Today, everyone uses Google Maps to navigate to any place. Similarly, why shouldn’t every company have their own maps to understand where its assets (people, bikes, scooters, cars, devices, products) are and where they should be?

The question of “where”.

If you are a company that has moving assets on the ground connected to the internet with a unique location at all times — chances are around 80% of your data has that spatial component!

But, we all know that location data is pretty futile when viewed in an excel sheet.

In order to make sense of the lat-longs, you need to be able to visualize them on a map, to understand where “they are coming from”, to identify the spatial patterns of your business metrics and how they change over sectors in a city and within cities.

An amusing anecdote!

One of the most important problems for an e-commerce (apparel) company was frauds and abuse of discounts. On plotting their data, they found that even frauds had a spatial pattern!

Most of the cases were in areas around universities, which meant that there were probably students abusing their coupon codes.

It’s economics 101!

Taking a step further from spatial patterns, visualizing your location data also helps you analyze your supply-demand gaps to run varied spatial models.

This buzzword, supply-demand!
The definition can vary amongst organizations. For Grab, a single unit of supply is a driver who is online and idle and a single unit of demand is a user checking fares for a ride. For Airbnb, a unit of supply is one Airbnb unit.

If you are a hyperlocal/ride-sharing/ hospitality company, your goal should always be to ensure users ordering a delivery/service/ ride, are matched to the driver/delivery/service partners who are closest to them, thereby bridging the gap in the shortest time possible. In other words,

Demand-supply matching is at the heart and soul of your business.

Aligning these gaps will inevitably lead to generating more money per unit — henceforth improving your efficiency and experience for your users.

You can implement planning actions to generate supply in high demand areas or examine what-if scenarios to increase demand in high supply areas!

Now that you know where these gaps are, what would you do with it?

More $$$ for your partners!

One of the simple ways to make this data actionable and bridge the gap is by directing your “excess supply” to high demand areas.

Companies like Grab, Uber, and Swiggy have built partner apps which show heatmaps of demand areas. They “incentivize” the partners based on the surge that they charge.

For example, in periods of high demand, Uber adds a multiplier on its base pricing, to ensure the drivers rush into this area to earn more and the riders who can afford the ride can take it, while others can wait till the fares go down.


Makes sense, right? Drivers or partners shouldn’t approximate where to traverse in their idle time based on their intuition. There should be real numbers backed by real maps behind those movements!

This proves very effective in times of bad weather, high traffic, rush hour or special events like New Year’s Eve or Diwali!

If you want to delve deeper into surge pricing,  check this out:

How does Uber do Surge Pricing using Location Data?
How location data helped in figuring out one of the best pricing innovations of modern times!

Lure your potential customers!

In the case of carpooling/bike-pooling companies, the deviation from the rider’s original route is not quite possible. Since supply cannot be tampered with, there is demand for the rescue!

You can send targetted promotions to your customers in those low-demand areas — providing them with offers and discounts to entice them.

Whenever your supply-demand ratio falls below a certain threshold, cluster all the low-demand areas and send automated promotions. Experiment and ideate to conclude what works on the ground, on which kind of customers, in which areas.

If you want to read up more about geo-marketing, check this out:

Things you need to know about Geomarketing in 2020!
In a world where “location” is the new cookie.

Handpicking the next route!

Supply-demand gaps also help in pinpointing the location of your new distribution centers, or dividing a warehouse into multiple centers, or aggregating all the centers to one common location.

Understanding how your current orders are clustered goes a long way! Clustering basically means breaking large data into smaller groups using certain classification criteria.

The locations can be pinned easily based on where your customers can be reached more effectively.

If your model is more like a shuttle service, you can carry out the same exercise when your routes are fixed — by identifying where your pick points should be or which set of new routes should be adopted!

Moving demand and playing God!

Demand prediction is very common in Data Science activities across companies. However, these models don’t incorporate the “location” element.

In case your company deals with the allocation of assets such as bikes, cycles, etc., the location should be optimized not only on the historic supply-demand gap but also, how much demand is predicted in that area at fixed intervals.

Uber, for instance, knows what times, day and areas have the highest demand. They crunch their past numbers for 3–4 weeks for a city, further breaking into specific hub/pocket by the hour of the day and day of the week.

Grab goes a step further — It has a Travel Trends widget on the rider app for the riders to know the predicted demand across time. The aim is to encourage time-insensitive demand to opt for a ride later, serving passengers with more urgent needs.

For a hyperlocal company with warehouses, understanding and predicting demand can help to stock different products in different distribution centers in the city.

Even a tier 1 city has pockets that behave like a tier 3 city. To give a very crude example, a Nutella would sell in a very different region than a Parle G!

Demand prediction is tricky, while it is a lot about your historical data, it is also about using external data sources in your models such as weather, traffic, etc in your models to be more precise and accurate.

Even a tier 1 city has pockets that behave like a tier 3 city. To give a very crude example, a Nutella would sell in a very different region than a Parle G!

Demand prediction is tricky, while it is a lot about your historical data, it is also about using external data sources in your models such as weather, traffic, etc in your models to be more precise and accurate.

Painting better recommendations!

According to this Airbnb article,

If you don’t have location-based dynamic pricing, you are simply leaving abundant money on the table.

A hospitality company has a fixed supply or occupancy. Demand here is a factor of seasonality. Even within a city, in case of an attractive season, the demand can change based on the neighborhood or location the branch is present in.

Supply-demand gaps help you with devising good recommendations and your pricing strategy.

How can we at help?

After all, being in the hyperlocal and ride-hailing space, your biggest goal is to ensure that your users get access to a ride, when and where they need it, as fast and early as possible, while providing an excellent service to your riders, and a better livelihood to your partners.

And we are here to help you do that. How you may wonder!

At, our mission is to bring the same level of granular analytics as Uber and Grab to every company collecting location data.

We do this using a system of hexagonal grids of 1 sq. km which we treat as the smallest unit of geography. This acts as a base for every ML algorithm, analysis, and service that works in the context of location.

Doing location-based pricing, promotions, and allocation at granular levels and different times of the day can prove to be very lucrative!

In the coming years, a lot of your success would attribute to being able to crunch data to predict what customers want, when they want, by what channel and most importantly, where they want it available.

Without adding the spatial context to your algorithms, you are ultimately leaving all these decisions for guesswork. Gain the “edge” that it takes to join the likes of companies like Uber, Airbnb, and Grab.

Similar Reads:

How we’re building our geospatial analytics product using first principles
Our philosophy on analytics at Locale!
Geospatial Clustering: Types and Use Cases
Deep dive into all the different kinds of clustering with their use cases.