An AI Driven Customer Dashboard is required to create an accurate Customer Health Score system

It will become the standard for software companies to utilize AI driven customer experience solutions in order to automate customer interactions with the most accurate and relevant information in a timely manner.

This means implementing Analytical CX with AI solutions early in the software start up phase to create automated processes for customers to operationalized the solutions faster. These CX solutions will need to analyze user usage patterns and behaviors by user types and provide recommendations to drive them to utilize the software in a way that will obtain the value quickly and on-going. Ideally, these will be solutions that can be integrated with your own software solution to provide user insights and recommendations within the software as much as possible.

How are we going to use this data and insights to provide the CS team with the right information and areas to focus on to ensure customer ROI and course correct when the customer falls off the plan to get to a planned ROI?

The Executive Leadership Customer Dashboard

I like to start at the top level and work my way down when evaluating dashboard metrics. The Leadership Customer Dashboard usually has the basic metrics of NRR, Churn, NPS, CES, CSATs and growth rates, which are KPIs tracked Quarterly at the leadership level. These are the indicators the leadership team will review to drive results to company targets for retention and growth.

When any of these metrics are not hitting the targets, CX leaders would then utilize their detailed dashboard with operational metrics and KPIs to help them understand why these top line customer metrics are not achieving the quarterly targets or the internal customer experience standards defined by the CS team. This dashboard might have metrics such as the following:

• Customer Onboarding SLAs (Sales Handoff, Customer login, training and initial launch)

• Customer Implementation Cycle times by stage (i.e. Plan, Discover, Configure, Test, Launch)

• Time to Launch

• Time to Value

• # of Open Tickets and Support SLA attainments. (i.e. based on support policy)

• Aging Bug backlog

• Customer communication (open rates, education webinar attendance)

• Training attendance, completion rates and test scores

• Customer Online Support center usage analytics (google analytics)

• etc.

The Customer Product Usage Dashboard

In many cases, the product usage analytics dashboard will be a separate dashboard owned and updated by Product Management. The basic metrics on this dashboard might be:

• Login rates

• usage rates by feature/functions

• Process cycle times

• Time in application per login session

• element creation counts (i.e. # vendors, #reports, #content, etc)

• Frequency usage by function

• etc.

There are many software solutions to help you collect and trend these various data elements over time.

Developing a Predictive Customer Dashboard

Once you have all of these basic metrics and KPIs in place where you are collecting and tracking customer activity with your software and services; you have the baseline to begin developing a proper, effective customer score that will have a high rate of predictability for customer retention and growth.

There are several popular customer analytics software solutions which allow you to build the AI capability into the predictive model. If the solution has machine learning abilities, I would recommend to utilize this as part of your predictive model to keep learning and evaluating which of these elements continue to have the highest correlation to the value lines.

In order to understand which of the metrics above should be used in your model, you will need to do some statistical analysis of the data. This will be where you test the various hypothesis around what data elements you believe influence the customer’s ability to achieve value and drive additional growth demands. It is critical to understand which of these metrics are highly correlated to customer retention and growth. The data cohort is defined by customers who have renewed at least one year or more, and tracked by each year of tenure and growth rates by year.

At Workfront, I worked with the product team and we hired a data and statistical analyst to help us do some modeling and help us understand which of the elements were the highest predictors of retention and growth. During one of our hackathons, the CX and Product teams worked together and built an application around this model. The results were not only 98.5% accurate in predicting retention and growth, but the levels of achievement required in each of these metrics was not what we expected. This data revelation helped the CX team to innovate our entire customer journey to drive the RIGHT key metric results which improved our retention and growth rates significantly.

One of my favorite Customer reports was one where we could select a specific territory and target market and it provided a bubble chart showing all the customers in that territory and target market. The size of the bubble indicated the Revenue $ and the color of the bubble indicated the customer health score status. When you hover over the bubble the summary customer details came up including Contract Value, # of users, tenure, renewal date. If you selected the bubble a detailed 360 customer report came up showing detailed stats for; implementation, consulting, training, support and product usage. This dashboard had over 40 detailed metrics tracking customer product usage and progress to their strategic roadmap plan from Sales handoff and initial launch (Time to Launch) through to completion of the customer roadmap plan (Time to Value) in which they should have achieved wall to wall operationalization of the solution (TLV/TCV).

Whether you build or buy a solution, you will want to ensure your solution has an AI capability whereby the machine learning abilities can work with the statistical modeling of the data to update your models regularly. This will help you to understand if new product released functionality influences key drivers for improving value and predicting future retention and growth.

The AI models will allow you to test the correlations of each of the various elements to see which elements have the highest correlations to the customer value elements and which are the highest predictors to retention. For example, at Workfront, we had a hypothesis defined for the use of specific customer reports that we believed had to have over 90% usage to ensure a renewal. The model actually showed that the usage rate of this report only needed to be over 67% with the combination of two other elements being over 50% in order to ensure 100% renewal. So, where we thought we needed to have over 80% usage rates in some functional elements, the model showed it actually didn’t need to be that high, but it did need to be combined with specific usage rates of other key functional elements, which is what produced the highest correlations to value.

Building the Customer Health Score

As you evaluate these results, it will help you to understand what data elements should be a part of your health score formula and how best to weight the various elements. These elements are what you want to analyze to determine the right elements to be included as part of your health score calculation in order for the risk analysis to be more accurate in predicting customers at risk and why.

I would suggest working with the product engineering team and some mathematicians on the team to create the right health score formula to ensure it includes the elements with the highest correlations to predict customer retention and expansion. Most health scores do not have enough or the right variables in the health score formula. The elements in the formula should have been tested as being highly correlated to the value the software is capable of generating for the customer.

The next phase of the AI model would be to analyze the scores for each of the data cohorts (by retention tenure, market segment) to determine the scoring ranges associated with various risk levels that make sense to establish to define the different levels of risk. The Customer Analytical model would reveal the score levels associated with percent ranges of probabilities of retention and expansion. Using the results of the model, you would be able create score ranges associated with risk levels. For example, if the health score ranges from 1 - 100. It may help define a scoring system such as:

• Score = 1-40; AT RISK (Probability of renewal is below 50%)

• Score = 41-70; Some Risk (Probability of renewal is between 50 - 75%)

• Score = 71-85%; Limited Risk (Probability of renewal is between 75 - 95%

• Score = >85%: No Risk (Probability is 100%)

With the dashboards showing each of the metrics results weekly, monthly and/or quarterly for each customer, the CSM will be able to check these details when the score falls below the standard level of predicted retention and see which metrics (associated with an element) have fallen below the standard. These are the specific areas that the CSM and services team can evaluate to learn what caused these metrics to fall below the standardized level and develop a plan to course correct and get the customer back to a high value state.

The beauty is that the CSM will be able to understand specifically what elements to focus on with the customer. They can actively work with the services team(s) to put a plan in place for their customers that are not in the 95% or higher probability of renewing or expanding.

Even with more predictive Customer Health Scores, the CSMs and/or AEs would still need to confirm the value has been or is being realized by the Customer. The customer dashboard can be used in the Customer Executive review meetings to share their results and allow the CSMs and/or Consultants to provide recommendations for where the customer could be realizing a higher business impact.

Once the customer acknowledges the improvements the software has provided using these dashboard stats, they are more motivated to work internally with their leadership and/or finance teams to confirm the dollar cost savings or revenue increases associated with these metric improvements. This information can be used by the customer champions to justify the business case for continuing their investment in your solution each year.

My prediction for the future of the CX organization and the value they will bring to any software company will be their ability to create these predictive dashboards and forecast retention and expansion revenue within 5% accuracy.