Digitalizing Sales: Understated Themes that Make the Difference

The Harvard Business Review (HBR) needs no introduction - it is widely considered to be the premier management publication. So, I was delighted to see our work at Microsoft featured in the September-October 2022 HBR Magazine in an article titled “How to digitalize your sales organization”. It took a few days to internalize - I was proud of our team's work and humbled to have played a key role in it.?

I was not surprised that the authors, considered authorities on this topic, had leveraged our learnings to be the cornerstone of their paper. The HBR article did well in capturing the key themes of a) boundary spanning leader at the helm with executive sponsorship b) accountability measures c) cross functional teams d) agile approach and e) enablement. In this post, I cast light on the understated - the subtle decisions that made a lasting impact, the counterintuitive choices that delighted users as we digitalized a global sales org with $10B+ in revenues and more than two thousand sellers.

There are a few nuances that made the digitalization of sales at Microsoft special compared to more storied cases of Amazon (Retail not AWS) and Netflix. First, the Microsoft ecosystem was a B2B sale involving a sales team (at times up to 7 sellers per account) and multiple decision makers at the buyer. Second, enterprise software decisions are not easily reversible vs. Amazon or Netflix, where you can simply return the product or start watching a different show. Third, Amazon and Netflix both had the advantage of building iteratively from the ground up over the last decade and a half. At Microsoft we encountered the benefits and challenges of legacy sales processes and old habits. Both of which had generally worked well, after all it is one of most valued companies globally. How do you transform something that is already working?

Leveraging AI to Digitalize Sales

At the core of digitalization efforts at Microsoft was an AI-based sales intelligence application named Daily Recommender, purposefully built to recommend the next logical product for the customer and next logical action for the seller. The app synthesized over 1K+ data points per enterprise customer and used a series of AI models ranging from deep neural nets to random forest to make recommendations (new business, cross sell, churn and more). The recommendations were prioritized for the seller and curated with relevant content and contacts. From a digitalization standpoint, the goal was to use AI to enable daily decisions - “which account should I call this week”, “who should I talk to”, “what solutions should I talk to them about'', “how do I frame the discussion”, “who else should I engage to ensure customer needs are met”. This would a) eliminate countless hours of manual work b) generate new business pipeline and c) improve conversion. And these benefits we did see: within two years of rollout, sellers reported productivity gains of 40% plus, conversion rates improved 5x to 30%, and the sellers generated more than $1B in incremental revenue.

The Understated Themes that Made the Difference

Shift the focus. HBR underscored the need for a boundary spanning leader at the helm with executive sponsorship. For digitalization to succeed both are a must, however, executive sponsorship takes on a new meaning. First, executives need to make space for AI-enabled decision making. Decisions in a sales organization are made in one of two ways: either a decision is made bottom up by the seller influenced by the customer or top down via sales priorities and operations. The room for a third dimension of decisioning does not necessarily exist in sales processes. At Microsoft, we made a hallmark decision to use Daily Recommender as the primary medium for discovery and lead generation. Second, executives should seek economies of scale and not just scope. “To bring more revenue, I need more sellers and better solutions to sell” - a common point of view, however one that centers on scope. Focusing on scale via digitalization entails a different set of questions?- “how do we use AI to help sellers sell more, sell faster, sell bigger” or in sales semantics, increase pipeline, velocity, and revenue per deal. At Microsoft, we struck on these - we wrapped the effort under the broader company strategy to scale its salesforce and we leveraged boundary spanning leaders to make space.

