Scaling Conversational Commerce

Go to the profile of Brandon Chu
Brandon ChuBlockedUnblockFollowFollowing
Product @Shopify, [email protected], co-founder of Tunezy. I have a strange love of frameworks.
yesterday8 min read

Scaling Conversational Commerce

We just enabled 250,000 merchants to sell in messaging, without building a bot.

I’m the product lead on a team that’s been working on conversational commerce at Shopify for about year now. Yesterday, we launched the ability for hundreds of thousands of Shopify merchants to get their products into chat threads and sell directly to their customers through Facebook Messenger. Check it out:

While we were thrilled to get some awesome press coverage, a part of the story that many got wrong was that Shopify had built a bot. We didn’t. In the experience you see above, there’s no natural language processing, AI, or even character in the responses. Hardly a bot.

We did something subtly different, but powerful: we simply enabled a merchant’s product catalogue to exist richly inside of Messenger, and then we got out of the way.

As much as we love the innovation happening in the conversational commerce space, our opinion today is that bots are not the right medium to drive it forward. Instead, we believe that fostering great human conversations between businesses and customers is the right approach.

To unbundle how we got to that line of thinking, it’s good to start from the beginning. Our team’s exploration of conversational commerce started a year ago with a simple premise, which I’m sharing verbatim from an internal doc I wrote just after we got started:

Commerce is a deeply human behaviour that our species has been doing since the dawn of civilization, and for millennia things have generally stayed the same: people walked into stores and then bought things.
Then a few decades ago, the internet happened and now we rarely go into a store unless we have to. As consumers, we’ve saved a lot of time, gained more product selection, and received cheaper prices as a result, but something we’ve lost was the intimacy with a business that we used to have: that interaction with a shopkeeper, whom we shared our needs with and then relied on to find us the right products.
Although the growth of online shopping has been tremendous, we have failed to bring back that intimacy into the user experience. In fact, as good as we’ve become at designing shopping websites and apps, we really haven’t evolved beyond a simple digitization of the Sears catalogue.
Messaging enables us to rekindle the conversations we used to have with businesses, and recapture that lost intimacy.

The key insight of the premise is that conversational commerce is not something new that we have to teach people to do; it’s something natural that we took away with the rise of online shopping.

This being Shopify, the natural thing we did next was to build some shit, in the form a hack days project on the concept. We used Facebook Messenger as the exploration canvas but this was way before the Messenger Platform even existed, so basically we made it all up using animation prototypes.

The hack days project that started it all

Exploration #1: Getting into a thread with a business should be seamless and in context.

We explored the scenario of asking a friend who’s a product expert in category X for advice.

Exploration #2: When asked a product question by customer, a business would need quick access to their product catalogues so they could easily respond.

We dreamed up this keyboard thing

Exploration #3: Since messaging a business still isn’t a common thing, we need a better way to foster conversations.

We landed on a simple button businesses embed on their websites to initiate a conversation

Aside from being hella cool at the time, the discussions and research from the hack days taught us something. Even though online shopping had suppressed our natural commerce-conversation behaviours, people were still finding a way to do it. They were doing it through Facebook comments on a business page, through email back and forth with the business, through messaging friends for advice, and by sharing links.

We recognized then that our work in conversational commerce should be less about describing the conversation, and more about enabling it. What enabling it actually meant, was facilitating the entire customer journey inside of messaging.

Parts of the journey that didn’t exist are in red

This customer journey became our roadmap:

  1. Discover — We needed a way to get people into conversations.
  2. Converse — The biggest one, we need a way for both customers and businesses to have rich commerce conversations so they can land on the right product.
  3. Buy — We need to facilitate a transaction.
  4. Support — We need a way to keep an order context in chat after the purchase has been made, in case the customer has a problem or needs to be notified of a change.

From idea to reality

In Shopify parlance, we set out to draw the owl. To be honest, at the time we really had no idea how to actually achieve this because the messaging platforms weren’t yet public. Around December 2015, we formally partnered with the Messenger team to make something happen, and got access to some of their private SDKs, as well as explored other technologies we could leverage.

