Listen to this story
In present times, we use navigation apps like Google Maps and MapMyIndia to explore new areas. From a consumer’s perspective, they are indeed good and provide all relevant information on traffic, restaurants, clinics, etc at a given location. But is there a Google Maps-equivalent for a businessman wanting to set up a store in a new location? Who does the recce for him? Who brings in intel?
GeoIQ does that and much more. A locational-intelligence platform based in Bengaluru, GeoIQ has achieved 10x growth in its ARR over the past four quarters and is on its way to registering additional 5x growth in ARR in the next two quarters.
Sign up for your weekly dose of what's up in emerging technology.
In an exclusive interaction with Analytics India Magazine, CEO Devashish Fuloria speaks about GeoIQ’s business models, flagship products, the challenges that the firm faces and also the future plans.
AIM: What is the story behind GeoIQ?
Devashish: Around 2018, we were discussing a lot of problems around data science. One of the biggest retailers in India was trying to implement their plan of opening a chain of 5,000 grocery stores across India, but faced problems in identifying the most suitable locations for their stores beyond metro cities. Generally, entities depend on information coming ground up in such cases. However, it becomes tricky to understand which information is good enough because entities generally do not know the whereabouts of a location. Now, when you set up the store, it’s a huge expense. So, if it doesn’t work after one year, you’re losing a lot of business value and capex.
Hence, we came up with the idea of providing location-related information in a centralised manner. We started checking out the data for various locations. We looked at government data, public records and the internet and eventually layered all the information on the map. Today we tell businesses everything about an address on a street without actually being there – whether a location is reachable, if it is risky, would there be demand for an expensive apparel, etc. We provide answers to such high-value queries in simple numbers. When a client enquires about a location, we give them a score between 0-100, which signifies how good that location would be for the business.
AIM: That’s interesting! Please elaborate on some of the latest use cases provided to your customers, particularly in deploying AI/ML?
Devashish: Our use cases are across industries, the largest being in fintech, retail and e-commerce. The biggest use case for fintech is risk prediction. Before disbursing loans, fintechs need to know about the credit-worthiness of their customers. When applying for loans, customers provide information like PAN card details, credit card numbers etc. Based on location-specific information derived from these details, we predict how credible a potential customer would be.
For e-commerce, we solve the concern related to return-to-origin by augmenting their user data with hyperlocal intelligence. Other use cases include accurate affluence prediction, fraud prediction, claims propensity, collections model, and lead prioritisation.
AIM: What are the challenges you have faced so far and how is GeoIQ addressing them?
Devashish: When we started out, we realised there was a strong need, yet a limited understanding of this data. We had to educate our users, unsuccessfully at times, on how best to use this vast data repository. Business analysts always looked for specific numbers that they thought were important. This was eventually addressed. Now we provide clients with scores.
Then there were challenges like sourcing valuable, accurate, and quality real-world data. For example, good-quality data for Indian boundaries were not easily available, nor accessible. Also, the pin code, city, state, and other boundaries were not very clear beyond administrative limits. Therefore, we built a GeoAllocation engine to define the boundaries better (20% more accurate than existing boundary definitions) and mapped the addresses on these new boundaries.
Another challenge was with respect to the sanctity of data. We had to check and validate data points from various sources to ensure the accuracy and truthfulness of information that we had gathered from public data sources.
At present, data discovery is a big challenge. Identifying which data is useful and impacts a use case directly is a herculean task and often lands on trial and error. The NoCode ML platform has been created to solve this problem. The solution has completed the beta phase and is soon to be launched. Businesses would be able to create multiple models in no time and experiment with data attributes to identify the best fit for their use case.
AIM: So, in a way, GeoIQ is trying to address the problem of bad data with the NoCode ML platform. Bad data is a global problem. How else do you plan to tackle the menace?
Devashish: Data is good. We do not believe in bad data. Bad data is good data not presented in a structured and consumable format. Yes, we don’t deny that there are many problems with data when you look at companies’ databases. But you can’t solve multiple problems. Thus, we are focusing on one specific area, i.e. location, that helps us bind a lot of different datasets together in one cohesive way.
AIM: What technology does GeoIQ use to analyse a broad data stack to provide customised solutions to its clients?
Devashish: We use our proprietary algorithms to transform real-world data from 600+ sources in a structured format, categorised under 3000+ attributes. We have built advanced machine learning capabilities that help us provide custom models to the clients, based on their use cases. The explorer and APIs, our standard offerings, are backed by ML capabilities. Explorer helps a user to get information about a specific location or compare up to three locations at a time for a specific attribute. APIs could be directly embedded in the code to augment it for location data.
AIM: GeoIQ raised USD 2.25mn in funding recently. How do you plan to channel these funds?
Devashish: As a product-led company, our core strategies are around product development. We are channelling a part of these funds towards building advanced tech capabilities and a team of skilled human resources.
Moreover, we want to take our methodology and build a similar system for different geography. As of now, it’s the US. We are opening our platform to data scientists, where they can create their own location models and get specific location answers related to the US.
AIM: Why US, and not the larger region around India?
Devashish: Southeast Asia has very similar problems to India, which is basically that we are all data poor. Southeast Asia, for us, maybe twenty geographic entities, which means you have to find 20×500 sources of data to build up a base. But US being one massive geographic unit, you can scale up the data very quickly, and also, the data is easily available.
AIM: At present, GeoIQ offers three products – No code ML, API and Explorer. Are there plans to expand the product range? What is in the pipeline?
Devashish: More than expanding into new products, we are trying to expand our solutions for new use cases, source more latest data, and improve ML models. Every new data source that we add enriches our offerings a step further. Our solutions are seeing wide adoption. New companies with very specific and niche problem statements are reaching out to us. We are excited about those and are expanding our use case scenarios.
AIM: In India, small businesses often suffer due to location issues. From the CSR perspective, do you have plans to engage with that section of society?
Devashish: We already have a freemium model for small businesses to access our solutions with a certain number of free credits. Also, some basic information about towns and villages and streets is being provided for free.
AIM: Do you think discussions around data regulation would impact GeoIQ’s operations?
Devashish: I think it was a very conscious choice we made four years ago that we’re not going to get into personal data. It is anonymised information that we deal with. Government data doesn’t name people. Information like Nielsen’s data talks about spent patterns at a location and doesn’t have any personal information. We don’t know who Person A is. That information doesn’t exist in the system. If anything, GeoIQ’s methodology sets a template for how data should be used across multiple systems while remaining ‘privacy first’. Therefore, there will be no adverse impact of the changing data regulations on GeoIQ’s operations.
AIM: What is the roadmap ahead for GeoIQ? What development can we expect in terms of new solutions and services?
Devashish: Modern AI systems that exist within companies are based on what users are doing with their apps. Nobody knows what the users are doing in the real world. For example, Person A is just an entity for Amazon, and all personalisation will happen based on A’s movement and behaviour on the app. It’s likely that A’s neighbour has similar preferences as A, but Amazon doesn’t know about them. We are seeking to tap these sorts of common interactions and make them part of the modern AI systems.
From a roadmap perspective, we are first targeting high-volume transactions, thus engaging a lot with fintechs, insurance, and e-commerce players. We are gradually increasing our verticals in India. In terms of new solutions or services, we are trying to strengthen our foothold in the insurtech and e-commerce sectors by addressing more use-cases and at a larger scale.