Des Traynor: Product Strategy – Revisited

Des Traynor, Co-Founder of Intercom, follows up his bravura 2013 Lightning Talk with an hour of precision focus on successful product strategy.

Successful product management requires frequent hard decisions and compromises, all based on a healthy balance of data and intuition.

In the absence of strategy, products are pigeon-holed into categories. They focus on fleshing out a feature matrix to win comparison wars with perceived competitors- who are doing the same thing! It’s a vicious cycle that yields the biggest, not the best, product. This fails; first you lose touch, second you lose relevance, and finally you lose customers.

This talk walks through an approach to product strategy focused on growing the value of your product – not simply the feature list. Very funny, pulls no punches and should change they way you think and act about product management.

Transcript

This is a continuation of what we talked about last year, but attending last year’s is by no means a requirement. Last year, some of you might have seen me give a lightning talk where I presented 15 slides, just to prove one of the points that keep coming up.

This is my deck from last year. As you can see there were in fact 15 slides. [laughter]

If anyone wants that deck, they’re more than welcome to. There are 180 transitions, but that’s not the point, you know. [laughter]

So, OK, for serious now. It’s actually wonderful to come to this event every year, ’cause the success of Intercom is punctuated by each boss, culminating today by me staying in the Seaport Hotel for the first time, which is a real great achievement in my opinion.

So, what a crazy time to be building product or managing product. Does anyone here own a product or run a product. I’m guessing it’s a fair chunk of you. OK, cool, so this isn’t the Intel conference. There was another conference going on in the hotel and I was like, shit, I hope I don’t go in the wrong room.

The world before social media looked like this. I think that’s a really nice cartoon ’cause it kinda reminds you of how much stuff we actually produce as software people. If you wanted Youtube, you looked out the window. If you wanted to Tweet you left a post-it down with just a quick 140 character summary of what you have to do or what you were doing. Google+ isn’t here yet. I’m sure they’ll find it somewhere. But more generally, software is replacing everything that we’re doing. When I was a teenager, this is what the world looked like. If you were, for example, listening to music, you were like, listening to music. You have ad vinyl on or a CD and you were sitting on the couch listening. Or if you’re playing music, you had a guitar. If you’re reading the news you had a newspaper. If you’re hanging out with friends, it was you with your friends. And, this is the world today. [laughter] What’s interesting to me is, that like, the people in this room and many more like us are responsible for creating the pieces of glass that people touch to do things. They could be doing anything, like, this guy could be reading a Twitter feed, or this guy’s shopping, hopefully for a better hat. This guy could be sharing a photo or fixing work issues, or any of those sort of things. The key point here is what Marc Andreessen said in 2011, software is eating the world and it is awesome people like, the people in this room, that get to pick what’s in the menu. Basically, most tasks are turning into software. Software itself is turning into little bits of glass that people rub their fingers on. Or soon, fingers crossed, pieces of glass on your wrist that you rub your finger along to do what you want.

It’s interesting to think about this because, you know, software is, you know, it’s gone so consumerized. It’s as consumerized as music is, but with one notable difference, if you ask either your children, or in my case nieces and nephews, what do you want to know about software? The first question’s always the same, it is actually, where does it come from? Which is really interesting, right. [laughter] It’s funny how, like, no one asks that question about, you know say, Miley Cyrus’ latest song or whatever, but as software, like literally, it’s a valid question if you tinker with stuff at 15 years old with a phone and you’re like looking at all these iPhones, you’re like, who does all this shit. It’s actually a fair question and it’s something I’ve been kinda looking at for a while.

