The need for Recommendation Engines in the Auto Aftermarket

Introduction

Recommendation Engine in general has received considerable attention in the market especially in the e-commerce and the OTT Industry. It is often said that Recommendation engine serves as a guide to predicting what your customers want and when. If we look at the statistics, we see that almost 60% of media choices on Netflix and almost 35% of sales at Amazon originate from the Recommendations which their platforms give to the customers. Also, a Wharton study of music recommendations indicates that receiving suggestions tailored to individual users widens exposure to new products and fosters human bonds. The increase in the volume of purchases can go up as much as 50%. Auto Aftermarket Industry is generally very complex which involves multiple SKUs which the companies must sell in the market as well as online through e-commerce. Here the Recommendation Systems become very important as the Sales Representative misses out on the key SKUs which can be upsold or cross-sold to the channel partner to increase the revenue of the company. Also, in the e-commerce space the Auto Component companies have the opportunity to increase the breadth of SKU sales through Recommendation Engine to the Customers. The Recommendation Engine utilizes Artificial Intelligence algorithm, usually associated with Machine Learning which can help users discover products and services which they might not have found on their own. Recommendation systems are trained to understand the preferences, previous decisions, and characteristics of the users using the data gathered about their interactions.

Types of Recommendation Engines

Collaborative Filtering: This is based on analysis of data on the user’s behaviour. This includes the users activity and prediction of what the user would most likely buy. There are two kinds of collaborative filtering techniques. The first is user-user collaborative filtering where the similar users are compared based on the items users have already liked or positively interacted with and the items are recommended on that basis. The item-item collaborative filtering takes into account the similarity between items calculated using the user rating given to the item.

Content-Based Filtering: Content-Based filtering methods based on the description of a product and profile of the user’s preferred choices. Products can be described using keywords, and a user profile is also built. As an example, in the Auto Aftermarket parlance if the user is buying Oil, Coolant and Air Filter the system can recognize that the user is buying maintenance parts and might recommend transmission fluid next time.

Hybrid Recommendation system Here the products are recommended using both content-based and collaborative filtering simultaneously to suggest a broader range of products. This kind of system is used by many e-commerce companies and OTT Platform these days

Types of Data used by Recommendation Engines

Recommendation Engine uses multiple data points for assessing the interaction of users. Some of these are given below:

• Historical Data: The Auto Component companies generally store their sales data in their ERP systems, and this is used as Historical data by the Recommendation Engine to understand the buying behaviour of the Channel Partners of the Auto Component Manufacturer. From Historical data the Recommendation Engine can identify the leading and lagging SKUs and the variability in the sales and quantity of these SKUs and identify the pattern which Channel Partner exhibits with respect to recency and frequency of buying the SKUs. The suggestion of SKUs can be made by the Spare Part sales representatives to the Channel Partner basis the Historical Purchases. Historical Data is also taken into account in the e-commerce purchases in addition to the ratings given by the purchaser of the parts on the website of the Auto Parts companies through which suggestions of SKUs can be made as a part of upsell or cross-sell.

• Demographic Information The Auto Aftermarket is a big Industry with multiple SKUs going to multiple Channel Partners. Hence Demography becomes important here with respect to Channel Partner buying behaviour. Demography is important with respect to location and type of Channel Partner buying the SKUs. For example, a certain spare part might be bought frequently by a cluster of Channel Partners in a particular Geo/Territory. The suggestions can be made by the Sales Representatives to the Channel Partner. In e-commerce also the behaviour can be taken into account by including the data of category and similarity of purchases through which suggestions can be made for upsell and cross-sell.

• External Data: The Data regarding the external factors like Vehicle Parc and seasonality, harvesting is very important when it comes to the buying behaviour of the channel partners. The Recommendation Engine takes all these external data points into account and can generate tailored Recommendation for the Sales Representatives to recommend to the Channel Partner which can lead to either Upselling or Cross-selling which can enhance the Revenue of the company. This is also important in e-commerce where the timing of the spares purchase becomes important when a certain spare part might be bought more because of the Vehicle Parc or Seasonality etc.

The format of recommendation can differ in different scenarios. For example, if the Sales Representative goes to the Channel Partner to recommend the SKUs this can be done through the instant messaging platform like WhatsApp in which the suggestions can be incorporated. The other format would be of the online purchase through companies which sell the Auto Components/Genuine Parts through their websites where the buyers can get recommendation to add certain items in their cart basis their purchase pattern or the purchase pattern of other buyers having similar behaviour.

