Tracking potential tax evaders on Instagram

The digital economy is booming in Medellín. So is digital tax evasion. To identify potential tax evaders, the local treasury department used to detect unregistered online stores manually. Since the use of social media for economic activities has grown exponentially, the agency has now developed a bot that automatically scrapes Instagram for such stores - and officers can use their time instead to work with store owners to formalize their businesses.

Innovation Summary

Innovation Overview

The digital economy has been a boon for Medellín, Colombia. The city is attracting capital and talent, and the digitization has helped it become one of the most innovative cities in the world. Yet for the local treasury department, online sales have become a challenge. Tax evasion, which is common offline, has increased online. The government took the first measure in 2017 and passed a new law on the digital economy that explicitly included e-commercial activities in the tax code. Still, many small businesses, whose online presence often is just a Facebook or Instagram page, do not pay the required Industry and Commerce Tax. In some cases, they are not aware that they should pay, in others they resist because of a sentiment that the government is already making enough money.

The small investigatory “Fiscal Intelligence” unit at the treasury department took up the challenge. To get a first impression, they began to manually search for stores on social media and check whether the owners had registered with them. According to their own estimates, they were able to identify six stores per hour and knew this was not a sustainable process.

They set out to find a new solution with the Government Innovation Lab of the city’s business and innovation center Ruta N. For two years in a row, the Innovation Lab has invited about a dozen public entities to participate in a facilitated process to address challenges that they felt stuck within an innovative way. In a number of workshops, public agents narrow down the challenges to specific questions they seek to address and then post an open call for solutions to NGOs, businesses, and universities in Medellín.

In this process, the treasury department decided to work with the local start-up Grupo Boötes, which had suggested to develop KBoot, an algorithm that scrapes Instagram profiles and matches data with those of local telephone operators to identify unregistered businesses. After the team at Ruta N selected this project as one out of three that would receive monetary support for a three-month pilot, they set out to test their idea in three phases.

In phase one, Grupo Boötes developed and ran KBoot, a software dedicated to identifying people and businesses who carry out online commerce on Instagram (Instagram was picked as the social medium to work with during the pilot phase due to its large user base in Medellín). The software was programmed to search Instagram for public profiles and publications in Medellín that use relevant hashtags, keywords, and names that are associated with online sales. It then downloaded relevant data, such as username, number of followers, number of publications and contact telephone numbers, to a database, only selecting profiles with regular activity after 2016. At the end of phase one, KBoot had identified 20,828 profiles associated with commercial activities, at a rate of 45 profiles per hour.

In phase two, the treasury department went to identify the individuals behind these profiles. To do so, it cross-checked the names with their own databases and compelled all telephone operators in Medellín to provide information on 9,080 users that had provided a telephone number on their profiles. Until early 2019, five out of eight operators - covering about 50% of the market - had responded to their request and in total, the office identified 2,683 individuals. Of those, only 453 were registered with the treasury department, and 107 of these said they were not operating at the moment. 2.230 individuals were not registered with the treasury department but had been identified as selling merchandise on Instagram.

The third phase, which is still ongoing, relates to the integration of these businesses into the regular economy. The treasury department early on decided not to begin by sending fines to these unregistered businesses, but to first include them in ongoing awareness-raising campaigns that it is carrying out with the local chamber of commerce. It also applied the new law on the digital economy, which allows it to grant special tariffs to new businesses. It does not want to scare away business owners, especially since “migrating” to a different municipality is simple online. The potential tax income is substantial, after all: 2.337 potential new accounts at a minimum tax rate of COP 30.000 per month makes COP 70.110.000 (or EUR 19.881,42) per month and 841.320.000 (or EUR 236.572,62) per year.

There is a downside to this approach using carrots yet no sticks: the sensibilization efforts have proven little successful. Out of 800 businesses that were contacted, only 30 took part in such efforts, and only 17 have registered business so far. Yet the treasury department remains undeterred and will now move to a different approach, in which they will contact all identified business owners via text messages and WhatsApp in a more personal manner. And if this really does not work, it can always resort to fines.

Innovation Description

Innovation Development

Innovation Reflections

Leave a Reply

Your email address will not be published. Required fields are marked *

Year: 2018
Level of government: Local government


  • Implementation - making the innovation happen
  • Evaluation - understanding whether the innovative initiative has delivered what was needed
  • Diffusing Lessons - using what was learnt to inform other projects and understanding how the innovation can be applied in other ways

Innovation provided by:

Join our community:

It only takes a few minutes to complete the form and share your project.