How to build a customer scoring tool with AI
Tony L
@tonilopezmr
Identifying which leads have the greatest potential to convert is key to keeping sales productivity high and staying on track with revenue goals. This is especially important considering that only 21% of leads will convert to sales, according to Gleanster Research. You can’t waste time on targets that aren’t a good match for your offering. It’s essential to have a lead scoring system to help you identify the best matches for your products and services.
In this step-by-step tutorial, you’ll learn how Stack AI can help you build a customer scoring tool. It will research the target company automatically, process that information with AI, and match the results to your offering. Using this data, you can identify the best fits and focus on closing them.
How to build a customer scoring tool
- Create a new Stack AI project
- Connecting inputs, search, LLMs, and outputs
- Score hundreds of companies with one click
- Try other interfaces
- Share it with your team
- Keep track of usage and analytics
- Improve your customer scoring tool
Create a new Stack AI project
Click here to sign up for a free Stack AI account if you haven’t already. After logging in, click the New project button at the top right of the dashboard.
As you can see, there are many templates for multiple use cases. We’re starting from scratch this time. Click New project to continue.
Once created, you’ll see the project’s canvas with input and output nodes.
Connecting inputs, search, LLMs, and outputs
Set up a Google Search node
The tool’s first action is running a Google Search to find more information about your prospect. Let’s set it up in Stack AI: on the left side of the screen, expand the Data Loaders section.
This section contains nodes that search for or hold data, ready to pass it on to an LLM to improve their reasoning abilities. Click and drag Google Search onto the canvas, to the right side of the input.
You’ll notice two settings inside the Google Search node card:
- The Number of Results dropdown selects how many search results to return into Stack AI. Increasing this value will increase processing time.
- Search Country lets you adjust the location of the search results to match your company’s operating area or target market.
Click the Settings icon at the top right of the Google Search node to reveal more controls. Most nodes in Stack AI offer this functionality, so be sure to look out for this icon as you gain experience with the platform.
You can turn on Advanced Data Extraction to retrieve data from elements like charts and tables. And while Stack AI already uses OCR by default, toggling the Text in Images setting activates a more advanced text extraction algorithm.
Connect the input node to the Google Search node to let the user pass the search term and start the search.
Add a research summarizer LLM
We need to summarize the search results with AI to improve readability and help match it with your company’s products or services later. On the left side of the screen, expand the LLMs section.
Stack AI integrates with all leading AI model providers, giving you access to the latest technology from OpenAI, Meta, Google, and others. For this tutorial, we’ll be using Anthropic’s Claude as the research summarizer: drag and drop the Anthropic node onto the canvas.
The LLM needs to receive the user input and the Google Search results. Drag and connect the user input and connect it to the Anthropic LLM. Remember to do the same for the Google Search node.
Add system instructions to the Anthropic LLM node so it understands what to do with the data. You can copy and paste the example below into the corresponding input field. Make sure the formatting, numbering, and spacing are consistent to maintain reasoning quality.
1. Company profile
- What is the sector?
- What are the core products or services they provide?
- Who is the primary customer base?
- What is their estimated scale in terms of revenue and workforce?
- Where is their main office located, and in which regions do they have a presence?
2. Operational Structure and Business Strategy
- How do they produce or deliver their goods and services?
- What are their primary methods of distribution?
- Are there any significant partners or suppliers they rely on?
- Any notable processes or technologies they utilize?
3. Potential Challenges and Needs
- What are the key issues they encounter that our offerings could solve?
- Where might they need to enhance efficiency, cut costs, or stimulate growth?
- Are there any regulatory or market challenges that make our solutions particularly beneficial?
Please distill the most important insights from your research into a succinct company profile, focusing on the key details to assessing them as a potential client for our business.
Next, we’ll structure the user prompt.
Copy and paste the following into the corresponding input field in the Anthropic node:
Consider these Google Search results about this company: {serpapi-0}
Analyze this company: {in-0}
The values in curly brackets are variables:
{serpapi-0}will contain all the data returned from the Google Search node.{in-0}contains the company name that the user typed into Stack AI.
When working on more complex projects, you can drop multiple nodes onto the canvas and add their variables to LLM prompts. This lets you provide more context to the AI engine, which can improve reasoning quality.
If you’re having unintended results with these prompts—or are editing them with your unique instructions—consider adding XML tags to separate each guideline. An example: <formatting> Format your results with bullet points in bold </formatting>
We want to display the summarized research to the user, so connect the Anthropic LLM node to the output node already present on the canvas.
Add a scoring and analysis LLM
This could be the end of the tutorial if we were only building a company research tool. But we’re going further than that: let’s add an LLM that will analyze the research data and match it to what your company offers. On the LLMs section, find the OpenAI node, drag and drop it onto the canvas.
As you can see, you can use multiple AI models from different providers in the same project, so you can always choose the best model for each task.
But before we configure the new LLM node, we will also add an automated way to analyze the products and services that your company provides. This data will be used to compare and score with the target company. On the left side of the screen, expand the Knowledge Bases section.
