Introduction

This project represents an internal development initiative designed to address a specific challenge in the accounting market. In this case study, we explore both the technical aspects and the business context that motivated the creation of this MVP.

Case Study Overview

Our target client is an accounting firm that frequently needs to convert bank statements, often dozens of pages long, from PDF to Excel format. These bank statements are provided by clients in PDF format, either by choice or due to limitations of banks that don't support Excel exports.

This isn't an isolated problem: many accounting firms face the same challenge. In Italy alone, there are over 10,000 monthly searches for PDF to Excel conversion tools. Further analysis using ChatGPT identified bank statement conversion as the primary need behind these searches.

To address this demand, I developed an MVP for a SaaS solution dedicated to this type of conversion, initiating a validation process with a beta-testing accounting firm. The MVP follows a freemium model with credit-based payments.

Over the next 3-4 months, I'll gather feedback to validate the idea.

The Initial Problem

Limitations of Traditional Conversion Tools

Free online conversion tools, like iLovePDF, often produce suboptimal results.

Here's an example of conversion using iLovePDF:

Instead of producing a single transaction table with 5 columns (Date, Description, Credit, Debit, Balance), the generated Excel file contains 8 separate tables, requiring extensive manual data cleanup.

Even with the paid OCR mode, while results improve, similar issues persist.

Limitations of Generative AI (ChatGPT, Claude)

Generative AI solutions like ChatGPT and Claude also have limitations for this use case:

• They require specific prompts, which aren't always intuitive for non-expert users
• They limit the length of processable files
• Cannot process multiple files in parallel
• Don't always correctly interpret every PDF

I tested the sample file with both platforms (ChatGPT and Claude) without satisfactory results:

It's important to specify that the versions of Claude and ChatGPT used for this test were the free versions. The paid versions, which both cost $20, can analyze most types of PDF files. However, some of the limitations mentioned above still remain (file length restrictions, prompt writing requirements, inability to process files in parallel).

Project Objective

The project aimed to develop an MVP for a WebApp capable of converting PDF bank statements to Excel with the following key features:

• Accurate conversion without the need for manual post-processing
• Ability to analyze any type of PDF file, including scans
• No prompt writing required for use
• No file length limitations and ability to process multiple files in a single request

These features were specifically designed to meet accountants' needs and reduce time spent on data preparation, allowing them more time for higher-value activities.

Development Process

1. User Interface Design

I designed the interface following the "less is more" principle, placing the file upload feature directly on the homepage to immediately communicate the service's value.

2. User Flow

Unregistered User:

1. Uploads file from homepage
2. Initiates conversion
3. Receives an informative banner about the free version and premium option

Registered User:

1. Additional multiple file upload functionality
2. Access to unlimited conversions

Authentication System

Registration is possible through two methods:

• Email and password
• Google Login

Payments

The SaaS is a freemium model with credit-based payments. The user can convert a PDF file of up to 5 pages for free, while for longer conversions, payment is required. Each credit can be used to convert one page of a PDF file.

Beta Tester Collaboration Approach

I provided the ready-to-test MVP to our first client, who reports any areas for improvement during their usage.

Conclusions

This case study demonstrates that it's possible to develop a functioning MVP in just a few weeks. Not just to validate an idea, but also to make it usable and facilitate user acquisition.

If you're interested in developing an MVP, book a free consultation by clicking this link: https://cal.com/luca-di-domenico-mvpfast/mvpfast-mvp-development-request-information