Our Wedding Chatbot
Using technology to ease a personal affair
In an effort to be practical, my husband and I decided to marry on our 10th anniversary and keep our anniversary date the same. He’s a developer and I’m a service designer, so during the planning we also agreed we would have a very practical and effortless wedding (or as we like to call it, the Minimum Viable Wedding). For us this meant fulfilling family expectations but spending as little resources and time as possible.
Being highly committed to my work made planning a wedding a bit of a challenge; I had to schedule one or two hours weekly on my calendar, and also had a Trello kanban and Airtable lists of providers and guests. Right after sending the e-vites and publishing the website, my inbox kept getting flooded with questions from our guests. I had little time to reply and often left some messages unanswered for weeks. Most of the questions were about very basic information that appeared on the e-vite or the website, and even if I told people that the information they were looking for was on the invite, they preferred to send a message and have a more “personal” answer directly from “the bride”. This soon proved to be very time-consuming for me and my idea of keeping it practical.
By that time, coincidentally, I started working on a chatbot project for the bank I work for, and while benchmarking chatbot platforms, I came to the realization that it was the perfect solution for the flood of wedding messages.
The very next weekend I started detailing the conversational architecture mainly including the frequent questions asked like the date of the wedding, the time of the ceremony or reception, the location, accommodation, the dress code, and also somewhat controversial topics like the registry or the food.
Cross-pollination between projects
The first stage of creating our wedding chatbot was to identify common questions and lay out a carrousel with suggested topics as a conversational starter.
It was very fun to think about all the possible questions, from where to stay or what to do in town (if they were visiting from another city), where to park, why we chose LA as our honeymoon destination, who our parents are, how to RSVP, what kind of music will be played, what does ‘cocktail attire’ mean, and even what to do if they wanted to bring their own alcohol.
Working on both projects (the bank’s and my wedding chatbot) helped me quickly apply new learnings from one another. And since the development of the bank’s bot was going to take longer, I was able to test first-hand the reactions of people and think about new problems and opportunities for the bank bot.
What else will guests need to know?
After writing all the possible questions and answers I could think of, I realized people’s doubts evolved through time. For example, as the wedding day was getting closer, people would rather ask how much time remained until the big day, instead of the date of the wedding.
I talked a bit more with a couple of guests to get a sense of what else will they need to know in the future, and planned some follow up questions (using the Quick Reply feature) on some of the topics that seemed to require more in-depth answers.
For example some people were worried about the dinner menu, since we’re vegetarian (well, actually pescatarian), most wanted to anticipate if they should grab a snack before the event. So on the ‘menu’ answer the chatbot explained there would be some veggie and fish options, and added at the end a quick reply to explain what pescatarian means.
Another topic where I included a quick reply was the registry, where I added a bank account for people to make a deposit if they want to contribute to our honeymoon but didn’t use PayPal.
Launch time! But who am I talking to?
After feeling satisfied with the range of possible questions to answer I sent an email inviting our guests to use our chatbot on messenger, and every time I got a new question on my inbox I would just copy-paste a link to the bot. But surprisingly it wasn’t as successful as I expected. Apparently guests felt it was too impersonal to talk with “a machine”.
I spent some time thinking about this, as it was also an issue I was working on the bank’s bot (I was defining voice and tone design principles and personality). I realized it was important to assign some personality traits and an explanation of who they were talking to.
So now every time anybody had a question, I wouldn’t tell them to talk to the chatbot, but instead told them to talk to my “wedding’s virtual assistant”.
I added a fun onboarding explanation and some light conversation answers and quips, like some trivia about our relationship, some emojis every now and then, and also the default reply for when the bot doesn’t have an answer or didn’t understood the question.
Takeaways
This was a very insightful exercise that allowed me to put into practice on a personal level something that I was researching and designing at work for a completely different industry. A lot of knowledge between both projects were cross-pollinated and helped me anticipate situations for both chatbots.
The feedback section on the platform was also very important, since it helped me learn more about use cases I hadn’t thought before, like what would happen if people had to cancel after they have already RSVP’d. So realizing situations like this and adding some instructions to the chat was very helpful for guests.
At the end it was more of a geeky personal project, but now that the wedding craze is over, I’ve had some time to reflect on many things that happened, especially around using technology and methodologies beyond the workplace to help me solve personal pains too. And the most amazing thing for me is to hear from acquaintances that some of the features that we designed for our wedding have become a reference for several wedding requests for designers and developers locally.
Our wedding chabot still gets visits from people exploring it — who later go on requesting their designer/developer of choice a similar chatbot. This may be the start of a new wedding trend.
Here’s some final stats from our wedding chatbot:
• 30% guests used the bot to get their questions answered
(mostly millennial friends)• 45% used it more than once
• 65% made open-ended questions
If you have any questions or had a similar experience I would love to hear it on the comments or twitter @gabrielasalinas
