Salvador Caetano is a portuguese holding company overseeing over 100 enterprises in vehicle assembly, production, distribution and retail, industrial equipment, and aeronautics, among other sectors. With a history of over 75 years, the company has evolved from a bus manufacturer into a major force in the automotive industry, employing over 8,000 people across 44 countries on three continents.

Salvador Caetano África, responsible for distribution and retail in African markets, has consistently faced challenges with competitive analysis due to the dispersed and hard-to-access information on the continent. Unlike in Europe, where they can access public databases to check competitors’ pricing for various car models, in Africa they must search websites, social media pages, and directly contact sellers via email to gather market information and set prices. The task is further complicated by exchange rate impacts and political issues, making price collection a pressing issue due to relatively high volatility.

Salvador Caetano estimates that market research makes up between 20% and 25% of their monthly activities, equivalent to 4 to 5 full working days. Seeking to automate market research as much as possible, they chose DevScope, with whom they had previously collaborated on other initiatives, to tackle this challenge. The project began with automating email processes.

“We spent around 25% of our time after information, and it doesn’t always arrive.”

– Daniela Soares, Product/Planning and Logistics Director, Salvador Caetano Auto, Kaseya

Mystery shopping is a method companies use to gather specific information about a market or competitor by replicating typical consumer behaviors. In African car retail, where public databases are scarce, this method is commonly employed for its effectiveness in obtaining high-quality market data. Salvador Caetano África has been using mystery shopping for many years and chose it as the first piece to automate in this project.

With information providers identified, our team developed a Microsoft Outlook add-in that leverages OpenAI and Power Automate to automate the process of writing and sending email messages requesting quotes. The add-in also extracts data from the replies, including any attachments. These attachments are stored in Azure Blob Storage, and the data is recorded in a CSV file. Power BI then generates a dashboard from this CSV file, featuring key performance indicators (KPIs).

After some adjustments, the system became more intelligent and is now capable of distinguishing between final price, price before taxes, discounts, taxes, and other pricing components. All of these details are accurately categorized in the CSV file.

This solution enabled Salvador Caetano África to expedite their mystery shopping process and reduce the overall hours dedicated to market research. However, there are still some limitations the company hopes to address in future iterations of the tool. Since this is an Outlook add-in, it must be connected to a single email address, which eventually becomes flagged as a competitor and obsolete. Therefore, our team has been working on methods to regenerate and increase the email addresses from which the tool operates. Additionally, the model currently struggles with different nomenclature, often treating the same vehicle model as different ones when sellers use varying names across the email body and attachments.

Although currently used only in English-speaking countries, Salvador Caetano plans to extend the tool’s use to French-speaking and Portuguese-speaking markets in Africa. This is not expected to pose a significant issue for the model. Additionally, the overall quality of the generated text is expected to improve with the transition from GPT-3.5 to GPT-4.

Although these updates are important, they are considered minor improvements to the overall project. The next significant advancement will involve introducing further automation methods for market research, aiming to substantially reduce the time the Salvador Caetano África team dedicates to this task.

“DevScope helped us a lot to guide the solution and one of the added values was being able to clearly understand how the system would work.”

– Nuno Faro, Business Development & Operations Support, Salvador Caetano Auto

Microsoft PowerAutomate

Microsoft Outlook

Azure Functions

Azure OpenAI

Azure Storage Account

Microsoft Power BI