A Data-driven Approach to Improving Sales Efficiency
Sales team plays an important role in growing and expanding customers of a company. This is especially true for a Business-to-Business (B2B) solutions provider like Moka, than a Business-to-Customer (B2C) focused company.
Improving the efficiency and processes around the sales team provides a sustainable competitive advantage for any B2B company. Providing the right tools and insights to our sales managers is critical in achieving our mission to Empower Merchants to Grow and Sell across Indonesia.
There are 4 critical stages a user goes through in any product/service-based company:
Awareness
The audiences are made aware of the product through marketing campaigns or events.
Registration / Installation
Interested users install the app and register for a trial account.
Lead Nurturing
Registered users are nurtured into leads by providing more context and product knowledge through demos.
Conversion
Leads that are convinced about the product value will then pay for the subscription and become a Customer.
Handling and converting qualified Registered User (RU) into a customer is an important business process that differentiates companies that blossom or diminish.
As a market leader of POS systems in Indonesia, Moka receives thousands of registrations on a monthly basis. While the major portion of them are genuinely interested in the product, we also see a notable portion of spam and unintended registrations. In this article we explain, an unbiased and scalable Machine Learning approach to recognize and nurture the leads that show promise and cut down the noise.
Being a data-driven company since its inception, Moka collects and processes several important business and operational information throughout the customer lifecycle. This data is used to improve the product experience and add substantial value to the customer journey.
Rather than using heuristics and gut feeling, we let data guide us to understand our users’ behavioral patterns.
We identify the important traits of a successful lead based on previous conversions from leads to customers. This knowledge is then used to differentiate leads that have a higher chance of realizing the value that Moka has to offer their business.
We incorporate several important factors (features) into our model which include the customer demographics such as business type, a location that is provided during registration. Additionally, we also integrate their interactions with our back-office and mobile app during their initial days of the trial period to measure their intent and interest with Moka.
The user activity is then categorized into different groups such as ‘Catalogue related activity’, ‘Reporting related activity’ based on the pages they view and the actions they perform. We discovered that our successful conversions shared some common behaviors such as:
- Actively engage with Moka platform during the initial trial period.
- Explore different features in the back office such as Reports, Account settings, Customer profile setup, etc.
- Explore various features in Mobile Application such as checkout process, item catalogue, etc.
We did a detailed study on the behaviors of our users based on their journey into the trial period and the nature of actions they perform. A snapshot of the proportion of their event categories is depicted below.
Catalogue and Reporting related events take up the major portion of activity during the trial period, which is intuitive as Moka’s core value proposition is to simplify sales and have a transparent reporting for the business owner. We also notice a sharp decline which is setting related events in the later stages of the trial period as businesses typically continue with initial settings and there is less need to modify settings.
Based on these observations we train a supervised Machine Learning model (XGBoost) to infer their probability to pay for the Subscription. The model then predicts the quality of a lead, whether they are ‘high quality’, ‘medium quality’ or ‘low quality’.
High-quality leads show the interest and intent of the user which is quite important for a sales executive to convince the user for a demo and ultimately a subscription payment.
We use this model on a daily basis to provide a ranked list for our Sales team. The lead ranking process has performed extremely well in the real world, where the high-quality leads identified by the model have 250%–300% higher conversion rate than low quality leads.
With this data-based approach, our sales team can work efficiently by focusing more on the high-quality leads and prioritize their daily schedule to pursue leads and do in-person product demonstrations.
Data is valuable and powerful, when it is processed within the right context and interpreted with proper business acumen. It has the potential to change critical business processes and add value to any Company/Organization.
We hope this article provides some insight and inspiration for your journey into data-driven decision making. Share with us if you have similar examples where you use Data to improve the business processes!
