As Easy as Chatting: Three steps on the way to Grub247
My partner, Tom, and I were sitting in a fast casual restaurant, doing a post-op on why our latest startup company, Gabriel, did not take off as we had hoped and anticipated. In a world enamored by lead generation, Gabriel was designed for automated retention. Call me old fashioned, or maybe old school, but I believe that providing great service to customers is the surest way to long term success.
Churn, the number of customers you lose month over month, is such a huge problem in today’s startups that it seems the accepted solution is to spend more money on lead gen, until churn becomes a tolerable nuisance you can ignore. (Yeah, this rarely ends well) A mentor of mine once told me, ignorance is the act of ignoring, so don’t be ignorant. As entrepreneurs we are optimistic by nature and it takes a strong will (and a heavy heart sometimes, truth be told) to look at your numbers fairly and unbiased. But I’m getting ahead of myself, so let’s take a quick trip back.
Tom and I, both graduates of Founder Institute, had previous startup experience and understood that ideas were cheap and plentiful. What mattered was quickly testing an idea and execution. But what idea would we take that next step with? We both are students of tech trends and knew that chatbots were quickly up and coming. What if we created a service bot, that made a service as easy as chatting? That was the beginning of what would become Grub247. We decided to both take a week and reconvene with our ideas on what industry we wanted to address.
Step 1: Choosing the right Industry
That week I studied Google trends, read articles on service businesses and industries and looked back at notes and ideas I have had over the years. With each possible business model I asked myself, “Will this make service as easy as a chat?” Many times, the answer was no, as the technology, from what I saw, was not there yet. The tech needed more time to develop, and, at this stage, we were looking to test a concept, and demand, not to code natural language.
When we met again we reviewed our lists and it came down to two. We could relaunch Gabriel as a chatbot, to help sales people stay in touch with their clients, or we could launch a bot in the food delivery space. We knew food was a crowded space already but we also knew two things. First, it’s a huge market that no one has yet dominated, and second, there was no chatbot in the market, yet. We decided then and there to launch a new chatbot in the food delivery space, and that is when Grub247 was born.
Step 2: What Platform To Build On
The next decision was what platform to build on. Given Facebook was making a big push into chatbots, and that it’s hard to argue with one billion users, we chose Facebook Messenger. While the goal is to eventually build on other platforms, including Kik and Slack, initially you have to choose one to start. If you try to be everywhere at once, you’ll quickly get bogged down in development. Remember, in the beginning you are testing your idea, so choose your ideal market, choose the best platform for you and get building.
We got to work the following week, first testing with existing companies that allow you to quickly and easily launch chatbots. After two weeks of building and testing, we decided to change direction and build a bot ourselves, which meant we would have to code from scratch. We lost a little time, and it cost more money, but we believed that it was better for us to have complete control of our product and source code.
Many people enter the bot space with a real belief in natural language, meaning the ability to have a fairly sophisticated conversation with a chatbot. I’ve seen some valiant attempts at this. This is definitely the ultimate goal, but for anyone who has ever tried to have a conversation with Siri or Alexa, the limitations are evident. We’re quickly approaching the time when artificial intelligence can and will handle these types of interactions, but we’re not there yet. We decided to focus our energy and resources on making an easy to use, and simple bot. And I say simple only compared to natural language.
Step 3: Building and Testing
Once we started coding we quickly realized that there are so many variables a bot needs to be ready to address and respond to. We needed to think of every possible scenario and be ready. Of course, there will always be new situations arising, but a well thought out bot will first have a well thought out development plan and testing phase.
After about a month and a half of coding, testing and placing food orders, we finally felt we were ready for our beta launch. Aside from a few unexpected glitches (delivery men do not like it when your phone number is not on an order), the tests went well and we decided to further promote Grub247 in the marketplace.
Launch and Marketing
Given that we are built on Facebook Messenger we had devoured any and all marketing materials related to advertising on Facebook. It was natural for us to target Facebook users, who would be familiar with Messenger. Facebook ads are a great way to reach an audience, and they are also a great way to waste a lot of money, if you’re not careful. As we progress I’ll be sharing my insights and experiences in marketing a chatbot.
We have great plans for 2017 for Grub247 and I’d love any feedback from people. I’m also a huge believe in helping out fellow entrepreneurs, so if there is anything you think I can help you with, please just ask. Wishing everyone an entrepreneurial 2017!
