Fashion

Tiger of Sweden

A side profile of a woman in a yellow flannel shirt

Tiger of Sweden Chatbot Solution

  • Email and phone inbounds decreased by 50%, while inbounds via chat increased by 400%.
  • First-reply time for customer support agents dropped from 17 hours to just 10 hours.
  • 35% of Tiger of Sweden’s customer inquiries are handled by their chatbot.
  • Support Customer Satisfaction score has increased in just one year from 73% to 96%.
  • The bot uses NLU to interpret questions from customers and provide the best answers.
  • The bot was built by Tiger of Sweden’s Customer Care Manager, with no IT background.
  • During a one-day, hands-on workshop, the agent learned how to build a Tiger of Sweden chatbot on her own.
  • The Bot Operator uses 1-2 hours per week on checking and updating the bot.

Who is Tiger of Sweden?

The Swedish high-end fashion brand Tiger of Sweden, founded in 1903, has become a great example of how traditional retail is embracing the digital era by expanding their services as an online, digital business.

Tiger of Sweden has several flagship stores in 8 countries and sells online to more than 30 countries worldwide. Tiger of Sweden represents a different and opposing voice to fast fashion.

Historically, the fashion house has always had a focus on crafting high-quality garments while standing for fair working conditions and ethical fashion production processes.

A screenshot of Tiger-of-Sweden's webpage with a chat tab to the right showing a conversation between the Tiger of Sweden chatbot and a customer
Tiger of Sweden’s chatbot messages.

Tiger of Sweden’s path to innovation

Catering to stylish and conscious buyers, the brand has experienced long-term growth in popularity. In their path to providing more purchase options and services online, Tiger of Sweden’s customer inquiries have been rising consistently year after year.

Questions from their customers were often very similar and responding to each individual inquiry would take up a large portion of the employees’ time in customer service, sometimes lengthening waiting time.

Innovation was needed to improve their customer support, and it would need to be fast to implement, easy to maintain, and well-integrated with their pre-existing business systems.

From problem to chatbot solution

The idea was there: to provide fast, automated, and accurate answers to all similar questions that customer support was receiving. But how?

The brand started considering using a chatbot to handle all first-line support.

The chatbot could be the initial gateway that would either carry out the whole conversation and resolve the ticket or pass the request on to an employee in case it was more complicated.

Tiger of Sweden chose Certainly because the platform was the best fit: it facilitated and allowed all the integrations they needed, and it was easy to use for employees outside the IT department.

The Certainly Platform, while being built on state-of-the-art technology, is easy to use for non-coders.

“Across companies and across industries, we find that development teams are extremely busy. One of the priorities for us at Certainly, was to build an advanced platform that developers would love, but businesspeople too could easily use, without needing to use developers’ time,” Henrik Fabrin, CEO of Certainly.

A woman and a man sitting at the office desk opposite each other and discussing
Nadin Kempel Sigh and CTO Dario Rodriguez discussing the chatbot.

But I can’t code!

“I remember my first reaction when my boss told me about the project was: Do you want ME to build a chatbot?? After all, I had no experience with coding or software programming, but I just knew a lot about what customers were asking us about. So, I was a bit skeptical about it all.”

Nadin Kempel Sigh

Even so, Nadin participated in a one-day workshop with bot expert Michael Larsen, at the Certainly headquarters in Copenhagen, where she learned everything she needed to know from ideation to building and training of chatbots.

Three people standing in the office and smiling, with a Chatbot poster on the right side
Nadin Kempel Sigh, Henrik Fabrin and Michael Larsen at the Certainly headquarters.

It turned out to be quite easy to get started, as the solution was modular, intuitive, and did not require any coding skills.

Tiger of Sweden began from a template and used out-of-the-box Natural Language Understanding with pre-made content, to have an MVP working bot – able to understand basic questions and to reply accordingly- from day one.

Building Tiger of Sweden’s chatbot – the beginning

After the workshop, Nadin gathered data on what customers asked about across the different channels and then categorized them into common themes – or as we call them in NLU/NLP, domains.

Screenshot of an instance of the NLU Training
Tiger of Sweden uses pre-made content in the NLU engine. On the bottom right corner, you insert the new sentence that the bot could not answer, and then the bot proposes the section to which the question should go and gives an indication of how sure the AI is (In this case 92).

“We could see that customers, for example, often inquired about their order status, shipping costs, and how to return items. So, we quickly set up some answers where Tiger of Sweden chatbot could respond to just those things. But unfortunately, we also made a classic beginner’s mistake” says Nadin Kempel Sigh.

Despite Michael’s recommendations to launch the bot and start editing conversation flows while the bot was live, Tiger of Sweden decided to carefully formulate as many individual questions and answers as possible, before launching. The modus operandi was based on their own knowledge and internal everyday language.

Unfortunately, the results were a product of that bias: “We had to shut down the chatbot the same day we launched it. It did not respond at all to what the users actually asked, as our own logic did not fit the users’ way of writing at all” says Nadin Kempel Sigh.

This was the first meaningful learning lesson for the newly appointed Bot Operator Nadin. She started copying customers’ questions directly into the chatbot and then writing answers that made sense in that context, in the brand’s tone of voice. The result, this time, was very different.

Building Tiger of Sweden’s chatbot from the users’ input

Day after day as more customers were chatting with the Tiger of Sweden chatbot, the company was able to train and improve its understanding of human input and the formulations of the answers.

The more customers communicated with it, the better the chatbot got. Answers to customers’ inquiries got quicker, and the company’s customer satisfaction higher. Nadin’s editing process also improved.

“The first month I spent 80% of my time training and improving the chatbot, but today I have a routine: I look through the customer interactions every morning and then train the bot based on the feedback we have received,” explains Nadin Kempel Sigh. She now uses around 1-2 hours per week on checking and updating their bot.

Tiger of Sweden now had an AI chatbot capable of answering many customer questions, ranging from opening hours to clothing sizes and order status.

Screenshot of Tiger of Sweden's webpage with a new version of a chatbot as a shopping assistant
Tiger of Sweden’s bot is about to become a shopping assistant that can act as a conversational size guide.

Results for Tiger of Sweden’s chatbot

Since starting to use the Certainly chatbot, Tiger of Sweden has seen an increase in overall inbounds because of the 24/7 customer service availability.

At the same time, the type of inbounds changed radically: inbounds on email and phone decreased by 50%, while inbounds via chat increased by 400%.

While the inbounds via chat increased, the first-reply time for customer service employees dropped significantly from what it was before the implementation of the Tiger of Sweden chatbot: from 17 hours to just 10 hours.

The chatbot works as first-line support freeing up time for their agents, who now have more time to respond faster to customers and more time to handle more complex tickets, that might require an agent to look into them.

Today, 35% of Tiger of Sweden’s customer inquiries are handled by their chatbot.

Perhaps one of the most uplifting results from the chatbot implementation is that the Customer Satisfaction score has increased in just one year from 73% to 96%.

Nadin Kempel Sigh stresses how the bot is playing a vital role in the improvement:

“We can feel the results in the business.”

Nadin Kempel Sigh

The Future for Tiger of Sweden’s chatbot

The future goals for Tiger of Sweden will be to expand the scope of their chatbot from Customer Service to becoming a Shopping Assistant. The idea is to have the chatbot help with more pre-purchase questions such as clothing size and materials.


Authors: Beatrice Carraro

Trending customer stories