Win more with eXplainable AI
I’ve spent 20 years analysing and making sense of sales data in CRM systems and in that time AI has helped us to extract more meaningful insights, enabling us to improve the processes that drive more sales. When eXplainable AI (XAI) is added to the mix, you’ll find that suddenly you’re able to answer questions like “Why will I win this deal?” rather than “Am I going to win this deal?”. This provides the basis for modifying sales tactics and strategies that will ultimately allow you to win more. In this article I’ll give you a real example of how an XAI system that I have built has benefitted a business.
A real world example
As CTO of Cloudapps.com I was talking to one of our customers: a large business selling high value deals in a B2B setting.
The solution they purchased assigns a "health score" to each sales opportunity and uses XAI to explain how that score is calculated. The person I was talking to was part of the Sales Operations team and he started to tell me about a recent experience which went something like this:
So we last quarter we had this really significant deal in our forecast and the sales rep had it in their commit with a probability of 95% likely to close. We were as confident as we could be that this deal was going to be won. The problem was that the AI “health score” was really low - like 5%. The funny thing is that we lost that deal meaning that the AI was right and the Sales Rep was wrong.
Now at this point with a conventional “black box” AI system, you’re left completely in the dark as to why the AI has decided to score this particular deal so low and so you’re left none the wiser as to what you can do to avoid this from happening again. Explainable AI adds the crucial piece of missing information to help you understand why the deal was lost. He continued with his story:
When we looked at the opportunity we could see that the deal health score was low because the explainable AI said the decision maker on the deal was not good. When we examined this particular contact more closely, we discovered that they had never bought any of our products despite having previously had lots of deals with us.
And so here is the crucial point. Without having the XAI to tell them where to start looking they would have had no idea why they lost that deal. They could have been searching around for days looking at the data and not discovered this particular issue. He added the following:
Now, when qualifying opportunities, we always check the decision maker's background. If the explainable AI flags a potential issue, we investigate further and adjust our approach based on what we find.
The fact that a possible cause of the deal loss had been identified, allowed them to take action that could produce a different outcome next time. Since XAI provides instant insights, sales teams can quickly experiment with different approaches—even qualifying out early—to prioritise the right deals and maximise their time.
How does it work in practice?
Machine learning systems, called models, are taught to find patterns in data. Let’s take a very simple example: A deal that we are working on has the following attributes: Lead Source, Selling Location, Industry and Product. A model has been trained to use this information to predict the likelihood that the deal will be won. The prediction is a number that we call the deal health score and is between zero and one hundred. Zero means that it has no chance of winning and one hundred means that it will definitely be won. In our example the score might be 80 - meaning that it has an 80% chance of winning.
XAI adds additional information to the deal health score. We also get a value for each of the attributes that the model was trained on. Values greater than zero indicate that the attribute is positively contributing to the health of the deal and values less than zero indicate that they are negatively affecting the health of the deal. So, for example a Lead Source of Referral with a value of +30, has a positive contribution to the deal’s overall health.
Black Box Predictions vs Explainable Predictions
So now, with XAI, we can see exactly how each input into the model contributes to the overall score. In our example we might conclude that the deal looks good, but we might want to check that the Widget product is really a good fit for this prospect. Of course this is just one example, but you could use the same approach to:
Better understand the effect that seasonality may have on the ability to sell a particular product.
As a guide to better optimise the price or discount that is being applied.
To gauge the effectiveness of a marketing channel to convert leads.
Unblocking the DADA loop
Data -> Analysis -> Decision -> Action
The DADA Loop stands for Data, Analysis, Decision, Action and is a decision-making framework used in complex and uncertain environments. It’s similar to the OODA Loop (Observe, Orient, Decide, Act) but places more emphasis on data-driven decision-making.
Here’s how it works:
Data – Collect relevant data from multiple sources.
Analysis – Process and interpret the data to extract insights.
Decision – Make informed choices based on the analysis.
Action – Implement the decision and monitor the outcome.
This loop is particularly useful in dynamic environments like business strategy, and sales applications, where rapid yet informed decision-making is crucial.
Sales DADA Loop - can form the basis for continuous improvement
Without the piece of data that tells us why the deal was in poor health. The analysis that follows, would have been significantly more difficult, slower or maybe not possible at all. In other words the DADA cycle would be blocked and broken.
So here the key is that XAI has provided the missing data (ie why the deal was lost) that allows us to analyse the root cause and to unblock the process of improvement. Ultimately this allows companies to rapidly trial new tactics that will result in improved win rates and over time lead to a competitive advantage.
Conclusion
Explainable AI was all the rage a few years ago, but in recent years has become overshadowed as LLMs and Chatbots have grabbed much of the attention and in my view that’s a shame because as we’ve seen here it’s a powerful tool that can have far reaching benefits that go beyond just productivity gains.
Contact us today to see how XAI can help your sales team win more deals, more predictably.
