for the here and now
To make the example practical, let’s try breaking down a typical e-retail scenario and getting into a customer flow.
What if customers are just abandoning carts in the middle of a checkout flow all of the sudden, even if there are no errors reported? Since the root cause of each particular abandonment at the Cart Review flow may not actually be reproducible, how can we figure out what caused it?
Looking at the aggregate observability data, we can see that the cart review data shows up on the page in under 1.2 seconds on average, with only a few anomalies, apparently due to bad wifi at the customer endpoint, so it doesn’t seem to be a page load time issue.
This is where the value of full-census telemetry data from a solution like Conviva really comes in handy, as we drill into particular user sessions that abandoned their carts but did not have a network timeout.
Figure 2. A single customer flow for Cart Review with upstream and downstream customer flows. (Infographic by Jason English)
Reviewing the smaller set of cart abandonments, we see that some are being presented with shipping costs, but some of the delivery dates are different, or fail to load for certain products in the cart.
Perhaps showing the estimated delivery date and shipping cost up front for the items in the cart might help resolve some of these issues.
To do that, developers also need to improve the performance of responses from logistics providers through the shipping API. A product manager could report a trouble ticket to those partners, telling them to pick up their own response times. That may work in some cases, but it doesn’t take responsibility for the problem.
Instead, the engineering team can take some of that effort into their own hands to avoid dependencies, for instance by calculating and pre-loading the most likely or ideal shipping cost and time estimates next to the items in the cart. The estimated or cached delivery costs and times could then be delineated as ‘guaranteed’ at checkout time when the actual responses from vendors and shipping partners come in.
If you have optimized all the flows as much as you can, and customers are still abandoning their carts, perhaps it is not an engineering problem at all, it is a design problem. Time to reduce required steps and delete some extraneous customer flows!
The Intellyx Take
Breaking down aggregated observability data into customer flows is not rocket science. In practice, it’s more like regular math expressions—putting brackets around each customer action to determine the order of operations in a long equation.
You may have an array of telemetry data sources, observability tools, and service and incident management platforms at play in your application environment. What matters is what you do with that information to improve quality of experience. That represents your primary brand advantage.
Don’t settle for a high-level overview of observability data, when you can live in the here and now with your customers, and get in the customer flow.
©2024 Intellyx B.V. Intellyx is editorially responsible for this document. At the time of writing, Conviva is an Intellyx customer. None of the other organizations mentioned here are Intellyx customers. No AI bots were used to write this content. Image sources: Feature image: Adobe Image Express; Product screenshot: Conviva; Infographic: Jason English.
