Data collection to verify merchant’s consistency

[Crowdworks AI] ONSITE DATA COLLECTION

Project

Offline visits to over 7.000 merchants in a specific region and verification of information

Request of client

• Visiting approximately 7,000 card company merchants in Gangnam-gu and check for matching merchant information (business name)

• Merchant industry classification (large, medium, or small)

What difficulties did the client have?

You’ve probably seen it before: you pay at a restaurant and get a receipt with a completely different business name than the one on the sign. It’s either because the business is using the same card reader as the previous one, or the information hasn’t been updated. This data is important because card companies use your spending data to personalize their marketing. If you’re stuck with bad merchant data, it’s hard to target your business to the right customers. In order to verify that the merchant information was correct, we would have to visit the restaurant in person and compare it to the information held by the card company, But given the number of merchants across the country, this was not feasible.

🧐 Reasons for choosing Crowdworks!

• Developed a custom UI to support a map screen in the CrowdWorks Worker app and to allow users to select pins on the map to view information directly.
• Crowdsourced data collection from 350,000 people: offline data collection in a short period of time 

Solution of Crowdworks AI

‍Offline visits (data collection)

First, we decided to verify the accuracy of the merchant data in Gangnam-gu, the neighborhood with the highest concentration of restaurants. We visited about 7,000 merchants in Gangnam-gu and verified that the actual information matched the information held by the card companies. 

The key was to keep workers from going to overlapping locations to make the project move faster. Crowdworks designed a taskscreen for the project that scattered pins across the map, and if one worker grabbed a pin, it would disappear from the other workers’ screens.

At the beginning of the project, there were a lot of restaurants for workers to visit and they were densely packed, so workers could do a lot of work with little travel. Later in the project, many of the restaurants had already been verified and the hourly rate would be lower for the distance they had to travel. Crowdworks anticipated this and started collecting at a relatively low unit price and gradually increased the unit price, doubling the unit price for every 50% reduction in the number of pins to motivate workers to finish the project.

Inspecting Data

Another mission was to take and upload photos of signage to verify that a worker had actually visited a restaurant. We found that some workers were searching the internet and uploading photos of their monitor screens without actually visiting the restaurant. We rejected all of these tasks through a manual review, increasing the accuracy of the data based on actual visits.

Client Comments

“Accurate merchant data is of paramount importance to us. With over 2.2 million merchants across the country, it was very difficult to verify the accuracy of the data. Crowdworks completed due diligence on 7,000 merchants in just two months, determining the accuracy of the data and turning it into meaningful consumption data. When they shared the first customized workspace with us, we felt like they had our back for the duration of the project.”