How Netmeds is leveraging AI & ML for operational efficiency, customer experience
In addition to customer-facing cases, Netmeds uses ML and analytics in its operational processes. The company has also been betting big on RPA, and has already experimented with blockchain in its supply chain process
Chennai-based Netmeds, one of the largest online pharmacies, has been building all core systems in-house, rather than using solutions available in the IT marketplace. The reason, as Advait Suhas Pandit, Director & CTO of Netmeds remarks, e-pharma is a new vertical and the speed at which the industry needs to adapt, and change is not possible if the company is dependent on a third-party to make changes to their products. Netmeds uses extensive parallelisation, CDN and edge computing on the customer front to ensure quick responses.
Speaking on the philosophy of its technology stack – appropriate tech for given problem – Pandit, explains, “While transaction processing systems are in .NET and Java, content-based systems are in PHP and NodeJS. Also, since solutions are composed as microservices, their management, monitoring, scaling and feature-rollouts are much simpler. We are in the process of getting started with serverless technologies such as Lambda and Fargate (serverless containers). Regarding the big data engine, a data-lake tracks all data using a combination of Redshift, DynamoDB, S3 and Athena. Excel, Tableau and some proprietary systems are used for data visualisation.”
Play of RPA
The company has been using Robotic Process Automation (RPA) in various operations systems. For instance, in order to keep the product catalogue up-to-date, Netmeds runs bots that lookup supplier data and match products and prices against its product catalogue. This helps in identifying new products, changes in pack sizes and prices. What could take weeks to get this information, Netmeds is now able to get it as soon as the suppliers have stock of the product or make changes.
Another RPA use case if that of onboarding of suppliers and interacting with their systems.
Pandit informs, “There are currently over one lakh pharma products available in the market. We use machine learning and fuzzy logic to map our catalogue to supplier catalogues. Over a period of time, on an average, we have been able to match 80 per cent of a supplier’s catalogue via our algorithms, which helps slash integration time tremendously. Our algorithms have also been able to detect new pack sizes, patterns in the way suppliers name products in their catalogue, which helps in on-going synchronisation between supplier systems and ours.”
Furthermore, in the area of digitisation and validation of prescriptions, every prescription received is digitised by a data entry team. Netmeds has built intelligence into the system to be able to understand certain prescription layouts and rate/score the validity of a prescription. This helps auto fill certain information and keeps improving. While this has significantly reduced the time for pharmacists to validate a prescription, it has also improved accuracy.
Role of AI
Ensuring validity of prescriptions and curbing misuse of prescriptions is critical in this tightly regulated industry. In line with this, Netmeds has built an advanced pharma module which facilitates digitisation of prescriptions, scores prescriptions based on certain mandatory data that a prescription must hold, reads patterns in prescriptions and finally puts out alerts to the team of pharmacists, if the algorithms pick out any suspicious prescriptions or orders. This module has been in use for the past three years, and the company is seeing greater accuracy and quicker turn-around times in terms of validating a prescription.
Explaining further, Pandit says, “Our pharma module relies heavily on AI to rate/score prescription validity. Apart from that, there are certain subtleties related to the pharma vertical which is very different from traditional e-commerce. A good example would be that of recommending products to a customer while browsing the site. We cannot recommend pharma products to a user. Therefore, we use machine learning and analytics to profile customers and recommend FMCG products or products that do not require a prescription that complement the pharma products the customer buys. For example, we would recommend a diabetic drink for a customer who falls under the profile of a diabetic.”
On how Netmeds is differentiating itself from other e-pharmacies, Pandit comments, “In this already crowded e-pharma space, the only other way to differentiate ourselves from our competitors is to provide a holistic health platform that includes online consultation, diagnostics and probably insurance in the future. These offerings do not make sense unless they are tailor-made to a customer. Therefore, profiling of a customer again becomes extremely important. Apart from just selling products and services to our customers, we also want to keep our customers engaged. We have a Health Library section which is a great source of content. We use the same customer profiling techniques to display relevant content and link the content back to products on our platform and vice-versa.”
In addition to customer-facing cases, the company uses ML and analytics in its operational processes. According to Pandit, these efficiencies are helping in reaching markets that the competitors cannot serve. He says, “There are currently over one lakh pharma products available in the market. There are also products that are only available in specific regions of the country. In addition to this, the prices of products change by batch, and manufacturers change the pack sizes of products frequently. To truly be a pan-India pharmacy, our catalogue and prices must be up-to-date. Also, the ability to quickly integrate with distributor systems is extremely important.”
In addition, Netmeds uses artificial intelligence to forecast demand and procurement requirements by pincode and tie this up to the warehouse that services the respective pincode. Forecasts are made depending on historical sales data, the type of the product, season, marketing efforts, etc. Based on these forecasts, demands are generated to suppliers, based on storage capacity of the warehouses, supplier ratings based on their previous fulfillment rates, time taken to fulfill demands, etc.
Netmeds has already experimented using blockchain in its supply chain process. It has worked on a proof of concept, but not put it in production yet. As Pandit informs, most of the customers search for products – even in the non-pharma category – rather than browsing through categories. In wake of this, the company is working aggressively on our search algorithms to show results based on the profile of the customer. If allowing generic substitution becomes mandatory, Netmeds could look at extending this to the pharma categories as well, but within certain boundaries.
“Subscriptions and refills are big on our list of things to improve in 2019. Subscriptions in e-pharma being a new concept in India, requires a certain level of human interaction to give the customer confidence. We are working on messaging bots and IVR solutions to play a part in this to improve efficiencies and remind/nudge customers about their upcoming refills and subscriptions, as well as help them complete payments with ease,” he adds.