I recently got to work on an exciting prototype for a 3 day hackathon at Spendflo where we built a quick prototype for a problem statement that SDRs face on a weekly basis.

Problem statement

SDRS typically spend 15-20 hours per work week in researching potential leads, gathering company profiles, financial reports, market insights and so on. Sales teams struggle to identify ideal customer profiles (ICPs) quickly, leading to inefficiencies in lead generation.

Salesforce's State of Sales report 2022, explains that 66% of sales reps feel bogged down by data entry and non-sales activities, impacting their overall productivity.

InsideSales 2018 Time Management Study reports that sales teams spend only 37% of their time selling with the remaining hours focussed on non-selling tasks.

Revolutionizing sales efficiency

Lyra aims to address this by automating the company profile search process and centralizing all sales-related information in one hub - thus reducing the time spent on manual tasks, and enhancing overall sales efficiency.

User research

We first started by understanding how sales teams currently find ICPs and understand the process steps that can be optimized.

Key insights -

• Sales reps often use multiple tools like LinkedIn, Crunchbase, Slintel, Company annual reports to differentiate between ICPs and non-ICPs.

• Exporting data into CRM systems is a common, time-consuming task.

• Users prioritize ease of use, relevant matches, and direct integration with their existing workflows (like Salesforce).

This was a sample data that we generated using ChatGPT and manually verified by one of our SDRs.

Prompt -

I need a table of companies who meet the following requirements:

1. the companies must be from: India, US, Canada, Singapore

2. the companies must use atleast 80+ technology tools

3. the companies must have employee count of 100+

4. the companies must have atleast series a funding and must have the potential to go till series d or above funding.

Jobs to be done

We went ahead with writing down the key requirements for achieving this goal.

Key requirements -

1. Users must search for companies by name or domain

2. A clear indication of how closely companies align with the user's ICP (with match percentage)

3. Export matching companies directly to Salesforce

Wireframing

User Flow:

1. 1. Users input a company name or domain into the search bar.

2. 2. The system returns top companies that fit the ICP, each with a match percentage.

3. 3. Users can either view more details or directly export these companies to Salesforce.

Implementation

We initially started with web scraping as direct access to public databases like LinkedIn and Crunchbase is restricted.

Web scraping -

We started with custom scraping solutions on publicly available data by using Scrapy and BeautifulSoup libraries.

Company websites - Many companies publish key financial metrics, leadership information, annual reports and business summaries on their official websites.

Annual reports - For publicly listed companies, we scraped data from SEC filings, annual reports and other regulatory documents that are publicly available. This helped us in extracting information like revenue, employee count and recent business performance.

LinkedIn and Crunchbase workaround - Manual research was done by sales reps to gather more precise ICP data.

Web scraping came with its own set of limitations where a person was required to manually clean the data and then classify them as test data to feed it to a machine learning model.

ChatGPT integration -

1. PDF readers were used to summarize complex reports and provide insights from unstructured data. We were able to carry out tasks like summarizing annual reports, providing company overviews, and highlighting recent company developments.

2. Custom prompts were carefully designed to ensure that we get relevant data for ICP matching.

3. Collected data was then stored in a MongoDB database of the collected information.

Export to Salesforce -

After collecting and summarizing the data, system was able to let sales reps to export selected ICP profiles to their Salesforce.