What does anyone's location have to do with individual risk profiles? You have not bought any property or a vehicle, yet certain algorithms have your location data to tell if you will repay the loan in time because you live in a specific geographical area. Sounds insane but it is true. Geospatial data is fast becoming the goldmine of information for companies that build predictive models for a variety of use cases. GeoIQ, a hyperlocal data intelligence firm takes geospatial data modeling a notch up by deploying pre-built predictive ML models. Analytics Insight has engaged in an exclusive interview with Ankita Thakur, Co-Founder & CDO, GeoIQ.

1.Kindly brief us about the company, its specialization, and the services that your company offers.

GeoIQ predicts where in the country business opportunities or risks exist, down to the exact point on a map. Consider this: a borrower asks for a loan online from a specific location – 143/1 Vishwas Khand, Gomti Nagar, Lucknow. In real-time, GeoIQ tells the lender how affluent/risky the customer is, just using the exact location.

GeoIQ has developed bespoke algorithms that churn vast amounts of scattered, unstructured data, to deliver instantly these ultra-fast answers, thus reducing the cost of manually checking locations by 100 times. This is possible by tracking data sourced from over 600 different sources that include government, open data, satellite imagery, surveys, and many more. GeoIQ's inbuilt feature store converts this data into 3000+ attributes and updates them at set intervals.

This data powers prediction models to solve issues such as credit risk prediction, fraud detection, sales forecasting, hyperlocal targeting for OOH and digital campaigns, site selection, and many more.

2.Please brief us about the products/services/solutions you provide to your customers and how they get value out of it (similar to above)

GeoIQ has two products – data and predictions.

Depending on the use case, our customers can access the data – through Explorer (a SaaS tool to select and download information) or directly through Data APIs. A few examples of the attributes at play are demography, socio-economic factors, infrastructure, healthcare, crime data, and much more. You can select from 3000+ attributes listed in the data catalog.

For advanced data science teams, the NoCodeML tool is one of the first global machine learning products with its own feature stores. It enables data scientists to find patterns in the location data, build an ML model and deploy it in real-time use cases within a matter of minutes. The model responses are available as APIs that can be easily integrated into existing codes for providing location insights. It is an AI for customer data augmentation at granular levels.

GeoIQ has also built RetailIQ, a pre-built ML solution for the business user in retail. It aids with site selection for retail expansion, forecasts sales figures for each store, merch mix prediction, and hyperlocal targeting.

3.What is your biggest USP that differentiates the company from competitors?

Coverage, accuracy, and a wide variety of attributes are available through a single API.  We cover the whole geography (all possible addresses) within the boundary of India and the USA. The accuracy of our prediction models is unmatchable. Also, we provide a wide variety of attributes that are available through a single API. Our No code ML platform enables businesses to try out 3000+ attributes in their prediction models with minimal effort and for free. It enables them to do exploratory analysis, shortlist attributes, build models and deploy them as real-time APIs in one click.

4.Which industry verticals are you currently focusing on? And what is your go-to-market strategy for the same? Would you like to highlight a few use cases where analytics has benefitted the organization tremendously?

GeoIQ has made its mark specifically in the retail and BFSI sectors. We are powering real-time decision engines at multiple organizations like NAVI, Paytm, HeroFinCorp, DMI Finance, Zepto, and Lenskart among others.

With GeoIQ's unique offerings, our fintech clients have witnessed a 10 percent reduction in NPAs with better risk profiling and a 20 percent increase in book size, especially for the New to Credit user segment. We are c expanding beyond the domestic market, beginning with the US market, as we believe our offerings are globally competitive and our growth momentum also showcases the same.

As our products and technology are industry agnostic and find use cases across verticals and sectors, in time, many other use cases will be targeted.

5.AI is projected to be the next market. How is AI contributing to the making of your products and services (for the US market)? What is your roadmap for the US Market?

The industry broke records during the pandemic, as AI funding doubled in 2021 compared to 2020. Artificial intelligence (AI) has the potential to transform the productivity and GDP potential of the global economy.

At GeoIQ, we use machine learning capabilities to draw critical insights and identify patterns and trends in real-world data applied to a use case. Such capabilities allow businesses to build accurate prediction models that drive informed decisions. We enable predictive analysis for revenue prediction, credit risk prediction, fraud prediction, demand prediction, site selection, user behavior analysis, and much more.

For a long time now, spatial data has been a domain for spatial data analysts. But by building an AI layer atop location data, we are bringing the solutions around spatial data and analytics into the hands of advanced data teams that are building large-scale prediction systems for fintech, insurance, e-commerce, D2C, and others.

Our data is already live in the US with more than 5000 data features. We have been actively pursuing data collection, processing, and transformation for some time now. Also, the ML capabilities that we have built are geo-agnostic and can be scaled up for any set of data for any location globally. Location data has been collected from all publicly available and government data sources and categorized under geo-attributes in our feature store for the US landscape. The variables or attributes are created and categorized keeping in mind the administrative divisions in the US which are not the same as in India.

Ahead of this, we are opening up our solutions for the US market in a phased manner, keeping in mind the ability of data teams to build models on the US datasets, and varying nuances of use cases and market appetite as compared to our current Indian market.

6.How do you plan to revolutionize the US market and what are your plans to tap the market?

The abundance of real-world data is a boon only when we have techniques and algorithms in place to make sense of it. And that is what we intend to do as we enter the US market. The data is already there, and that too in abundance; the problem is in finding insights from them. That's where GeoIQ's platform steps in. We set up pipelines from multiple data streams and transform them to be fed into ML models that generate critical business insights.

Unlike the Indian market, the challenge that the US market faces is not the lack of quality data. The bigger challenge is sorting through these tens of thousands of data points and identifying which ones add value to a specific use case. GeoIQ not only makes this data available in one place but also provides mechanisms to quickly test and identify the ones that bring in the most value for businesses' specific needs.

With hyperlocal insights about user behavior and location dynamics, businesses can operate at scale and back all their decisions with logic. We eliminate guesswork and gut calls while making important business decisions. We intend to empower the data science teams with data and tools that extract value from data, thus reducing their time to geo-based insights massively.

7.How does your company's strategy facilitate the transformation of an enterprise (US perspective)? How does your company's rich expertise help uncover patterns with powerful analytics and machine learning for the US market?

The US market is one massive geographic unit with millions of locations having their individual characteristics; some of them make them similar, and some of them make them opposite. Till you don't have any data, these correlations are nigh impossible to draw. The good news is that the US is extremely data-rich. The second good news is that technology now allows us to find patterns that were humanly impossible to identify. So, we are sitting on the cusp of data and AI systems that can be enabled to create a new kind of AI – location AI – that is unlocking higher efficiencies in automated decision-making.

A brief explanation of how is it done?

Our ML engine identifies location patterns in clients' data – what makes a street good for business, what makes it bad for business, the high-value users – are their locations similar in some way, etc. The correlations are drawn with GeoIQ's datasets. Once a model is built, it takes less than a second to identify the similarities between existing and new locations, at a scale and speed unseen in the geospatial world.

8.Could you highlight your company's recent innovations in the AI/ML/Analytics space?

Our NoCode ML platform helps businesses create models using their data and GeoIQ's data within no time and for free. It's one of the first ML products in the world with its own feature store, specifically tuned for identifying location patterns. It's a new kind of AI infrastructure platform that will allow other data teams to build their own models on top of it.

For business users in Retail, our RetailIQ product is suggesting streets open stores across the country within a single click – a process that was fraught with errors and used to take months or even years.