The need for Recommendation Engines in the Auto Aftermarket

Given the scenario of the Auto Aftermarket, there arises a need for the recommendation systems due to the following problems faced by the Auto Aftermarket:

• High Number of SKUs: It becomes difficult for the part sales representative in the field to sell the entire range of parts to the channel partner. This can be due to many factors like difficulty in recalling all the parts, channel partner might not have much time and repeats the parts which he has ordered in the past to save time. Thus some SKUs which could bring revenue are left out in this process.

• Lack of analytical skills: The sales representative and the Distributor might not have much analytical thinking when it comes to which part to sell to which Channel Partner. This is because the data which is being captured is huge and there is an element of external factors which leads to variation in the orders by the Channel Partners be it due to geography, seasonality, festivals etc.

• High Number of Channel Partners: Today in the Auto Aftermarket, the industry has a lot of players selling to a lot of Channel Partners. In this scenario the aspect of segmentation becomes important and customized recommendation for each Channel Partner is the need of the hour which the Distributors or the sales representatives are not able to do, and this ends up with the same SKUs being ordered with a little variation and no tangible increase in revenues.

Challenges to Recommendation Engines in the Auto Aftermarket

Given the context of the Auto Aftermarket, many companies could find it hard to adopt the recommendation engine given the scenario of this industry. The companies will face challenges at many levels while implementing and executing the Recommendation Engine. These are highlighted as follows:

• Physical Interaction: The first challenge which arises for the Auto component companies is that the Parts Sales representative has a physical interaction with the Channel Partner. Since the Channel Partners have to sell the parts as well as cater to different sales reps, they may not have enough time which leads to the Channel Partner just repeating the parts which he bought earlier. Also the sales representative might not know which parts to upsell and which parts to cross-sell to the Channel Partner which leads to him taking a repeat order.

• Traditional ways of doing Business: The Auto Aftermarket is still running on traditional ways of doing the business and the Channel Partner operate in a traditional manner where they order only those parts which sell and shunt out the other potential selling parts.

• Implementing Recommendation Engine-based systems: Some organizations may lack the requisite IT skills to implement an Recommendation Engine-based system. This increases the cost as well as effort for these companies as the vendors charge the cost of implementing Recommendation Engine-based system separately from the Model building. Our interaction with industry experts leads us to the conclusion that the IT skills generally tend to drastically dip as we move from Tier 1 companies to Tier 2 and Tier 3’s.

Addressing the challenges of Recommendation Engine through Digilytics RevUP

RevUP by Digilytics provides easy-to-use Recommendation System for Revenue Growth Management in the Auto Aftermarket Industry.

Digilytics RevUP provides the following:

• Delivery of Recommendations:To ease the challenge of physical interaction, the Recommendations are delivered to the sales representative through widely used instant messaging platforms like WhatsApp which makes it easy for the sales rep to look at the recommendations and have a dialogue with the channel partner regarding the same.

• Model Management: Digilytics RevUP provides AI/ML based Recommendation Engine model and its management through setting up of robust data pipeline with a team of data science experts with faster implementation time.

• Easy consumption of model: Digilytics RevUP provides easy-to-consume AI/ML based Recommendation Engine model for various stakeholders of the client through visualizations, robust reporting through client systems as well as instant messaging platforms like WhatsApp. This would make it easy for the sales rep to visualize and see the recommendations which were considered and the recommendations which were not considered by the channel partners.

• Auto Aftermarket specific Focus:Digilytics product RevUP has built-in knowledge of Auto Industry as a lot of inputs from Auto Industry Experts has gone into it. RevUP has also been implemented in Auto Aftermarket space for clients. Combined with data science and our experience implementing solutions for the use cases specific to the Auto Aftermarket client for Revenue Growth Management, this takes RevUP in a sweet spot regarding Recommendation Engine for the Auto clients. Also, the insights provided by RevUP in the AI/ML models are specific to Revenue Growth in the Auto industry which makes it easy for the Auto Industry Stakeholder to understand.

• Right Data Strategy : RevUP by Digilytics has a well-architected solution where data follows through a well-defined pattern of creation, ingestion, storage in the right format and consumption layer. This improves the visualization results and the sales rep is able to see the result in a standard output template which consists of SKU-wise recommendation for the retailers.

About Digilytics AI

At Digilytics AI, we aim to drive business value leveraging our platform. In an ever-crowded world of clever technology solutions looking for a problem to solve, our solutions start with a keen understanding of what creates and what destroys value in your business. Founded in 2014, by Arindom Basu, the leadership of Digilytics is deeply rooted in leveraging disruptive technology to drive profitable business growth. With over 50 years of combined experience in technology-enabled change, the Digilytics leadership is focused on building a -values-first firm that will stand the test of time.