top of page
Click.AI: Clicksign’s Artificial Intelligence

Building the AI for the “Automation” Product at a Digital Signature Company.

Overview

Within Clicksign’s product portfolio, there is a product called Automation, which allows users to automate the generation and sending of contracts and documents at scale.

The product had issues that directly impacted users—for example, a lack of clarity and understanding about Automation itself, as well as how it differed from the automations users were already performing through Clicksign’s traditional setup.We decided to focus on understanding what users truly perceived as problems within the product.

From the start, it was clear that this project would require significant updates, investment of time, and the introduction of more technology and process automation.

We began by conducting studies with both internal and external stakeholders. We created scripts to guide usability tests and map the user journey, developed a test model for direct interaction with the site to simulate a real environment, and, after conversations, interviews, and testing, we arrived at a solution leveraging artificial intelligence.Below is a step-by-step overview of how this product was built!

Principal Problem

Solution

Outdated product, old, without significant updates.Users of the Automation product:Did not understand the difference between automating existing processes and the Automation product itself.When they encountered a specific problem with the product, there was no direct channel to resolve these issues; the only route was through Customer Service, along with all other support requests.Faced difficulties attaching documents, sending files via WhatsApp, and other tasks.The document submission process was slow and very lengthy.

A dedicated chatbot entirely focused on the product, where the document submission process became faster, more technological, modern, and efficient, eliminating the delays in submissions.In addition to the changes in support, which are now directed toward exclusive assistance for the product’s clients.Regarding the lack of understanding of what the product offers, and the confusion between automation and automatization, we implemented strong communication through marketing, targeted product training, and videos with clear explanations for each step of the new process.

How did we arrive at the solution to these problems?

Before creating the product and determining whether an AI-generated chatbot would truly be the solution, we opted to conduct usability tests and interviews with both the product users and the stakeholders who were working on the front lines of this project internally at Clicksign.

This way, we were able to understand: What are the users’ real needs with the product? What features would they like the product to have that would help them in their daily tasks? Are the problems we observe internally also visible to these users? If so, which aspects of the product do they consider daily pain points? Is an AI-powered product with a chatbot truly the best solution?

image.png

Interview with the product's clients

We created a structured script with questions to understand the value and validate the flow (medium-fidelity wireframe). But first, we wanted to learn more about who these clients were—their habits, understanding of technology, current role, among other aspects.After understanding the client, we began with questions focused on the Automation product and their experience with Clicksign.

Internal usability test with the product stakeholders

We initially conducted a test with the people involved internally in the project, using the medium-fidelity wireframe, in order to understand the main gaps and questions these individuals had before making the test public to clients. In total, ten people completed the test.This test was conducted remotely and unmoderated. We sent each participant a link to access the test, and at the end, we provided some follow-up questions.In the link below, you can access the questions, detailed results, design team recommendations, and the main insights generated.

Key points of attention and insights from the internal test

  • Sometimes the system asks the user to “type,” and other times to click buttons; it would be useful to have a consistent standard.

  • Indicate to the user when they can end the conversation with the bot.

  • Some users did not understand certain texts, for example, the difference between “add a new signer” and “add the signer to the flow.”Users requested clearer instructions.

  • Provide prior explanation of how documents should be formatted.

  • Provide prior explanation of what to expect in the next steps.

  • Messages should clearly indicate what the next step in the flow is.

  • Avoid requiring two documents, and if there is an error related to submission or the documents, the error message should be more objective and clear.

  • Indicate to the user that information has been saved when clicking exit.

60% of testers reported having difficulty finding the icon.

40% of testers had difficulty creating the template from the documents.

70% of people found the tasks complex.

Usability Test with the clients

image.png

After the usability test with the people involved in the project, we were able to identify some conflict points and significant errors, which we were able to correct before conducting the test with clients.The goal of this test was to understand areas for improvement, technical feasibility, feedback, and insights.We had several user profiles and chose to conduct the test with all of them.We spoke with a total of six people, and the test was conducted remotely and fully moderated by me (UX Researcher on the team).

User Profile 1

 

Client using the Automation product.

User Profile 2

 

Client who has never used the Automation product.

User Profile 3

 

Client who has never used the Automation product but is familiar with it.