Measure the invisible. The article emphasized the need to incorporate accountability measures via objectives and key results (aka OKRs) coupled with a governance framework (such as a business review). At Microsoft, we implemented this from day one, examining metrics on model accuracy, adoption, and impact. And yet in the early days, we saw what most organizations see - adoption lagged even though the models passed the quality controls. Some were quick to dismiss the role of AI in B2B sales. So, what was missing? In the digitalized sales ecosystem, there are two actors, the sellers who master the unspoken, the customer sentiment in between the lines, and the AI that reasons and masters the data. Frequently the two arrived at conflicting positions. A measurement of the linkage between the two was missing. In a machine-to-human AI ecosystem it is necessary to measure the actual AI quality (counterfactuals) and the perceived AI quality - or “how the human reacts to the AI”, “do they agree passively or actively with the recommendations'', “do they dismiss them off the bat”.?We course-corrected, integrating metrics on perceived AI quality into the business reviews, holding the AI teams accountable, and using the metrics as inputs into the roadmaps. A step change in user sentiment occurred in three months, sellers agreed with the recommendations 75% of the time, disagreed 7% of the time and were neutral 17%.?

Embrace a product mindset. The article touched on the need to deploy cross-functional teams. At Microsoft we did take this approach, however it was the product mindset of the cross-functional team that made the difference. The team had an unapologetic and uncompromising focus on working backwards from the user and evaluating decisions holistically based on outcomes (not outputs). In the formative stages, we decided to ensure every transaction within Daily Recommender could be completed in no more than 3 clicks - after all you can buy a product on Amazon with one click, you can book an Airbnb in three. So why not here? As the system evolved and features were added, we adhered to this concept religiously to ensure the end result was simple yet powerful. And in the end, it was: the entire sales prospecting experience supporting more than 29+ products came to life in two highly curated pages. Every seller transaction ranging from looking up contact details, to viewing an accounts marketing engagement, to logging an opportunity could be completed in less than three clicks. Decisions were evaluated holistically: once the AI models supported all products, a decision was made to use the same for account planning, and consequently digitalize that process as well. The sales cycle consists of conjoined tasks and processes i.e. planning is an input into prospecting, prospecting leads to sales, sales to consumption, which in turn leads to renewal or churn. The cross-functional team needs to embrace a product mindset to digitalize sales.

Agile or hyper agile. HBR touched lightly on the need for agility, but it is one that needs more emphasis. From its formative years, the team operated with hyper agility - in practice our mental model was 10-10-10, ten hours to define a problem, ten days to design a solution and ten weeks to pilot. This does not mean that the problems were tactical or limited in scope. In fact, it meant that large problems needed to be broken down into bite-sized measurable chunks that could be addressed quickly. It also provided the team the luxury to run many pilots with a low cost of failure vs. a handful of large ones with critical impact. It was not uncommon for 100 plus features to be deployed within a quarter. Getting here wasn't easy; through the journey we encountered headwinds from IT partners where six-month semester plans, gantt charts, and linear processes ruled the day.?Digitalization of sales requires an agile or hyper agile approach. The reason is grounded in human psyche and the nature of sales - if you are going to ask a seller to rely on AI (or another data point) to change their behavior with a customer and put their credibility on the line, you are likely to get a handful of chances and a very short timeframe to prove your worth.

Don’t just enable, design for context. HBR emphasized the need to enable the organization and we did this at Microsoft. However, we designed the solution for context and that is what paved the path for grassroots adoption. Initially, when adoption was low, we dug deep and realized it was not necessarily because the recommendations on the next best action for the seller were incorrect, but more so the underlying context was missing. Instead of just mandating usage top down or furthering change management, we made the choice to invest in AI explainability, a difficult and costly problem. Sellers would now see the reason for the recommendation and the reason to call the customer. These features not only provided context, but they created an aura of trust and transparency since the users could physically see how the AI was processed. Sellers, particularly those on variable compensation plans, are known to act autonomously in quest for quota, consequently it is imperative to design for context first and then enable the organization.

As the HBR article stated, digitalization of sales is proceeding at a breakneck pace, however digitalization efforts also have high failure rates. At Microsoft, we achieved transformational outcomes: seller productivity increased by more than 40%, conversion rates by 5x. By following the broader themes and the subtleties summarized here, you can dramatically increase the chances of success of digitalizing sales.