Finding the BestPlace: Meet an AI Solution that Knows Your Next Business Location

Willing to open a brick-and-mortar store and not knowing where to start? Or perhaps you are looking to supply your product to a local grocery shop and having no idea which one to choose?

You are not alone. Every year thousands of new entrepreneurs embark on a rocky business journey of understanding their customers and their behavioral patterns while well-established companies do their best to outpace their competitors. To find the best location for accomplishing their goals, the two often turn to methods, such as deploying “field agents” or gathering data manually.

A Moscow-based mathematician and business consultant Alexander Kiryanov was among those who first noticed how obsolete these methods were. He decided to fix it by creating an Artificial Intelligence-powered predictive geo analytical tool that provides end-to-end visibility of local shopper patterns.

Dubbed BestPlace, the solution became a recognized industry name, securing millions of dollars from investors. Its clients include several major companies such as Leroy Merlin and PepsiCo, which benefitted from BestPlace’s assessment of sales potential and audience coverage for every point of sales in 2019.

Having scaled dizzy heights, BestPlace is ready to cast light on its path to success.

Modernizing the industry

Although the consensus is that online retailers and Instagram have killed the brick-and-mortar stores, the demand for them, as well as physical supplies of goods, remains important for many businesses.

Having worked with multiple companies, Kiryanov noticed that the reliance of companies on old-school, borderline obsolete data collection methods when opening and choosing new locations was strikingly high.

“Most retailers analyzed everything manually, presenting information to the CEO on an Excel sheet with map screenshots. That is plain ineffective in my opinion,” notes Kiryanov.

After observing the businesses for several years, he decided to explore a potential business niche. The watershed moment took place after he started collaborating with a large client who aimed to open five hundred stores a year.

“Their goal was to make fewer mistakes place-wise. It was important for them to find the best location and gauge their revenue. That is when I realized that it is time to develop a new solution that caters to their needs,” says Kiryanov.

Back then, the market lacked a tool that would become a functional, general analytical platform for collecting tons of data while verifying its accuracy and making predictions. Since Kiryanov had a business partner who was Head of Machine Learning Department at Yandex, Russia’s equivalent of Google, he decided to join forces and create an AI and machine learning solution that would fulfil the business’s demand for geographical accuracy.

This is how BestPlace was born. Already in 2017, the startup raised $1 million from business angels worldwide. The solution also successfully participated in the Alchemist Accelerator, a San Francisco-based initiative focused on accelerating the development of early-stage ventures that monetize from enterprises.

To make predictions and deliver results, BestPlace collects a bevy of data relating to geographical patterns. For example, which sort of transport is used in the area, what kind of stores are around, the number of households and cars, etc.

“Say you are planning to open a store or start supplying your goods to a certain shop. To help you do that, we analyze over 40 different consumer patterns. We use mathematical models to make predictions. Our team features four data scientists and more than ten data engineers who oversee the process,” explains Kiryanov.

At the same time, he notes that the company put a lot of thought into its actions when developing its analytical model: “For example, there is a gym nearby. It could be hypothesized that the person working out in it pops to the shop nearby to get a soft drink. We test this hypothesis. If confirmed, we recommend that the relevant company acts accordingly.”

How accurate is accurate?

Although BestPlace made headlines in the business space, at first, the company faced several challenges. One of them pertained to data collection and accuracy.

“Imagine opening a map and zooming in on it. You will find all sorts of stores there. Yet it does not necessarily mean that they exist. For example, the map might claim that this shop is a bakery, but in real life it is something entirely different,” notes Kiryanov. “Because of this, our accuracy rates at the beginning stood at a mere 60%. Some of our clients were dissatisfied with the result. So, we had to look for ways to improve it.”

To tackle the problem, BestPlace turned to the crowdsourcing data labeling platform Toloka, which is popular among AI solutions around the globe both due to its unique expertise in spatial crowdsourcing tasks and cost-effectiveness.

In BestPlace’s case, Toloka’s performers, tolokers, — as many as 50 thousand people at hand — collected publicly available information about the neighborhoods of interest. Their task is to explore the area, take pictures of relevant public places, such as shops, and feed it into the system. As a result, BestPlace got access to ample volumes of information that helped improve its machine learning algorithms, improving the accuracy of predictions.

In fact, the data turned out to be so precise that BestPlace managed to reach an accuracy level of 80–95% depending on the industry type.

“The clients received analytics and recommendations as well as photos of the places of interest. This turned out to be a game-changer. They literally saw who their competitors were and made decisions accordingly,” emphasizes Kiryanov.

As the company continues to grow, so do its ambitions. BestPlace is already providing high accuracy results for multiple big companies and unique insights into customer behavior. Its automatic insight generator, for example, helps companies find out which flavor is likely to be popular in the neighborhood.

“We help companies better understand their customers, their behavioral patterns, and how they can improve their performance. The average monthly bill for our service is $500, yet it is worth every penny,” asserts Kiryanov.

With the world slowly emerging from the pandemic and the economies reviving, BestPlace is positive that its future looks brighter than ever.