Intercom’s Brian Donohue on Good Product Design and the Future of Customer Service

When I bought my ticket to Inside Intercom: Lessons Learned months ago, I assumed it would be a nice meet up with a presentation and maybe some snacks. The morning of the event I was emailed an itinerary which stated that the “show” would begin at 8PM. I arrived at the Park Avenue Armory not sure what to expect.

The armory is a giant, commanding space and the Intercom team filled it with a huge screen, extensive bleacher seating and incredibly high-production lighting and sound. 90 minutes of presentations followed where the audience heard from Co-Founder Des Traynor, Senior Customer Support Lead Sabrina Gordon, Senior Director of Marketing Matt Hodges, and Director of Product Design Emmet Connolly. Together we laughed at tech jokes, related to office anecdotes, and listened to Intercom lay its own growing pains squarely on the table.

Intercom is a customer messaging platform that has made a name for itself by focusing on humans and their humanity. Intercom’s incredibly rapid growth solidifies their initial theory that there was a better way for businesses to communicate with customers. They’ve grown from $1 million in annual recurring revenue (ARR) to $50 million in ARR in just three years, making them one of the fastest growing start-ups in the game, rivaled only by Slack.

But what strikes me most about Intercom is their radical transparency and bold opinions, something I’ll discuss with Intercom’s Group Product Manager, Brian Donohue, in this interview. From their own well-known blog to the Inside Intercom World Tour to their book: Intercom on Starting Up, the leadership is not afraid to share their thoughts, call for change, admit mistakes, and even use colorful language. Brian and I discuss what it’s been like to be part of Intercom’s wild growth, the company’s vision for the future of live chat, and what phase AI development is actually at in the world of Customer Service.

Intercom is one of the fastest growing start-ups from $1 million in ARR to $50 million in just 3 years. Click To Tweet

KRISTINE COLOSIMO: Brian, thanks for joining me. You’ve often talked about Intercom’s Product Story on the blog and at conferences – what’s your own Intercom Story?

BRIAN DONOHUE: I’ve been at Intercom for about 3 and a half years. When I started, the company was around 30-40 ppl. We’re now at over 350 employees so I’ve been lucky to see a big part of the ride that we’ve gone on. It’s been super cool.

As for joining, I had actually worked a little bit with Eoghan [McCabe, the CEO and Co-Founder of Intercom] in the earlier years and then I worked with Des [Traynor, Chief Strategy Officer and Co-Founder] for a year. I followed them really closely as they set up their agency and then when they set up Intercom. I was a big admirer of what they’d accomplished in the super early days and I distinctly remember when they hired Paul thinking, “Holy Shit! They got Paul Adams to join!” They were really picking up momentum and about 6 months later they decided to break product into multiple teams and to hire for the Product Manager Role. I jumped at the chance and haven’t looked back for a second!

At a company like Intercom, the opportunity that you have for growth and the amount you learn in just 6 months might is equivalent to a couple years at another company. It’s similar at a lot of other tech companies that are really ambitious and trying to do something interesting. It’s been the typical crazy rollercoaster but a really fun ride.

Brian Donohue of @Intercom talks the future of support chat, Steve Job's jeans, and more Click To Tweet

KC: What does your job role entail?

BD: At Intercom I’m a Group Product Manager, which means I work across several teams. I focus primarily on Respond, Educate, and our Messenger Teams. My job is to help the product manager on those teams and to help figure out what the most impactful thing to build is. Then to execute on building that and then validating if that product is solving the problems we set out to solve. Which is mostly what product managers do just across several teams

KC: When I think of Intercom, I think of transparency. You guys openly talk about your own growing pains and mistakes so that others can learn from them. How did this approach develop?

“I think Eoghan’s perspective is that he’s going to keep pushing it and putting trust in everyone.”

BD: You’ve really spotted something because when I think about it, this theme really does run through so much of what we do and how we think about things. What’s clear since I’ve joined is that transparency is just something that is really important to Eoghan. Internally we have weekly all-hands meetings where everyone gathers together. He’s put a priority on being transparent with our business numbers, our challenges, and sharing what’s going well. As you scale, you really test your values – how much can you sustain this value of transparency when you’re 20 people versus when you’re 80 people? I think Eoghan’s perspective is that he’s going to keep pushing it and putting trust in everyone.

