How to use CAPTCHA in a conversational interface?

In today's digital world, robust authentication mechanisms are an absolute necessity. With the increase in cybercrime, it is essential to protect personal and sensitive information. As stated in the last OWASP 2023 Top 10 Vulnerabilities report, authentication mechanisms are the first line of defence against unauthorized access to online accounts. Furthermore, the new ISO 27001:2022 certification standard dedicates specific controls and clauses to ensure secure authentication procedures.

A robust authentication mechanism is the key to unlocking a great number of self-services. For example, making financial operations, changing insurance policy details or consulting medical test results are operations that individuals can perform online if they pass a strong authentication screening.

However, there are situations where robust authentication mechanisms are not possible. This is the case of quote&buy journeys, where customers are not registered and identified apriori, but also when agents and advisers are about to request some actions on behalf of their clients.

In such cases, the list of self-serve services experiments a physiological reduction because some critical actions won’t be accessible to an unidentified user. Reducing the risk of robots and Denial of Services attacks is also important by using additional security measures like CAPTCHA codes and other techniques.

CAPTCHAs, in particular, are a popular security measure used to prevent automated attacks by requiring users to prove they are human and they nicely fit conversational interfaces. A handy-style text over a noisy background is generated and displayed to the user as an image. Automated Optical Character Recognition (OCR) detectors won’t be able to easily guess the keyword by reading the image.

The main three requirements for a good CAPTCHA code generator are:

All Posts

The Spixii effect: designing products around customer problems

4 min read

According to Gartner, 85% of future consumer interactions will be supported without human application by 2020. How can you make sure your chatbot thrives? Here's a deeper dive into Spixii's customer-centric approach from UX designer, Matteo Schgor.

 

How to apply agile frameworks and customer-centricity to your chatbot

 Agile frameworks such as Scrum or Kanban, as well as Eric Ries' lean startup methodology, are designed to help you shorten product development cycles and develop around customer feedback. Yet, when you apply them to designing chatbot conversation, this combined approach will help you look at your digital customer journey from fresh eyes.

 

"Iterate, iterate, iterate. By iterating we validate

our ideas along the way because we're hearing

from the people we're actually designing for"

- Gary Brink, Founder of Tomorrow Partners 

1. Understand the challenge

A common human tendency is to dive into building a solution before analysing the situation. Before you start considering solutions and technologies, analyse what problems your business is facing right now. 

For instance, let's take the story of Minnie* (names changed).

When the Beast from the East hit the UK earlier this year, many insurance companies experienced a significant increase in claims: customers reporting damages to their roof and gutter from ice and snow, collapsed ceilings, burst water pipes and water damage.

 Minnie returned from a 10-day holiday to find six inches of water in the house. She tried calling an emergency assistance number but could not get through for three and a half hours. The longer she waited, the more stressed she became and the more damage was caused to her home.

So, let's look at the problems.

Minnie was unable to reach customer service at her time of need and had to spend hours on hold, wasting her time and making the situation even more stressful. The company was unable to scale its call centres when facing a high number of calls, and the resulting poor customer service caused low customer loyalty and high customer acquisition costs.

Screen Shot 2018-04-12 at 15.33.55

What are the why's, how's and what's facing your company?

 

Your company and customers are unique. Investing time understanding them early on  will help align your business strategy with the needs of your customers.

The resulting solution? Your technology will be more than just a nice-to-have or nod to digital transformation. It will be designed with a clear purpose. The moment you find this is when you find your power.

 

2. Formulate your hypotheses

We all bring ideas and doubts to the table when introducing something new.

Once you have defined the challenges, step back and gather a list of concrete hypotheses to test. A testable hypothesis must be as specific and detailed as possible. For instance, you could say that 'people aged between 24-35 living in London will need to use a chatbot because they have no time to wait on the switchboard'.

prototyping-min

 

3. Understand your customers

Formulating hypotheses goes hand-in-hand with research.

  • Who are you designing for?
  • What are their needs?
  • What value can you bring them?

You might discover that the customer you focused on is not your actual target audience, or that their problems are different to what you expected. Once you have a clear idea of who your customer is and what their problem is, you are ready to move into prototyping. 

 

4. Validate your hypotheses

After building the first draft of your conversation, the next step is to test it using the build, measure and learn loop: the foundation of the lean startup methodology.

Screenshot 2019-03-26 at 13.34.53

In small groups, invite people to test your conversation on multiple devices.

For instance, on a recent client project, we invited 20 people to try the conversation out in small groups of 3-4 people. After each testing session, gather the feedback and edit your conversation accordingly. This great guide explores this in-depth.

 

5. Continue iterating

The result? Once you have the first iteration of a conversation, you can follow this loop to continuously improve your digital experience. Once you have some data, this is where you can start applying machine learning and artificial intelligence to get behavioural customer insight.

Ultimately, it takes time to design great chatbot conversations. But more than technology or experience, the most important feature lies in a very human value: empathy.

Today, many insurers are struggling to emotionally connect with their customers. This approach to design may help you reach the next level of customer-centricity: bringing your business into a new era of personalisation.

Recent Posts

How to use CAPTCHA in a conversational interface?

In today's digital world, robust authentication mechanisms are an absolute necessity. With the increase in cybercrime, it is essential to protect pers...

Read more

The sweet spot between smooth operations and customer experience

3 min read Gone are the days of interactions where customers and businesses would have to come face-to-face for a transaction. Thanks to the mighty po...

Read more

How can customer service operations save money?

5 min read Efficient and effective customer service operations can significantly impact a company's finances. By providing customers with prompt and a...

Read more

What are the 3 steps to reduce the stress experienced by customer service consultants?

5 min read As a customer service consultant, many factors can cause stress and anxiety in the workplace. One of the biggest causes of stress is dealin...

Read more
Spixii White Paper

Download your FREE Spixii White Paper copy

Discover how to execute high-value processes sustainably. Included in the White Paper:
  • Expanding propositions with digital
  • Conversational Process Automation (CPA) deep dive
  • De-risking the transition to automation