Dubai, United Arab Emirates: Fuelled by the COVID-19 pandemic, when shoppers chose to avoid crowded places, the online grocery delivery market in the Kingdom of Saudi Arabia is booming. Founded in 2016, Riyadh-based online grocery platform Nana is on a mission to have a 40% share in this market by the end of 2026. It has already expanded into Egypt and is looking at other regional markets.

“What differentiates us from the competition is that we serve all of our customers’ shopping needs, whether that’s a small drop of day-to-day essentials delivered within 15 minutes through our dark stores or their regular supermarket big shop with a choice of over 22,000 products,” explains Samer El Sahn, Chief Product Officer at Nana. “And our Nana Business app gives small and medium-sized businesses the opportunity to buy and sell their goods too.”

Attracted by its technological maturity, the propensity to scale, the suite of tools on offer, and its cost-effectiveness, Nana launched on Google Cloud in 2016. And, to stay ahead in the market, the company is constantly looking for ways to improve its customers' offerings and services. As with any new entity, there are always upgrades and refinements to be made and that is what El Sahn and his team focus on. The team is carrying out what El Sahn calls ‘back-to-basics,’ assessing each and every link in the user journey to see where improvements can be made. 

Fine-tuning for scaleup with Google Cloud
“When you move from a very early stage startup to a scale-up, you start seeing where your infrastructure can be improved,” reveals El Sahn. “We have a very mature system when it comes to functionalities and features, but what the team is focusing on now is going back and making more efficiencies and better experiences for our customers.”

Once the Google Cloud team was aware that Nana wanted to make improvements, they worked closely with the company’s engineering team to ensure that they deployed the Google Cloud solutions necessary to take it to the next level. 

“Google Cloud really helped us to rethink our architecture and provided us with Google Cloud credits, which meant that we could experiment with lots of tools without being afraid of costs mounting up,” says El Sahn. “We were even given access to the not yet launched Cloud Fleet Routing tool, which lets us use machine learning to optimize our truck dispatch management. It has vastly improved our service as it’s much more intuitive than other systems we had looked at. To me, it’s like magic and we haven't even scratched the surface of what it can do.” 

By using Cloud Fleet Routing, Nana was able to improve the utilization of its fleet by almost 40% and cut its fulfillment costs too. Now, it can group orders per truck according to delivery location and delivery time, automatically providing drivers with the most efficient route for delivery using Google Maps. Next it plans to use Google Maps to let customers live track their deliveries.

Centralizing data with BigQuery
Having optimized delivery, El Sahn began looking at how else he could use the vast amounts of data that Nana held to benefit other areas of the business. For this, he turned to the scalable data warehouse BigQuery. “All our data points from finance to marketing are now sent to BigQuery,” he reveals. ”And we're really leveraging this data by allowing different teams to access it to build their own dashboards, so that they can analyze and understand different parts of the business.” 

Nana has dashboards, for example, that show how each dark store is performing, how much stock they are holding, and the average delivery time for each of its trucks. “BigQuery is now the central point for all our daily operational metrics from which we draw all our business intelligence. It helps us to enhance all of our divisions,” says El Sahn.

Nana is also using BigQuery to segment its users based on their behaviors, so it can see how regularly they use the service and what their average spend is, for example. Based on this, it creates automated personalized campaigns to send to customers to increase their engagement. In the future, the company also plans to make it possible to analyze what people are putting in their baskets to personalize campaigns further.

And, to ensure customers get the best out of the Nana app, the company uses the app development tool Firebase to A/B test features. “We’ve done some interesting experiments to validate different hypotheses,” says El Sahn. “We wanted to see, for example, if our customers are more focused on which store their products come from or the product itself. So we gave them two options: they could build their shopping cart first and then choose which supermarket we source those products from or vice versa. What we found, which surprised us, was that our customers had specific preferences to where their items come from and we wouldn’t have discovered that without Firebase.”

Predicting demand with Vertex AI
To more accurately forecast demand to reduce waste and better control stock levels, Nana is harnessing the power of Vertex AI. “We have always had to juggle the availability of stock and the time that it spends on the shelf and we have trialed other solutions, but none have delivered. Vertex AI has solved this problem for us by automating the purchasing of items from our suppliers,” says El Sahn. 

“When looking at how much of each product to buy, it takes into account seasonality, the time it takes to receive delivery from our suppliers, etc. and it means that we always have the right amount of products in our warehouses and are only paying for the storage space that we use.”

According to Nana, the new Vertex AI-backed machine-learning solution is now predicting stock levels with 70% accuracy. As it continues to learn from new data that is being constantly gathered, this is only expected to improve.

Rapidly scaling up and down
While Vertex AI helps the company forecast demand, the automated and scalable Kubernetes service, Google Kubernetes Engine (GKE) helps the platform efficiently and economically manage that demand. By automatically deploying and managing Nana’s Kubernetes capacity, it ensures that the company’s infrastructure can nimbly scale up and down to meet demand and that it only pays for what it uses. 

As with any business, Nana is very focused on maximizing the return on investment for all of the technology it deploys, and Google Cloud Premier Partner Rackspace helps it cost-effectively apply all of its Google Cloud products. “Rackspace has helped us consolidate our billing and shown us how to optimize all our Google Cloud solutions so that we get the best out of them at the least cost,” says El Sahn.

Working with continually developing technology, the Nana team will never be able to say that they have finished optimizing their platform, but El Sahn says that with the support of Rackspace and Google Cloud, he feels that they are creating a future-proof infrastructure.

“Google Cloud now feels like an extension of my team,” he concludes. “By working with us on architecting our own solutions, they have helped us greatly improve our delivery response times and forecasting, speed up our development cycles, reduce our cost per order, and, most importantly, improve our customer experience.“

-Ends-