I think these values from the leadership are important for setting the tone for the whole company and as a result, transparency runs through not only how we work internally, but also what we build and how we act. The blog, for example, has carried an almost implicit mission of transparency where we can say, “Here’s what’s worked and here’s how things we’ve tried have messed up.” There are a lot of people who know us for our blog even more than our product. In fact, there’s a blog post by Sharon Moorehouse, our Customer Support Lead, who wrote a post about how to say no to customers, which moves into how transparency impacts how we ourselves act as a company.

Sharon writes that being transparent is not most human’s default – overpromise and underdeliver is kind of the way that most of us act without thinking about it. When people ask for feature requests, the default is to say, “Oh yeah we’ll get to that soon” – that feels like the nicest thing to say. The person may leave with some optimism, but if you know your team is not actually going to get to it, then that optimism will just turn into frustration. It’s much better to be more honest up front, even if the immediate impact is more negative. The upside of doing this is actually really big because when you tell them why it’s not going to happen, you start to build trust. Trust is the hardest and most valuable thing to get.

KC: You mentioned transparency was also apparent in the things that you build. Can you give us an example?

BD: It might sound cheesy to map things to your company vision, but there are sometimes where we directly do that for our product. Our mission is to make business personal and on our Messenger product last year a big goal was trying to embody that mission in a live chat messenger.

It was in the big details, like having big profiles to make it clear that you were talking to a person. If you can make your support team seem more human, it can change the tone from frustration to a more human conversation. A smaller detail is the inclusion of “read receipts.” Most consumer messenger apps have this but for businesses, their initial thought is, “Oh no, we don’t want to let [customers] know we’ve seen their message!” They like having a curtain to operate behind because it’s more comfortable to have that barrier between them and the customer. We made the decision to not let them turn this feature off, because then the customer knows that this conversation is continuing and isn’t left in a state of uncertainty. It’s this little signal that’s really important for conversations. The transparency here might be a little uncomfortable, but we actually think that initial discomfort is worth it.

Another detailed product example of this is the inclusion of the teammate’s location. The customer does not need to know the Customer Support rep’s location, but it personalizes the person in the business that they’re talking to. There’s now some identifiable thing that helps make this person more human. It’s a signal that grounds this person in the world, versus a fake avatar with a fake name that maintains that curtain between business and customer. None of our customers were like, “Oh I’d love if you could show our location to our end users,” but that’s another place where we believe that transparency, despite the initial discomfort, is a good thing.

Read @brian_donohue on the value of transparency & how it runs through product design at @Intercom Click To Tweet

KC: I love hearing about intentional product decisions like that! Why do you think this ultra-transparency works so well for Intercom?

BD: From my own experiences speaking at Inside Intercom, I think people often think that they’re the only one struggling with these issues. They might look to Intercom as a thought leader and what really resonates with people is when we pull back the curtain and say, “Hey we’re struggling with this as well. Yeah we have a lot of strong opinions and we’ve thought about it a lot, but we’re constantly changing.” And just that level of transparency and honesty to your peers in the industry helps relax everyone. Everyone is trying to figure it out.

Intercom on the Future of Support Chat

KC: You wrote an article for the Intercom blog called “Live chat was great, but now it’s history.” What are some of the problems with the way business live chat is now?

BD: There are 2 points here:

  1. Real-time communication can be incredibly powerful – it can be magical for sales and support efficiency – but by having it only be real-time, live chat is basically like a digital live phone call. You’re still put on hold waiting for someone to come talk to you, then you have your chat and, similar to a phone call, they ask you if they can close the chat. Which is similar to hanging up the phone.
  2. Second, there are some other points that make this a less than optimal experience. For example, once the chat is gone, it’s gone forever, you can’t get back to it.

To address these issues, we’re trying to bring what consumer messaging apps have brought to the table and apply it to business communication. Fluidly moving between synchronous and asynchronous messaging means that there might be 30 seconds between answers or 10 minutes or an hour. No one thinks about this when using consumer apps like WhatsApp or Facebook Messenger. But when you apply that to the live chat world, it is actually a really different model. First, because it’s on a desktop, and second because end-user’s expectations are: “This is either live chat or it’s email.” It’s been a multi-year challenge to make that work. We really do think this is a better way to communicate for everyone and that it really will become the new messaging model – we’re just in that transition period.

“We realized notifications are actually the secret sauce.”

KC: Changing live chat is not a task many would take on because as you said in your article, many people don’t think it’s broken. How do you go about reinventing something that is already so well-established? What are some of the biggest challenges?

BD: We thought about why asynchronous messaging works so easily on our phones and yet is so different from the live chat model that exists on the web. We realized notifications are actually the secret sauce. That’s why on your phone, you message someone and you just put your phone away. You know you’ll get a notification when they reply and you can just reply in real time or you just say, “I’ll get back to that later.”

The challenge with bringing this model to businesses is ensuring that the customer can get notifications when a reply happens on both desktop and mobile. If you have a mobile and web app then you’re in the magical world because you can do this. The customer can go on your website, start a conversation and they don’t need to wait for a reply. They’ll either get a push notification on their phone and then continue the conversation there in real-time or whatever suits both parties. That’s our vision. The problem is when you don’t have a mobile app or they’re not yet a customer and they haven’t installed your app yet. That’s where it gets more tricky and we’re relying on the crossover to email rather than pushing a notification to their phone. Of course, email will most likely send a push notification to their phone, but we have to account for every scenario to make this a really seamless experience.

This cross-channel fluidity is the main challenge we’re facing, and the other main challenge is the mental-model that end-users have about live chat. We’ve done loads of user-tests and even though they say, “Oh this feels familiar, like WhatsApp or Facebook,” whichever messaging app they use, they still are thinking “live chat” simply because they’re on a website. We’re doing a lot of work to try to set clear expectations for end-users. For Intercom’s customers, the businesses using this chat, their mental model also needs to be adjusted. We’ve got a lot of work to do but we really believe this is where messaging will go.

Live Chat as we know is not the future of CS if @Intercom has anything to say about it! Read more: Click To Tweet

KC: You guys are basically disrupting yourselves – what is the benefit of doing this and what are the pitfalls?

BD: You have to make a bet on where the world’s going to be. If you make that bet too far out, you’ll fail. Usually, the most forward-thinking, innovative companies get screwed because they had the great idea but the timing was wrong. You’ve got to make that bet but not have it be so far out that it’s too big of a shift. I hope that this idea is one where we’ve got the timing right!

Customer Service Connectivity

KC: One issue you talked about is the loss of a chat after the conversation happens. Internally, we created a Workato integration to move Intercom chats into Salesforce automatically because you’re absolutely right that chats contain important customer data. How can companies better utilize their chat data?

BD: I think what’s interesting here is that something that sounds very non-techy – our mission Make Business Personal – is actually very techy when it comes to delivering on that. You have to connect a boatload of stuff in the background. To be personal means not being dumb or asking for the same information again. It’s amazing that we’re still in a place where the first thing you have to do when you want support is tell them who you are and give them all this information that the company should already have. The worst thing is when you get transferred and you have to do it all again. Being personal is trying to get to the point where you have a relationship with each customer and you have many previous interactions with them. Being personal means the business can talk to the customer like you would to a human you have a history with.

If you’re talking to a customer and they’ve contacted the business 5 times in the last week, that’s a completely different space for the support team member to be thinking about than if the customer hadn’t written in for a couple of months. What’s the tone of the conversation? Is this someone just pinging in many different places because they’re impatient? No company needs 4 people working on the same issue. All of those scenarios require all of your systems talking together.

“To do this well you have to be willing to invest some effort into integration.”

Being able to plug into Salesforce like you mentioned is critical for a lot of companies. Trying to be human as a business talking to its customer is surprisingly hard when you’re doing it at scale and when you have these siloed channels: phone, email, live chat. Not pulling these together in one place is a fundamental barrier to support teams being able to operate intelligently.

KC: Are there any communication challenges you think workflow automation and bots will be able to address in the future?

BD: Absolutely. AI and Machine Learning are still on the upward swing but I think specifically for bots for customer service, the initial hype is winding down a little bit. We know there’s value that bots can bring, but it’s different than what was initially promised.

Automation and automation through bots can be great when you use bots in a modest way. When you try to pick off the simple, repetitive tasks and it feels ok to the end user, then it’s just speeding up certain parts of work. We launched our bot Operator and we framed it as the first bot with manners. This bot is not gradually going to overtake humans, we’re just trying to bite off some of the simpler things to save time for teammates so they can do more meaningful support work.

A great use for bots is to provide resources. The bot can say, “It seems like this question could be a good one for self-service,” and give you a link while you’re waiting on a support teammate to come. Then the bot can figure out if the article helped. Even that simple interaction takes a lot of work to get right and to make it happen in a way that doesn’t feel intrusive. You want it to feel natural.

It’s actually very hard to build a bot that isn’t annoying and that doesn’t seem dumb. Just getting to a place where everyone feels like, “this is useful!” is actually really difficult. That’s why we think modesty and taking those smaller steps is the best thing to focus on. The state of modesty we’re in now with bots will gradually grow, it’s just not nearly going at the speed that people are talking about.

It's actually very hard to build a bot that isn't annoying and that doesn't seem dumb. -@brian_donohue Click To Tweet

KC: Absolutely. We always say that you’re not automating a job; you’re automating the parts of your job you don’t want to do yourself. Any general advice for Customer Service teams?

BD: First, I’ll say that I’ve seen support teammates actually have two computer screens up with different customer service programs on each, where they have to go back and forth. Those decisions to use multiple apps all feel right at the time and have purpose, but it’s hard to estimate the real cost of the siloed support system. The hidden cost of multiple systems will come back to haunt you if you don’t do anything about it.

Second, it’s difficult to handle a large volume of requests and I think companies have this human tendency of trying to control the volume. There’s a temptation to turn off a channel of communication. We’ve had our customer say, “Hey, I only want to turn your messenger on during office hours. I want it off when we’re out of the office so we don’t come into an influx of messages,” but they’re not thinking about the customer and how they might want to write in at the time that’s convenient for them. There needs to be more nuanced views on how you enable your customers to reach out to you. Far better to set a realistic expectation of a response time — just tell them it will be tomorrow or Monday — but still let customers reach out when it suits them.


The Future of Customer Service

KC: Intercom is a B2B company, but on the product side you also think about what consumers on the other side of the screen will enjoy using. Do you think customers will have different expectations for support teams in 5 years?

BD: I feel like right now, a customer’s expectation of support is almost always, “Oh I hope this won’t be a bad experience.” I hope in 5 years that baseline expectation has changed. Instead of just not being bad, most companies’ support will be good and you have to differentiate by being exceptionally good now.  

I also think and hope there will be a shift in customer behavior and in the perception of self-service. Everyone’s default is to try to get a human to answer their question. Even in daily business at the office there’s a default of, “Hey, can I interrupt someone to answer my question? That would be sweet because then I could then I could get an immediate answer and finish up what I’m doing.” This is, of course, great for me, but it’s just incredibly expensive for everyone else.

So, I hope in 5 years there’s a shift to where self-service is the normal starting point for most people. It already is for lots of folks, but right now it’s often perceived as painful. This shift will really come through the products we design as we get better at helping people self-serve and delivering it at the right place and the right time. That will shift people’s mindset to self-service being the norm and where customers are totally happy doing it. This will naturally cut down the repetitive, low-value requests.

KC: I’m sure that sounds like a wonderful vision to customer service teams everywhere! You’ve probably seen the stat from Gartner which predicts that customers will manage 85% of their relationship with an enterprise without interacting with a human by 2020. I’m curious if you agree with this prediction.

BD: I think that is really optimistic. We’re not totally sure that businesses would want to cut out that much human interaction.

“We’re not at all working on AI – we think people way too loosely use that term.”

KC: So how do you think artificial intelligence will change the customer service landscape and does Intercom plan on utilizing any AI technology?

BD: We’ve definitely invested in this. We’re not at all working on AI – we think people way too loosely use that term. Machine Learning (ML) is the more relevant technology. Machine Learning is like a more modest version of AI – when you don’t describe to the computer how to solve the problem, you just say ‘here’s the problem, use your methods to see if you can come up with a solution for it’ in a way more narrowly defined space.

The challenge here is that most people aren’t honest about the scale of the technology challenge – the amount of time it takes and the amount of data required for ML to really work. If we’re talking about specifically building support tools and the challenge is to have a per-company, specific set of ML technology then it’s super hard because the volumes are way too low. The best solution would be a service that aggregates learning across all the companies using your service and then you can actually provide value. But it’s not tailored to each company.

I find that most purveyors of ML, particularly if they’re framing it as AI, are overhyping. It’s really hard, it takes a long time and it takes a lot of data. The people we’ve hired in ML have said, “Ok, what are the smaller, simpler steps we can take without ML, before we start investing in the heavy tekkers route?”

Of course, there’s also massive progress made too. Apple was sharing how they got Siri to talk with more human-like intonation and, though subtle, it’s great. She seems way more human with just little inflection changes, but the challenge they were required to go through was massive! We’re still in that curve where we haven’t hit the takeoff curve.

Some Advice

KC: How can companies get ready for the changes to come in the customer service landscape?

BD: Transactional customer service will soon simply not be good enough. Customer Support teams in forward-thinking companies will soon own more than just responding to customer problems, and will start to meaningfully contribute to anticipating and preventing problems. Being reactive should just be one part of the job. Motivation, not efficiency, will increasingly become the main focus of support managers.

“There’s a great clip of Steve Jobs from the late 90’s wearing a truly ugly pair of jeans…”

KC: As a Product Manager, what advice would you give to other project managers who are trying to create disruptive technology?

BD: I think the starting point is to think of technology last when thinking about disruption. There’s a great clip of Steve Jobs from the late 90’s wearing a truly ugly pair of jeans and answering a snarky question from the audience with an incredibly eloquent, off-the-cuff response that captures this.

If you haven’t seen it, watch it.

So disruption is not primarily about technology. And the second angle here is to focus not on the technology, and not even directly on the customer, but instead on the job the customer is trying to do. This is one of the primary insights of Clayton Christensen. Focusing on the job often unlocks lots of innovation, and innovation that actually is valuable, not just novel.

KC: What’s different about being a Group Product Manager vs. a regular Product Manager?

BD: The main difference is how close to the ground you are and how close to the customers you are. When you’re working across multiple teams, the huge challenge of it is figuring out your zoom level. I think the default is to think: “Well the teams are the ones who need to be close to the detail and I need to be one step removed and see the big picture.” But often the details are the difference between success and failure! So, a real challenge for this level of product management is figuring out how to zoom in and out between big picture and detail usefully for the team and across all the teams. Finding that balance based on where your teams are at is the most challenging part and the fun part about the role.

The other thing from an organizational standpoint is that, in order to move you need to put up some silos. Then once a team puts up a silo, there’s a huge risk of where you’ve broken things because, of course, your customers don’t work in silos, they’re working across your entire product. So it’s a huge challenge as you try to scale teams to be able to do more and make sure you’re doing it in a way that doesn’t break the coherent experience for your customer.

KC: Can you describe a professional or personal practice or ritual that has contributed to your success?

BD: Before Intercom, I was wary of the goal-focused approach. But as a PM, you almost certainly have a slew of things coming at you every week, from all directions. If you feel like you’re on top of everything, you’re probably not doing your job very well. And so the default mode is the be reactive. I mean, in part the PM’s job is to be reactive, but if you’re only reactive then you’ll only be mediocre. So committing to a small number of goals (not tasks) each week is how the PM can ensure they’re actually working on the most important, impactful things, and not just serving as a human ping-pong paddle. | END


Subscribe to our newsletter for our latest interviews and productivity tips >



Leave a Reply