Utilizing No-code and OpenAI's Whisper and ChatGPT for Sales: My experience

Hello everyone!?

Today, I will share how I use No-code + AI to manage my sales team. These methods are relevant to businesses of all sizes.

Background

As previously mentioned, I am involved in the development of a small manufacturing company that specializes in parking lot equipment and storage systems.

I have previously told some stories about implementing a calculator in Excel, hyperlocal targeting in paid ADS, and many other things that made our lives a little easier.?Today I would like to talk about sales.

I have built the sales process from scratch and have been responsible for sales control for some time.

The First Phase: Listening to Call Recordings day by day

Tools: Planfix CRM, notepad, pen, and Google Spreadsheets.

I started to listen to calls with just a pen and paper, but this method was time-consuming and lacked regularity. I fixed some issues, but new ones kept appearing.

Eventually, I increased the playback speed to x2 and started entering data into a Google Spreadsheet. The task still demanded tedious and systematic work, which I sometimes lacked the time and resources for.

We could only meet with the sales team once a week at best, which was infrequent.

By the way, i’ve got some results and made a plan.

The Second Phase: making processes?

Tools: Planfix CRM, Make, Slack, Telegram, GoogleSheets, Looker, Google BigQuery

I have noticed a few issues that keep happening repeatedly:

- Not sticking to the scripts while communicating with clients.

- Not recording crucial information from client interactions.

- Poor identification of client needs and inability to handle objections.

- Not discussing the next steps with the client.

- Inadequate prioritization leading to the loss of important clients.

- Not fulfilling promises and losing production/project details.

Here is a list of the areas that I have been consistently working on:

1. Setting up a CRM system. I consolidated all the processes into a single system, created useful screens and dashboards for different scenarios, and automated everything possible. It was a long journey. It has not stopped yet.

2. I continued to listen to calls independently and successfully implemented sales scripts, which were finally integrated into the CRM.

3. I established a motivation system for the sales team, which includes a grading plan and a coefficient system.

4. Monitoring CRM system. I made some alerts at key stages of the deal: no one has called back for a new request within a few hours, or when a call has been missed.

I did not have enough time to constantly control the process, but I identified all breakpoints. I realized that I needed to reevaluate my approach, but I was struggling to come up with new ideas.

The Third Phase: OpenAI

Tools: Planfix CRM, Make, Slack, Telegram, GoogleSheets, Looker, Google BigQuery, OpenAI (Whisper / ChatGPT)

In May, I commenced an investigation into the capabilities of OpenAI and discovered its potential for various tasks in my company.

Following a month of testing, I determined the optimal parameters and developed a prototype system that functions with our PBX system. This system transcribes text and extracts vital information.

I transcribed the text utilizing Whisper API and saved it in the CRM.

Subsequently, I utilized ChatGPT to analyze the dialog. This was the first version. Despite its elementary form, this approach proved to be useful.?

Within 10 seconds of the call concluding, the sales team was granted access to the conversation's text and a summary. This approach obviates the necessity to remember the conversation's details or re-listen to the call.?

This was a significant victory.

Afterward, I integrated Slack into the process. I ceased checking the CRM to remain informed of the most recent developments.?

All summaries from conversations with the sales team were automatically posted to the relevant chats.

Results: The response time to problems decreased, and the manager was receiving feedback and assistance from me almost instantaneously.

The Final Stage: Going Deeper

Tools: Planfix CRM, Make, Slack, Telegram, GoogleSheets, Looker, Google BigQuery, OpenAI (Whisper / ChatGPT)

The subsequent stages of the sales funnel necessitate a more thorough approach. In June, I took a deep dive by fine-tuning ChatGPT, resulting in a customized version of gpt-3.5 tailored to my sales scripts and products.?

In addition, I began to consider the context in which information was conveyed to clients at various points in the funnel, recognizing the importance of appropriate handling, qualification, and addressing inquiries.

Two steps were employed in this process.?

The first step involved defining the context and situation, while the second involved selecting an appropriate GPT model for the given situation, either a fine-tuned version or separate instructions.?

The resulting analysis provided the requisite outcome, which was then organized into fields in the CRM or transmitted to the corporate messenger or telegram bot.

The latest iteration of this approach involves the extraction of structured information, such as the width, height, depth, and number of shelves, in response to qualification questions.?

It also identifies key parameters such as timelines and budgets, as well as objections and needs.?

Tracking agreements on the next steps, highlighting complaints and negativity, and ensuring proper CRM completion are equally essential components of this approach.?

Finally, recommendations for the next steps are provided, taking into account the context and dialogue.

The utilization of a comprehensive set of tools has led to a significant reduction in the reject rate to 3% and an exponential increase in the speed and quality of customer processing.

These results demonstrate the vast potential of such tools, which can be harnessed by businesses and individuals through imaginative application and availability of resources.