Managing a product, product scope, feature creep, focusing a product and delivering magic

So, in this talk, we’re gonna talk about, we’ve got 5 points. It’s gonna be about, managing a product, product scope, feature creep, focusing a product and delivering magic. I’ve been looking at how do products make their way into this world and there basically seems to be 5 reasons and I want to thank a guy called Vankatesh Roa who has a great blog at RibbonFarm.com and he identified 5 reasons. I kinda couldn’t do any better than that if you like. It made total sense to me. You can either be a product focus company, in which case, your entire thinking is creating this holistic piece of software or this product and you have a clear vision of what it is that you want to create in this world and what it will do and how it will be used. Or you can be a customer focused company which is, in which case, the customer is your king and you will do whatever they will buy, right. You can be an Auteur, which is somebody who really wants to put your own particular flavor on things and there are some very successful Auteurs in software. You can build a product to scratch your own itch, in which case, the itch is your king and once you’re finished scratching that itch, you’re generally done because your passion wasn’t to create that next great billion dollar company. Your passion was to solve that itch. Or lastly, the thing we’re seeing a lot of at times like these when venture capital is freely available and people are chasing the next big thing, so often. You can follow patterns and patterns are interesting ’cause everyone kinda thinks when you hear of software patterns that they’re kind of a bad idea. And I don’t mean software patterns in terms of gang of four or any that sort of stuff, I mean, like maybe following patterns to produce products.

This is the icon for Tinder, so I’ve heard. [laughing] Yeah, look at you all playing dumb. [laughs] But, Tinder is a wildly successful product and now there are things like Tinder for professionals and there is Tinder for jobs and there’s Tinder for creatives, and there’s Tinder for dresses and there’s even Tinder for dogs. BarkBuddy connects dogs with other dogs. Isn’t that cool. [laughter] But, this is like, if you’re ever familiar with the theory of mapping problem solving, this is kinda like what their seeing, is sort of a core principle that makes sense and mapping it to money domains and this happens a lot. So, we have Tinder for X but we also have Birchbox for X. That’s the idea of like, stuff that delivers, you know, like food to your door on a regular basis. It’s what Joel Spolsky used to call Netflix for cabbage. There is also Airbnb for X. And Airbnb for X would be something that, where you have access to something where you don’t intend to use, and it’s a way for you to profit off of it. And lastly, this one like, Uber for X.

What actually happens for these things to get created is, someone looks at Uber and goes, holy shit, and 18 billion dollar company. I want some of that, and if they’re really, really, I wouldn’t say dumb, but really, really tactless I guess, they try and start an Uber, but the big advantage of Uber was, it didn’t have to compete with Uber, you know. Any second Uber is gonna have to deal with the fact that there is an Uber and it is worth 18 billion dollars. So, the smarter thing to do here is to abstract, if you’re familiar with how the Laplace transforms can help solve complex mild molecule charges, you abstract it, right. So, what is Uber? A car is an expensive device, that are rarely used by city living folk but it is used somewhat and it delivers no value when it’s not in use. And Uber, replaces the need for that car with an on demand car delivered by professional company that you can rely on. So then you think, how do we map that to similar problem demand. You say, well, what else is expensive, rarely used and delivers no value when it’s out of use, and then which of these could a professional company deliver on demand. And then you start to realize there are lots of, does anyone want to share any ideas, by the way?

Audience: Programmers.

Programmers, exactly. [laughter] Who was that? You are on to something.

But, I was thinking about, I was trying to solve this question loosely and I came up with something. For example, one of them is like, tuxedos or suits, right. Like, is it nice, basically, we all have a suit that costs more than $80 that sits in a wardrobe somewhere that only delivers value for the 4 weddings a year you attend. And, it turns out, there had to be, there’s already Uber for suits. Other examples are things like, let’s say lawn mowing, right. Every one has this expensive gear that you buy in order to manicure your garden, but it turns out there’s already one for lawn mowing too. The point being, I guess, there’s actually Uber for X, which I love. This is from, so I was at TechCrunch Disrupt last week and this was, this dude’s startup literally is, he gives you a platform to build Uber for anything. [laughter] And I can’t even say if he’s wrong, to be honest with you. So, it’s interesting, but when we talk about product management, you manage the product in accordance with the principles that create the product.

