HEMA is innovating: chatbots for better and faster service
The number of customer questions and complaints sent through social media is continuously increasing. These increases can not be ignored by organisations. Therefore, organisations are always looking out for scalable solutions that can still provide an excellent service. HEMA, whose mission is to make people’s everyday lives more easy and fun, was faced with a challenge.
Their solution was found in incorporating a chatbot as a conversational agent. We spoke with Ronald Gerrits, Webcare Coordinator at HEMA. We spoke about the process surrounding the innovation and the influence of OBI4wan in this process. In the development, training and implementation of the chatbot for a better customer experience.
Big in exceptional simplicity, the sky is the limit
HEMA is one of the largest Dutch brands on social media. With more than 750.000 fans on Facebook, 60.000 followers on Twitter and 307.000 followers on Instagram, they have shown to be a stable online brand. A strong social media strategy, campaigns and engagement have helped this classic Dutch brand to connect with thousands of fans online. HEMA is able to connect their products to reality in a humorous manner. They are good at picking up on actualities, back-to-school campaigns and have strong webcare responses.
The growing number of fans on social media resulted in a growth in the amount of client contact moments through online channels. A webcare team was set up, which has now grown to a full department and is taken very seriously. HEMA won the CCDNA award in 2016 for most customer focused offline retailer and was nominated in the Best Social Awards in 2017 for Best Brand. The simplicity boasted through HEMA has no borders. The chain is active with about 200 stores in Spain, England, Belgium, Germany, Luxembourg, France and Austria. 3 stores will be opened in Dubai this year, as first stores outside of Europe.
The growing importance of webcare
The amount of (social) media messages that HEMA receives has increased immensely over the past years. They experienced an increase of 60% in 2017 compared to 2016. Especially messages received through Facebook Messenger have increased quickly. HEMA now answers about 30.000 messages per month, of which 8000 come in through messaging channels.
The questions HEMA receives vary from questions about products or shops to questions about their newly launched loyalty program. Each conversation is tagged with a specific subjects, which helps detect which questions are asked most frequently. This helps to structurally identify points of improvement.
“Webcare is important for HEMA as it provides insights into how customers experience products and/or campaigns. Webcare provides us with unfiltered opinions and great compliments about our products and campaigns. We collect this feedback and incorporate it to improve our products and campaigns.” – Ronald Gerrits, HEMA
Innovating: chatbots for customer contact
Good customer service is an important aspect for all organisations. In 2017 HEMA did research into new solutions to improve their customer contact through (social) media channels. Their growing number of messages and wish to renew their technology, led them to the chatbot solution.
“A chatbot aids us to work more efficiently. It picks up the simpler questions. This helps your webcare agents to focus on resolving other, more complex questions.”
As all incoming messages had been tagged with a specific subject, it was easy to find the most frequently asked questions. For HEMA most questions were focussed around ‘Meer HEMA’ their loyalty program and online orders. In these questions important information such as order numbers or e-mailadres were usually missing. Which meant the webcare agent had to request this information, in order to fully help the customer. This is unfavourable for the customer as the conversation time increases. Ultimately, this influences customer satisfaction.
A chatbot as job preparator provided solutions: the chatbot can aid the customer service agent by asking the customer for additional customer information. The service agent that will pick up the question afterwards, will be provided with all necessary information and will be able to provide the client with an answer quicker.
The combination of humans and technology is therefore implemented optimally. Simple questions are automated, which leaves time for the agent to handle more complex questions. This does not decrease the customer experience, in fact, it makes the service offered by the customer agent more personal and valuable. A chatbot provides you with the opportunity to go the extra mile for your customers.
Determining the goal is the first step in the development of a chatbot, together with the amount of tasks. Which questions should the chatbot answer? And what is the goal? In this case, the goal was to decrease the workload of the employees and to help customers faster. After this has been decided, you must determine what rules your chatbot must oblige to. What questions must the chatbot ask to receive the right answer? When should the chatbot pick up a questions? And especially; when should it not?
“We taught our chatbot to recognise order numbers. For instance a question about an iTunes order, which has a different order number than a regular online order, will also require a postcode from the client, to answer the question. This teaches the chatbot what information is necessary for which question.” – Ronald Gerrits, HEMA
When a chatbot is able to recognise an order number, it will also be able to recognise incorrect order numbers. As an organisation it is advisable to decide on what the follow-up should be. If the chatbot recognises the order number to be incorrect, but the customer is convinced it is correct, there is a possibility that the chatbot will repeat itself which could create annoyance for the customer. These types of decisions are called business rules, are a set of rules which decide exactly what the chatbot is allowed and not allowed to do.
HEMA decided that messages with an incorrect order number would be passed on to a webcare agent. This means the contact with the chatbot will end once the incorrect order number is received. This helps to avoid the chatbot repeating itself which may frustrate the customer. The webcare agent can then decide what is necessary to answer the question correctly.
The chatbot was launched internally by the CEO of HEMA. Mainly to create internal awareness for the work conducted by the webcare team. After only a pilot of a few months, Ronald shared the following results with us:
- Of all Facebook Messenger messages, 29,4% regarded the loyalty program. Of all these messages, in 17,2% of the cases no order number was attached. Meaning the chatbot could pick it up, this saved the webcare agents time.
In the overall client conversation, a complete time saving of 2 hours was measured in order to solve customer problems.
- In addition to this, HEMA launched a second chatbot, which automatically processes messages in which people tag each other, which does not require human action. During the last period this included 11,7% of all messages. Manually handling and processing these messages is no longer necessary for webcare agents.
In addition to this, HEMA launched a second chatbot, which automatically processes messages in which people tag each other, which does not require human action. During the last period this included 11,7% of all messages. Manually handling and processing these messages is no longer necessary for webcare agents.
Optimization as a continuous process
The HEMA chatbots are continuously improved. By manually assessing whether the chatbot has done something right or wrong or if the chatbot might not have activated in a situation in which it should have. By doing so the chatbot can learn to evaluate when it should pick up a conversation.
“By implementing your own tests, you will gain a better view on what a chatbot can do for your organisation. Take time for this, as the success of the chatbot directly influences the customer satisfaction.” – Frank Smit CIO at OBI4wan.
Look to the future
If it is up to Ronald, HEMA is ready to continue to innovate in the future.
“We are looking into how we can expand the chatbot within Facebook Messenger, but our focus is also on a broader application within our customer service. For instance the FAQ on our website.”
Ronald also advises other organisations the following: “Just get started with a chatbot. The technology makes it possible to answer customer questions more quickly. It will help to experiment with the technology to see if it is suitable for what your organisation wants to offer.”