We started with a quantitative research

 

We divided the questions into two sections:

Mapping the use of Automation and Mapping AI in the use of Automation.

 

We asked the following questions:

 

  • Do you use any other part of Clicksign besides Automation?

  • If yes, how frequently?

  • How often do you use the Automation area of Clicksign?

  • Which features do you currently use in Automation?

  • Does Automation help you with your tasks? How?

  • What do you think could be added to help you even more?

  • Do you use Artificial Intelligence in your daily activities?

  • Do you believe AI could help you in your activities with Clicksign? How? At which stage?

  • If I told you that we are considering implementing Artificial Intelligence in Automation, would you find it useful?

  • What would you like it to do?

Usability test execution

We divided the test into four distinct phases:

 

First Phase:

Find the icon to start creating the automation through ClickAssistant.

Read the suggestions and tell me what you understand you can do and what will happen.

 

Second Phase:

Upload two PDF documents with the same text structure.

Create a template from the documents you uploaded.Identify the variables in the created template.

Make some edits to the created template.

 

Third Phase:

What do you understand at the moment of adding a signer?

Which options are available?Add a signer to the flow.

Remove the signer.

 

Fourth Phase:

What do you understand happened, and what can be done from this point on?

Upon completing the flow, copy the link to the form.

Finalize the creation of the flow.

Pain Point 1

 

The user uploaded documents that were different from what was requested, which triggered an error message. However, the message did not explain what the error was or what the next steps should be.The user asked the assistant for guidance, but it was unable to provide an answer to help them continue the flow.

Pain Point 2

 

We understand the high level of frustration experienced by the client when trying to interact with the AI, and it does not respond or allow them to go back to the previous step.

Pain Point 3

 

When interacting with the chat, the user feels the need for buttons that can guide her with the correct terms/words, and she constantly resorts to “copy and paste” text.

Key points of attention and insights from the client test

 

Clearly, basic usability and interaction points with the chat experienced issues, leading to the conclusion that the product is still not mature enough for production. We have opportunities for improvement in:

 

  1. Copywriting.

  2. Provide a more efficient explanation of the AI’s purpose (considering that the persona using it may or may not have knowledge of Automation).

  3. Allow easy access to FAQs and/or provide answers through a conversational chat.

  4. Explain the fundamental requirements for creating a good document.

  5. Offer predictability of what can be generated after interacting with the AI.

 

Correct usability issues, such as:

 

  1. Provide clear instructions for document upload.

  2. Review the rule of allowing only 2 documents to be uploaded.

  3. Offer help and documentation in error scenarios.

  4. Provide routes for returning to previous steps and completing stages of the journey.

  5. Enhance conversational freedom with the AI.

  6. Introduce a standardized action pattern using buttons.

Final Product

image.png

Greater efficiency with automatic title and message generation

 

When documents are sent for signing, Click.AI automatically generates titles and personalized messages based on the document’s content.

image.png

Document summary for better understanding

 

Click.AI directly enhances the signing experience by summarizing the key points of the contract for your client, making the content easier to understand.

image.png

Documents signed more quickly

 

With features focused on better document comprehension, Click.AI increases confidence in the signing process and accelerates contract execution.

image.png

A more transparent and seamless signing experience

 

Click.AI enables a more optimized signing experience, allowing companies and their clients to gain productivity and speed when closing contracts.

Learnings

 

  • Fewer steps, fewer clicks, and more autonomy for clients.

  • Increased client confidence in the product.

  • Greater flexibility in the process and its stages.

Next steps

 

  • Keep investing in communication and training.

  • Keep identifying features that will enhance the product.

  • Keep testing with every new update.

Business Impact

 

The project delivered significant improvements, such as:

 

  • Reduction in complaints and/or messages to Customer Service (SAC).

  • Greater product adoption and understanding. Some clients had the plan with the product but did not know how to use it or found it unintuitive compared to the old manual process.

  • Increased client engagement with the product following new marketing communications and interactions with the User

  • Experience team.

53%

 

Fewer messages reporting issues to Customer Service (SAC) about the Automation product

32%

 

Product adoption by clients who already had the plan

45%

 

Improved engagement with the product following communications, promotion, and training.

bottom of page