Since the beginning of this year, we’ve launched a set of products that at first glance may seem like experiments or small features, but upon closer inspection were really all pieces towards enabling the conversational commerce journey.

Feb 22nd, 2016 —We launched Shopkey, an iOS keyboard that gives Shopify businesses their product catalogue in their pocket.

Incredible how close this was to our imagined hack day project

The impact to the customer journey:

Businesses could now seamlessly access their shop from their pocket, and the coolest thing was that as an OS-level keyboard, they could do this in any messaging app and even email. We felt this feature gave us only 50% progress on the “Converse” part of the journey, since it only enabled the business to access their products, and not the customer they were talking to.

April 12th, 2016 — We launched Messenger Channel at F8, enabling businesses to start conversations with customers by sending them order receipts and delivery notifications in Messenger.

Businesses could embed this widget into their checkout flow
Our launch video

The impact to the customer journey:

This clearly gave us a check mark on the “Support” section of the journey, but also added 50% to the “Discover” section because it created a chat thread with existing customers, who often used it as a starting point for their next purchase. The downside was still that there was no great way for a new customer to enter into a conversation.

May 30th, 2016 — We quietly shipped a feature that enabled businesses to embed a “Message Us” button onto their online stores

What a business configured in Shopify’s admin
Example implementation on a store

The impact to the customer journey:

This bumped up “Discover” to green since it finally enabled new potential customers to have their first transaction with a business go through messaging.

Oct 5, 2016 — We launched the ability to shop in Messenger.

To try it out just message the page:

Customers can now access a Shopify business’ products in Messenger, share them with friends, and checkout via web-view from the app.

The impact to the customer journey:

This completed the second half of the “Converse” part of the journey by also enabling customers to explore and bring up products in conversations. It also made progress on the payments front, but the experience can still be improved so we’re leaving that at yellow for now.

A complete journey, if still imperfect

Those four product launches collectively completed the customer journey we set out to build nine months ago. In retrospect, it’s crazy to reflect on how close some of our earliest — completely made up - hack days ideas came to fruition, and that’s exactly what I was doing when I wrote this tweet:

Human conversations are still the most important

Our team isn’t close to done, we will continue to dive deeper into rekindling the natural interactions of commerce and conversation. But today, we can confidently say that hundreds of thousands of businesses on Shopify can facilitate an entire customer journey through messaging, and they will do so with their own words.

Only time will tell how both customers and businesses adopt these interactions at scale, but if there’s anything that Shopify’s product history has taught us, it’s that businesses will use the tools you give them in ways that you cannot predict, and will ultimately unlock more potential from them that you ever anticipated.

Wait, what about bots? Why didn’t we build them?

Because we want the developer ecosystem to do it for us.

Shopify is a platform for hundreds of thousands of businesses, and with that comes enormous diversity. There are two key insights that build off that which make building generalized bots for all so difficult.

  1. Shopify businesses care deeply about their brands, and their customers care about the people behind them. Bots can take away the voice of your brand.
  2. AI isn’t ready for the diversity of businesses on Shopify. We all love Dominoes pizza bot, or Uber’s bot, but the reason they can delight users is because are for a single use case and brand. Shopify’s 300,000+ merchants sell products accross hundreds of product categories, and in as many languages; there is no generalized NLP and/or AI that can currently handle that diversity.

To build great bots would mean building hundreds of them, specific to the needs of each niche, which is something that would take us too long to do.

But that doesn’t mean the developer ecosystem can’t.

If you liked what you read be sure to ♥ below — it means the world to us.

Learn more about Shop in Messenger here, and if you’re interested in product management, check out some of my writing on it.

  • BlockedUnblockFollowFollowing
    Go to the profile of Brandon Chu

    Brandon Chu

    Product @Shopify, ex-@FreshBooks, co-founder of Tunezy. I have a strange love of frameworks.

    Previous Post
    Next Post