Drag the URL node onto the canvas, close to the two LLM nodes.
The URL node lets you gather information from your website’s web pages. Access your company’s product and service offering pages, copy their links, and add them one by one into the node.
You can adjust the scraping frequency to daily or weekly with the Re-scrape setting. This will make sure that the data is up to date at all times.
Next, connect the main input to the URL node.
Then, connect the output of the URL node to the input of the scoring LLM node.
Since this LLM is in charge of matching the target company with your offering, it needs a different set of system instructions. Copy and paste the following into the corresponding OpenAI node input field, or adapt them to match your company data:
Analyze the company profile and determine if they match the target audience for our business.
Consider how their needs and operations align with the various products and capabilities we offer, which include: {urlemb-0}
Highlight key areas where our products could provide value to the company based on their profile. Assess how much there’s a match between our products and the company profile. Assign a score from 1 to 10 based on the strength of the match.
The {urlemb-0} variable stands in for the data scraped with the URL node.
We need to add the previous LLM’s output as the input of this node. Drag the output connector from the Anthropic LLM into the input connector of the OpenAI LLM.
Edit the prompt in the OpenAI LLM node to reference the Anthropic LLM’s output. You can copy and paste the text below into the Prompt input field:
Analyze this data: {llm-0}
Again, the value in curly brackets is a variable. {llm-0} stands for the data generated by the Anthropic LLM.
The user needs to see the scoring results. On the left side of the screen, expand the Outputs section.
Drag and drop the Output node to the right of the OpenAI LLM. As you can see, you can add multiple outputs to show content to the tool’s users or diagnose functionality when building more advanced tools with dozens of nodes.
Finally, connect the OpenAI LLM to the output to finish setting up the workflow.
Score hundreds of companies with one click
Now that you’ve taken care of the tool’s functionality, it’s time to set up the user interface. We’ll configure the batch interface to run this logic for multiple companies. This way, all you have to do is upload a list of companies, click Run Batch and wait for the full results without any other manual work. Once ready, you’ll get everything in another list that you’ll be able to export as a CSV or Excel file.
First, save all the workflow changes: click the Publish button at the top right of the screen.
Important: whenever you make changes to your project, remember to click the Publish button to save them.
Tool deployed. To start editing the user interface, click the Export tab at the top left of the screen.
Click the dropdown displaying the Form interface and change it to Batch.
Important: to update the preview window on the right side of the screen, click the Save Interface button at the top right.
You can add a custom domain to make the link easier to read, type, and share. Below these controls, add a tool name and description to help your team understand what it does and how to use it.
To improve the user experience, you can label the inputs and outputs as well. For this tutorial, you can change:
- in-0 to Company Name
- out-1 to Match Score
- out-0 to Company Research. Tick the checkbox to show it on the interface.
At the very end of the side tab, you’ll find the Security settings. Stack AI has high security and privacy standards, offering controls to keep your interfaces safe:
- Enable password protection for a simple, direct way of securing the interface.
- For more granular control, you can force SSO and select who can use this tool: password bearers, organization members, or specific email addresses.
This way, external users won’t have access to this tool even if the link is leaked and made public.
Try other interfaces
Stack AI has other interfaces that may be a better match for other use cases:
- Form turns this tool into a single-use interface.
- WhatsApp/SMS enables access via these two channels.
- Slack App connects the tool to your internal communication platform.
- API integrates this AI-powered workflow with your ERP, internal tool, or CRM.
Share it with your team
Sharing the interface with your team is as easy as copying the link at the top of the preview window and pasting it in an email body, internal channel, or in a text message.
If you want to share your Stack AI project with other tool builders in your team, you can do so by clicking the Share button at the top right of the screen. This will share a copy of your workflows: other people can’t change your work.
Keep track of usage and analytics
Take a look at the performance of your customer scoring tool by clicking on the Analytics tab.
Here you’ll find an overview of total runs, users, errors, and tokens consumed.
Under these top-level stats, you can see a list of recent runs. Change which columns display on the analysis by activating or deactivating them using the Columns dropdown.
Improve your customer scoring tool
Use the insights of the Analytics tab in Stack AI as well as your team’s feedback to adjust AI prompts, add new nodes, or change the connections. As you do so, stay up to date with the latest AI technology: you can change the providers and AI models with a few clicks. First, click the Settings icon at the top right of each LLM node.
On the side tab on the right, open the Provider and Model dropdowns to select the model you’d like to use. Remember to hit Publish when you’re done and run a quick test to ensure everything is working as intended.
Wrapping up
It’s easy to be bogged down by too much research work, and as a result, take the shortcut and focus on leads that aren’t ready or willing to buy yet. But with this AI-powered customer scoring tool, you can quickly assess which companies are most likely to want to do business with you, as well as exploring the approach angles your sales team can take.
But there’s a lot more you can automate with Stack AI. Create a free Stack AI account and explore the AI tool tutorials: