Captain Train’s Obsession Over Customer Service

The Blogpost Edition.

This post is an English translation of my original article. Kudos to Jérémy, François and Hugo for their help.


Exactly a year ago, I gave a talk to the first batch of Lion, a bootcamp from The Family to train future startup employees.

I told them about my experience working in customer support at Captain Train, and shared what we learned during those 5 years. Just before the talk, I put my slides online:

While I thought most of the content was common sense, those slides almost went viral within the French startup ecosystem. People are still asking me if the talk was recorded, if a video is available somewhere.

Instead of doing the presentation again and again (talks don’t scale very well), I thought I’d turn it into a blogpost. I’d have a shareable link for those who want to go further than just slides.

So here we are. The article is a bit long (the talk was 2 hours long), but I hope you will learn a few things.


Obsessing Over Customer Service

July 23rd, 2016. 4 p.m. Say hello to my French slides.

For those who don’t know Captain Train, we are a startup that sells train tickets online. One of our core values is to always think about customers first.

I usually introduce myself as the first non-technical employee at Captain Train, because it says a lot on the values of this company. When the first non-technical hire is a customer support job, it gives you a clue about the DNA of the company. From the beginning, it was all about obsessing over the customer experience — up to our business model.

Unlike most web services that optimize time spent on their website to monetize your attention, we optimize time-to-value. We want our customers to get out of our website — with a train ticket, if that’s what they wanted — as quickly as possible. We want them to go back to their lives and do more interesting things than buying train tickets. The faster the people leave, the happier we are.

The Family initially named the talk “Customer Service Like a King”. I allowed myself to edit the title because, honestly, we do not pretend to do customer service like kings.

Yet there’s one thing we’re absolutely sure of: we’re genuinely interested in this topic. We’re obsessed with it. This is the main reason why I wanted to join this company. So I’m going to tell you what we did during the last 5 years to improve the customer experience.

That Awkward Gap

When I tell people I work in customer service, they usually picture something like this:

Business side. Everyone thumb up for the stock photo.

It pretty much sums up what many companies think about their own customer service. These overjoyed agents loving their job, doing everything they can to deliver a fantastic service to their customers.

In real life, when you look at what customers think, it’s a bit different:

Customer side. Just picture a compilation of angry French people complaining about customer service.

I could go on for a while — just open Twitter. Or don’t even go that far: ask around. We’ve all had the most horrible customer experience stories to tell. We all know about this gap between what companies and customers think.

But if we want to find numbers that match our feeling, let’s dig into some research conducted by serious people. They surveyed companies about the quality of their customer service.

80% of companies surveyed say they deliver “superior” customer service.

🤔

Hi Cedric.

Then, they asked the same questions to the actual customers of the companies previously surveyed. Enjoy the results:

Yep, Cedric Daniels was right.

So it’s not just an intuition. When it comes to customer service, most companies feel like they’re doing the right thing. Meanwhile, the actual customer experience is quite different.

To get going, let me tell you a little story I found on Tumblr. Be careful, it’s almost a horror story.

Cdiscount Killed Me (Cdiscount m’a tuer)

Don’t try to learn French with this sentence, it’s not even correct. Long story.

It starts in May 2014. This is the story of Damien, a regular online customer, who blogs every now and then. He buys a fridge on Cdiscount, one of the e-commerce leaders in France. Great fridge, worth over €700.

The fridge gets delivered to his home — it doesn’t really matter how.

He picks a “Garantie Sérénité”, an AppleCare-like guarantee sold by Cdiscount. You know, just in case. This gives him a right to exchange his fridge or get it fixed for free, if it happens to break down.

Summer is coming. It’s getting warmer. Also, his wife is pregnant.

Step 1 — The Purchase

Damien decides to buy a fridge to refresh his family.

Step 2 — The Failure

3 months later, the fridge breaks down. Things go South, starting with the food Damien was keeping in the fridge. Summer is here — we’re in mid-August.

Step 3 — The Maze

Damien’s (still pregnant) wife begins suffering from the heat. To appease her suffering, Damien calls Cdiscount’s customer service.

Cdiscount picks up, listens to his problem and tells him to go talk to Haier (the manufacturer of the fridge). Too bad for the Garantie Sérénité™ paid to Cdiscount, that’s just the way it is. Damien has entered the dreaded maze of customer service.

Our dear Damien calls Haier. He actually reaches ClicTel, an outsourced customer service. A subcontractor. Apparently, it’s pretty common.

Step 4 — The Technical Intermediary Waltz

On the phone with ClicTel, Damien tells his story a second time. ClicTel offers to send a technician to fix his fridge.

Within 10 days.

10 days go by. (Damien is probably older than this kid.)

Step 5 — The Missing Part Needs to Be Ordered

The technician, vaguely informed, arrives at Damien’s. He goes to the kitchen and dismantles the fridge.

The compressor is dead. So is the memory card. They need to be replaced.
I don’t have these parts in my truck. We’ll have to order them in. Should take 15 days.

Still no fridge for Damien. His wife is still pregnant, like, at the very end of pregnancy.

15 days go by.

Step 6 — Honey, I Forgot to Order the Missing Part

15 days later, the technician comes back. He tinkers with the compressor of the fridge. It works great. Solid work — he’s proud of himself.

For the memory card, it gets complicated: he forgot to order it. Oops. Of course, he also missed the deadline for the closest order. Not so proud anymore.

I’ll order right now. The memory card will be there in 15 days. That’s the best I can do.
When I had to buy a memory card for my GameCube, it was a bit faster.

Step 7 — Ping-Pong Between Services

Damien is getting tired of calling everywhere and seeing his dead fridge lying in the middle of his kitchen. His wife is now very pregnant and she feels weaker by the hour.

So he decides he won’t wait for 15 days. He picks up his phone and calls Cdiscount customer service to request a full refund for his broken fridge. Cdiscount picks up and quickly starts to wade in the clutter of procedure. They tell him they can’t do anything for his fridge. They advise him to reach the consumer service (which is entirely different). Barking, relentlessness, endless call transfers. The whole spiel. 10 calls total.

Step 8 — Someone Promises a Replacement, Finally!

Damien is sent back and forth between two services. It’s like table tennis, without the fun and the shorts.

Out of the blue, a phone agent agrees to exchange Damien’s fridge.

I’ll call you back…

Damien begins to catch sight of the light at the end of the tunnel.

24 hours go by.

24 hours go by, and nothing happens. Cdicount remains silent. This is a ghostly customer service for a customer in a rush. At this point, Damien is furious and he starts losing his shit. Now his wife has to try and calm him down. It’s been a month since his fridge is broken.

Step 9 — The Last Chance Call

Damien plunges back into the “Garantie Sérénité” he had paid to Cdiscount, desperately looking for something that could help them out. While diving into the fine prints of the terms, he finds that a device cannot remain out of order at the customer’s home for more than 21 days. In his case, it’s been 26. 26 days during which he asked over and over again for his broken fridge to be replaced.

Revengeful and disdainful, he contacts customer service once again. But this time, he’s calling with an unstoppable argument.

Customer service responds, stammers, apologizes, and agrees to exchange his fridge. He will be called back within 24 hours.

24 hours go by.

Damien waits. The hours feel like weeks. The day passes. The phone has yet to ring. They lied to him. Again.

Summer is still on. Damien’s wife is now super pregnant. He gives up and buys an extra temporary fridge.

Step 10 — The Tumblr Rage Post

Gathering his last remaining strength, Damien plays his last card: he publishes his story on his blog. He asks for help. Help from others, from Cdiscount, from anyone.

And he waits. He starts to wonder if his son’s birth will happen before he gets a working fridge.

Eventually, his story reaches a decent level of virality online. Like all things gone viral on the Internet, it unleashed people’s creativity:

The right one reads: “In customer service, no one can hear you scream”.

Step 11 — The Outcome

Red flags have probably been raised at Cdiscount. They finally do something to take care of Damien. After weeks of frustration, the social pressure, and mockery from thousands of people online, they agree to replace Damien’s fridge.

End of the story.

It took 6 weeks to get there. Needless to say Damien’s food expired.


What can we learn from this?

Yes, this is probably an extreme story. And it’s probably not the typical Cdiscount customer service experience (let’s hope not). I have nothing against Cdiscount, by the way. But this story is interesting because it’s a perfect compilation of all the worst things you can get in a customer experience.

These scenarios usually come down to the fact that nobody can do anything for you, because of…

  • Standardized answers. The customer faces people who essentially follow scripts. They can’t take any initiative.
  • Employee’s disempowerment. People constantly put the liability on each other. Ping-pong style: it’s not their problem.
  • Inexistant follow-up. The customer spends his time re-explaining his issue to multiple people who have no backstory of the problem.

Let’s play a little game. Let’s imagine what could describe a bad customer service. It could be something like this:

If you think about it, we already have a pretty good idea of what a good customer service could be.

What if We Did the Exact Opposite?

We saw the horrible Cdiscount case. It’s a good starting point to think about how we could do better.

That’s one of the founding ideas of Captain Train: instead of considering customer service as a cost center, what if we built a service that would be so helpful to customers that it’d attract others? It could serve our customers, but also ourselves. To get new customers.

So instead of answering emails in 2 weeks… Why don’t we try to answer within 2 hours? Instead of using corporate lingo and scripted answers… Why don’t we talk to our customers like real human beings? Instead of outsourcing our job to outdated companies… Why don’t we hire our own experts? Instead of leaving our customers alone on weekends, why don’t we find a way to be there when they travel?

Those are good intentions. Everyone would like to do that. But how does the implementation work? How do we actually do it? Before diving in, I have a few disclaimers to make. Because we are dealing with a pretty specific environment.


Specificities and Challenges of Customer Support at Captain Train

We are in a heavy, slow, and extremely complex industry. This has a huge impact on our everyday life. It’s important to understand this context to figure out how, from those constraints, we are trying to build the best customer experience.

1. High Growth

Like every — still alive — startup, we are growing fast. However, the support team cannot triple every year.

2. Low Margins

We live on the fees carriers pay us for every ticket we sell. We’re talking about a tiny cut off of the ticket price. That’s peanuts. We need high volume to make it. Which means our business model only works if most of our customers are autonomous.

3. Obsession Over Quality

I might have mentioned we are kind of obsessed by the quality of our customer service. There is no way we will ever outsource such a crucial activity. Bye-bye, offshore call centers!

And yet, this is how most companies think they can “solve” this issue. They find a cheap call center abroad and voilà, problem solved. Not even an option for us.

4. Train Tickets Are F***ing Complicated

This is the most important thing. Our product (basically an e-commerce service) allows you to buy other products (train tickets). We have to do support for both. It means our skillset and responsibilities cover a wide spectrum:

On top of the “technical support” of our own applications (web, iOS, Android), we’re also doing the work of travel agents by responding to requests about the specificities of each carrier. The complexity from carriers, mostly related to pricing, but also to after-sales or disruptions, eventually lands back on us when it comes to customer relationship.

But we prefer it that way. We could have chosen to send back our customers to carriers to handle every problem we are not responsible of. But we think this will give them a poor customer experience. They have better things to do than understanding that a bug in the Deutsche Bahn API — that we cannot correct ourselves — prevents us from helping them. That’s just not their problem. They bought the ticket from us. We should be the ones helping.

So we deal with problems that are not caused by us. Those external dependencies to carriers are painful for us, and difficult to understand for our customers. Those are definitely the most delicate ones to handle.

When a customer finds a bug in our product, that’s the easy part. We “just” need to fix it — it’s on us and us alone. But if they spot a bug that needs to be fixed by someone else… That’s where the real challenge begins.

I love that picture. It gives a rough idea of how complicated train tickets are.

Those external dependancies have huge consequences for us. Today, we have 14 people in our support team. We handle around 1.8 million customers. If you look at industry examples:

  • Buffer handles ~3.5 million users with a support team of 15–20 people.
  • Trello handles ~15 million users (2018 note: way more now!) with 5 people in customer service. Please read this amazing article where Brian Cervino casually explains how he used to handle, on his own, the 4 million users of Trello back in the day.

Do you see where I’m getting at with my external dependencies to carriers and their outdated APIs? Trello poeple don’t have to bear with the SNCF API and its 1,600 fares. Neither its discounts or its multiple discount cards. You can multiply that by a dozen of carriers we offer to our customers.

For us at customer support, this train complexity really is the key element in our sprint against growth. I’ll come back to that later.

5. Permanent Emergency State

Trains run every single day. Our service never stops. The incoming stream of requests from our customers via email, Facebook or Twitter, is steady. Their travel conditions can be stressful. That’s another specificity of our job.

There is no PAUSE button for us. Postponing is not an option. Even on the weekend.


Secret Sauce (Not So Secret)

Now that you understand the context a bit better, let’s dive into the details of what our everyday work looks like, and how we try to do better than Cdiscount did with Damien.

Here are a few leads. Spoiler: nothing extraordinary in there.

We Try to Answer Quickly (No Shit)

I’m stating the obvious. We have this page where we list what people say about us on Twitter. I looked at the tweets talking specifically about our customer support.

There’s this one thing that keeps coming back: speed. It feels like this is the only thing people talk about.

Speed matters so much. Speed is everything. It’s not even a cliché.

I showed you the tweets because they are public, but we get roughy the same feedback in the emails we receive.

When it comes to customer service, most of the feedback we get is about that. Speed. Is. Everything.

It sounds obvious — because it is — and yet we can never stress this enough. Even when dealing with the trickiest cases (no Sir, I really can’t refund you this non-refundable ticket), we quickly realized it goes much smoother when we respond within a very short timeframe.

KPI Stuff

Speed became so important that the median response time became the one and only indicator to measure our “performance” in customer support.

We didn’t want to bother our customers with boring and automated satisfaction surveys — which are often only used as vanity metrics.

Not only our customers have better things to do than answering satisfaction surveys, but moreover, I find these scores range from delicate to flawed. Are we measuring the satisfaction towards our product? Towards our customer service? Towards the efficient resolution of a problem? Towards a combination of all those? You can offer an outstanding customer service on a shitty product. Or the opposite. Anyway, it’s quite a slippery slope. And I’m certainly not the first to feel this way.

If we wanted to sum it up in one sentence: our customers are happy when we answer quickly. (And well, but this is obvious to everyone in the company.) In the end, it doesn’t get much more complex than that. Hence, the median response time is the only indicator we found relevant to us.

Fast + Human. Thank you for smoothing up my transition, Simon.

We show a human face. We speak like humans.

We talk like human beings. Meaning: not like a brand or a company.

Customers often expect a certain tone, when contacting a company. When they get a human answer, they’re pleasantly surprised. We are a customer service with a human face. This is where I tell you about Captain Train’s tone.

Once more, I’m gonna start from what most companies do, and the way they talk to their customers. Here is an example from SNCF, our main carrier in France (this company was picked randomly, obviously!):

We have been facing some tensions around the maintenance operations of the rolling stock since mid-August. This resulted in unusual train compositions.

— Tristan Rouzes, SNCF.

(The French version sounded sharper, but I still didn’t recover from the “rolling stock” part. Unforgettable.)

That’s how much these kind of sentences make us laugh.

In these kind of messages, you can feel a clean, politically correct tone, that is common in our era of massive consensus. All these messages read the same. Spontaneity just isn’t the norm anymore. It can sometimes look like it, but more often than not, it’s crafted by an army of PR and marketing teams.

So what do we do? We humbly try to stand out from the hypocrisy. The same sentence, from Captain Train, might sound like:

There’s still just as many of you, but we’re short on carriages. Most of them are in the repair shop, getting fixed. So people are unfortunately crammed in like sardines.

— Human, Captain Train.

We talk to our customers as if they were our friends. Of course, we cut the slang and stay polite. We are talking from individual to individual. Not from a brand to a consumer. Our company talks like a person, not like a brand. And this changes everything.

Our customers are not investors: they don’t care about our numbers or any other corporate information. What matters is if we’re reliable, honest, funny, or trustworthy.

What would a friend do? Every now and then, he would take some of his time to write you a personal letter.

Friends write letters to each other.

You can also ask yourself: what would a friend not do? He wouldn’t lie to you. He wouldn’t brag. He wouldn’t be boring. He wouldn’t try to avoid tough questions.

A friend is someone who shows consistence. Who comforts you through their benevolence. That’s exactly what we’re trying to do when we speak with our customers.

Our tone sets us apart. Jokes and absurd humour, amongst other things, reflect our identity. Even when a customer sees our voice and tone elsewhere, he recognizes us:

Friends recognize each other.

When Voyages-SNCF speaks like that, it just sounds wrong. Because it’s not a tone that sounds like them.

The great enemy of clear langage is insincerity.
— George Orwell

We Are Honest

Once again, this may sound stupidly simple. But it’s one of those things that define us. In our tone, in our image.

When we launched the tickets exchange feature months after our competitor, we weren’t exactly boasting.

Friends are honest.

As humans, we make mistakes. I’m really fed up with those companies that can’t admit they’re wrong or they made a mistake.

Of course, we’re trying to make things right. And I believe we’re doing them right most of the time. But we’re the first to admit when we’re doing something wrong.

Humans make mistakes. Friends can admit their mistakes.

In its early days, Captain Train was only available to a limited number of beta-testers — mostly our friends and family. When we were talking to those first “customers”, we didn’t even think about using a corporate tone. We didn’t want to be SNCF. We were already talking to our customers as we would to friends, precisely because they were our actual friends. When growth came, we saw no reason to change that.

We’ve always been close to our customers. From the start, without making an effort. It all came from the inside without being thought out or posed by an agency. This tone emerged from our culture. That’s why it is spontaneous.

We quickly realized that speaking without overthinking led to a simple and human relationship with people. For example, we never asked for that kind of poster that a customer sent us (and actually made!):

We also receive postcards, drawings, and love letters. I always feel like this is extraordinary. When people take the time to write a letter to a company, it’s usually to insult them.

Much love. Such kindness.

We are not curing cancer. We are not saving the planet. We’re just selling train tickets online. So why do we get this? I’m still not sure. But the way we talk to our customers might just be working.

By talking to our customers from human to human, we’ve built an audience. Not only customers. Some people love our newsletters and are genuinely waiting for the next one. Our blogposts where we share issues in full transparency, how we build features or how we deal with design or technical choices, are very appreciated.

Every boring piece of text can be made a bit less boring.

Internally, communications and customer service share this authenticity, this freedom of speech. Those roles are handled by the same people, because the support team will always be our first megaphone, the first people responsible for our image. Separate roles would ruin this authenticity.

As you can see, it turns into free advertising for us. We have a page on which you can find tweets about us on Twitter. This page makes the entire team proud, because even if we didn’t ask anything from those people, they just recommend us to all of their friends. Because they actually like us.

We quickly understood that taking good care of our customers would give back a hundred times over.

#lovewall, our Slack channel for love.

On Slack, we have a #lovewall channel where we post our favorite quotes from customers. Love messages are incredibly empowering for a team that runs on empathy. On this channel, we see dozens of messages per day. It reminds us why we wake up in the morning. If you’re having a bad day, reading the #lovewall is the best boost I know of. It’s just impossible to be depressed when you see so many authentic gratifications as a direct result of your work.

Beyond every KPI, the only thing we care about is people satisfaction. Because a happy customer will never leave us if we treat him well. It sounds like an advice from the grocery store next door, I know. But this has been so true for us. We’re here to take care of the people who pay our salaries. To provide a human and authentic service, in a soulless industry.

Not so long ago, this is what Voyages-SNCF customer service looked like.

Everyone on Support

Changing the topic. I’d like to tell you about what we call here the “Everyone on Support” philosophy. It simply means our customer service culture is not only the business of the support team. It’s a strong concern for the whole company.

This is our main support channel.

We are around 60 people in the company, of which, 56 subscribed to the Slack channel for support. I think it means a lot. Everybody is interested in customer support. Everybody knows our customers’ problems are very important and play a crucial part in our business.

We also have a channel where we automatically get tweets addressed to us. Anyone can jump in and help with the problem. Especially the most qualified people on specific topics.

This proximity between customers, support and developers, helps us answer in minutes to our customers with absolute clarity. In the above example, I didn’t have to find the best-qualified person to answer this customer’s problem: they were just here and handed it for me. I didn’t even have to ask. Even better, he shared it to the entire team (since the channel is public), and they’ll be able to use it as well.

Now, imagine how this request would have been handled in a structure where everyone is not so involved in the customer experience. Remember this ping-pong-fridge-horror-story? There you go. I feel sorry for this customer support agent, equipped with his headset and his scripted scenarios. The journey he’ll have to go through to get the crucial information. Precisely the one he would need to give an accurate answer to the customer, instead of feeling pressured and powerless in front of them.

EoS for insiders.

Every employee — from developers and marketers, to the accountant and the CEO — spends one afternoon doing support every month. This won’t break anyone’s schedule, and it makes sure this culture sticks in the company. People quickly understand that being in direct contact with the customers’ issues just makes them better at their own job.

But the main thing I learned is that there’s a huge difference between just “reading support tickets” or “listening in on calls” — that many managers say they do — and actually be in charge of answering them. As in actually writing an answer with your own fingers, being responsible for what you say and having to follow through until the issue is dealt with. The commitment that comes with answering a customer problem yourself makes all the difference.

I’ve experienced it while we grew. I was spending less and less time answering customers — to take care of the support team, amongst other things. When you don’t answer customers yourself, you have a terrifying natural bias to minimize their problems.

We Listen to What People Say

For real. I feel like I’m stating the obvious again, but here’s what you can find in a French Court of Auditors’ report about SNCF’s spending in communications:

Customers’ requests are not taken into account by the SNCF communications directors.

Customer Notifications

On our end, we listen carefully to everything people tell us. Every time a customer takes some of their time to report a bug or suggest a feature to us, we keep track of it somewhere. We call it “customer notifications”. The idea is straightforward: when the bug is fixed or the feature implemented, we can just let them know (and thank them for their contribution!). Once again it sounds ridiculously simple, but based on the feedback we get, it seems that almost no one does that. It brings tremendous value and only costs a bit of discipline and organisation.

Personally getting back to a customer — even years later — to tell them we implemented what they suggested, creates a lot of enthusiasm among customers. It’s one of our favorite tools to generate a WOW effect.

Exchanging tickets on Android was one of the most requested features.

I admit I was quite hesitant to do this when the suggestion was a few years old. Aren’t we just slow? Is it still really worth it?

Unearthing record: 5 years.

The answer is YES. A big yes. The feedback is amazing.

Those kind of small touches of care and kindness really helped us earn the trust of many, many customers. In a slow, but very sustainable way.

The way you track customer feedback is amazing! What kind of CRM do you use?

That question always made me smile. A notepad could do the job. The value lies in the genuine interest to do it, not the tool (in our case, it’s Trello).

When people like your service, you get flooded by — more or less relevant — feature requests. The hard part is then to sort out all of these suggestions or requests. As Phil Libin said:

Customer feedback is great for telling you what you did wrong. It’s terrible at telling you what you should do next.

Listening to our customers carefully allowed us to not only answer them properly, but also to build a better product. Which makes a perfect transition to the next part.

Customer Support Is Linked to the Product

Of all the things we do to improve customer experience, it’s probably the most important.

People who work on support are in the best position to know what could improve the customer experience. We put everything in place to have as little friction as possible between finding a product problem and solving it.

Here are a few simple examples of what we could call support-driven development:

You can’t begin to imagine the number of support tickets this tooltip spared us.
You have no idea how much drama this (ugly) yellow saved us from.
10 words changed. Hundreds of emails avoided. WOW + free advertising on top.

I could give you many more examples. So many great product ideas came directly from customer support, precisely because of their insight. We think the support team has full authority when it comes to customer knowledge. No one knows better how to improve the product.

Support also plays a key role in copywriting. Every piece of text that appears on the site is usually written — or at least proofread — by someone from customer support. Our daily interactions with customers often help us to find the right words to improve their experience with our product.


Growth vs. Support

One of the common challenges with customer support — because it’s hard to scale by design — is facing growth. Especially if you believe in a genuinely human customer service, which is by definition the opposite of industrialized processes. We’re talking about an endless race between issues to tackle on one side, and the issue-multiplyer growth on the other side. It leads to the question everyone keeps asking me:

How do you scale customer support?

I have no secret recipe, but here are three leads.

1. The Best Support Is No Support

This is where my previous point on a tight relationship between support and product makes most sense.

Our job is to think about how to work less. Every incoming email is seen as a failure of the product. Behind every problem faced by a customer who makes the effort of contacting us, there could be a solution in the product (e.g. perfectible UX, poorly worded error message, incomplete help section, etc.). For every email we get, replying to customers and solving their problem is only band-aid. Our actual goal is to think about a sustainable solution so their problem never happens again, and won’t happen to others in the future.

To deal with growth, we try to keep our contact rate as low as possible. Don’t get me wrong: the goal is not to talk to our customers as little as possible. It’s about being there for them for all the edge cases where a human, with human skills, is actually required. In a nutshell: remove the cases the product should solve on its own, to free up our time for the (many!) cases where a human help is essential.

The main role of our support team is to take care of our customers by answering their emails, but ultimately, we need to improve the product so that they don’t need to contact us at all. In the end, this is our true mission: help make the product so good that there is no need to get in touch with us. The best support is no support.

Only the product can save us all.

On top of improving the product itself, we’re putting a lot of work on our help center. It plays a crucial role in making our customers autonomous. Once again, we’re not trying to avoid them; but their experience will be even better if they can solve their problem immediately and on their own, without having to wait for us to reply. We may answer very quickly, we’ll never be better than a search engine and static resources available 24/7.

The help site is our baby: since we can make edits ourselves, we improve the customer experience without relying on developers, who, on the contrary, are needed for most product changes. Our main focus is to make the help content as clear and straightforward as possible — which also makes it easier to find.

Those help pages also save us a crazy amount of time to consolidate knowledge. Once you’ve explained the same thing to 3 different customers, it might be time to write a help article about it. Next time, instead of re-explaining from scratch, you can just put a link to the corresponding help page. This is especially convenient for tweets and their limited length.

We also implemented a contextual help button, which suggests the most relevant help articles based on where you are in the app.

2. Internal Tools

We have two developers dedicated to the customer support team. Their job is to help us be more efficient in our daily job. And their roadmap is full until 2020.

They work on our internal tools (business back-office, tools to handle tickets, etc.) and track every useless clicks in our daily tasks. Their role is to automate everything that shouldn’t be done by humans, so that we can spend most of our time tackling customer issues, efficiently and autonomously.

This aspect is often neglected — pretty much like customer service in general — although it’s truly essential. Both for answering customers quickly, and facing growth without getting overwhelmed. Back-office and internal tooling developments are so underrated.

In addition to internal worfklows, we are constantly looking for productivity tools that can help us be more efficient. That’s how Alfred became our best friend.

3. Recruiting

Even if hiring is key in the support race against growth, I think it should be seen as a last resort. The least you hire, the more your team will get creative about scalable solutions to improve the product and internal tools. Some would say “hire when it hurts”.

However, building an amazing team has been essential to face growth and keep improving our customer experience. I won’t teach you anything by telling you recruitment is probably the most important topic in any startup. Our hiring funnel is tight because we do not compromise on the degree of passion of candidates, their writing skills or their level of empathy. Culture fit is our most important factor — we don’t take any risk when it comes to our values.

As our vision of customer support is quite unusual on the market, and requires many different and unrelated skills (writing, empathy, tech culture, knowledge of the train market, etc.), it leads us to hire atypical profiles. Our team is a mix of people with very different backgrounds. It goes from journalists to bike runners, including teachers, former Apple employees or psychologists.

Our Vision of Customer Support

Remember this picture describing the issues in a traditional customer service? Here is our vision of it:

In summary.

By the Way, We Didn’t Invent Anything

Even if I think our vision of customer service is quite different from most companies, we didn’t come up with it. Companies like Basecamp, Help Scout, Trello, Buffer, Automattic, MailChimp, Wistia, Slack and many more, have been doing this for a long time.

But our biggest source of inspiration is probably Zappos.

Delivering Happiness

When we think about customer experience, Zappos has almost become the textbook example. Zappos is a company that sells shoes online. Nothing crazy here. Something has made them truly stand out though. They were one of the first companies to radically place customer service as their number one priority. Their motto: Delivering WOW through service.

While many companies were spending their money acquiring new customers, Zappos took the opposite bet: they spent everything on retention.

That strategy paid off pretty well for them: they got acquired by Amazon for 1.2 billion dollars.

Tony Hsieh, their visionary CEO, made employee happiness a true management concept. He wrote a book about it, which inspired thousands of entrepreneurs. If you haven’t read it yet, you should!

It was a great source of inspiration for us as well. It’s the book I was offered when I joined Captain Train 5 years ago. So when this customer wrote us this tweet:

“Is Captain Train the French Zappos?” Best. Compliment. Ever.

It was probably the best compliment we could ever receive.


This article is an attempted transcription of a talk I gave in July 2016. Captain Train since became Trainline, the company that bought us that same year.

I am no longer working at Trainline since November 2016. Therefore I’m not in the best position to discuss the Trainline of today. This article only covers what I lived between 2012 and 2016.

I left Trainline to work on a new project — which I may talk about in a future article.