So, going back to our 5 points here, you have to understand what area it is you follow when you’re trying to manage your product, and this kind of massively informs your product strategy. It’s all well and good to say product strategy is about saying no, but that’s only true unless you are literally building what ever cus- you know, your strategy is we build whatever the hell customers want, and we’re gonna try and sell it and we have this baller sales team who can do it, then yeah, don’t listen to Des, you know. Similarly, if you’re following a pattern, if you are gonna do it like in- my favorite example of following a pattern was Slideshare whose simple pitch was, we’re gonna do Youtube, but for Powerpoint. And you hear that and you’re like, shit, that makes total sense, and now look at them, 120 million in the pocket, you know. Such is life. So, you have to actually follow the principles that cause you to create the product in the first place.

Let’s talk quickly about scope.

Scope is kind of the bane of most of our lives. It’s definitely the thing that causes most arguments in any office I would bet. Hands up if you had a fight over scope. I guarantee half of you guys are trying to expand it and half are trying to shank it. It’s great. We should all go fro a drink or something.

What I love about a scalpel is that it’s a single purpose device. It’s very easy to tell if it works. If I run my scalpel across your arm and we don’t see blood everywhere, it’s not a good scalpel. Simple. I don’t suggest that you test it that way, but it’s very easy to understand what it is. It’s easy to adopt it. It does one thing very well and it’s quick to build and quick to test. The opposite of a scalpel to me is, the Swiss Army Knife. So, Swiss Army Knife has other traits. It’s hard to market, it’s hard to adopt, it does nothing well, and it takes a really long time to build. People often push for this Swiss Army Knife approach and the danger of this is that, like, if, for example, I have your aunt for a dinner party and you say, hey, I’m gonna open a nice bottle of wine. And I say, OK, cool, here’s our bottle opener. You’r like, what the fuck. Similarly, if you’re on a date and you’re lady or guy says, hey do you have a toothpick, and you’re like, yep, you know. [laughter] Or if you get into a knife fight and you pull this out, the other guy will die of laughter, you know.

What’s interesting is, it is a valid approach to have this, we do absolutely everything particularly badly, but it’s all in one piece. It is a valid approach and there are suites that will do that, but what’s interesting is, you have to fight to get there. You don’t get it for free. If you actually want to be the successful, big bulky platform, all things to all people, nothing to no one, one size fits none, type product, you can’t disobey simple systems engineering principles, such as Gall’s Law. So, Gall’s Law briefly states that, any complex system has evolved from a simple system. You cannot go to complexity straight away. You have to fight to get there. Basically what I’m saying is, even if you want to build shit, you have to start off good first, you know.

So, what that means in terms of scope is that your first or your first feature release, and I can see scribbling and this is gonna go fast, but I will give you the slides at the end. Your first product, or your first feature release if you’re immature product who are still working on feature, is, has to be a solution to a real problem that exists today and it should be easy to build, test and refine and easy to explain and then easy to adopt. Most of the time when you go wrong, it’s because you violated one of those principles.

But we say simple a lot in this world, and one thing I realize is, there’s a difference between making a simple product and making a product simple, and that’s not a typo. There’s is a difference between making a simple product and making a product simple. Making a simple product, you have to decide where it starts and stops and you pick a small scope, ’cause people don’t use your product for a living. Sorry, that might be the case for like, 1% of you, but most people have other shit to be doing in their life. And, what actually happens is, there’s a whole workflow of things that have to happen and your product serves one piece of that workflow. So you might start over here and say, hey, I have to pay my team and then out the other side you might go, everyone’s paid! And there’s a whole chain of events that actually has to happen. You check the bank, review the product, bill the clients, check the time sheets, verify rates, draft payslips, issue transfers, tell team, and now everyone’s paid, right.

The naïve thing to do here is to say, right got it, 18 screens. Let’s just build them all, you know, straight in a row. That’s ignoring a lot of what we just talked about. You should start your product at the first place in a workflow, where you can add some new value. And new value looks like, letting people do it faster, easier, cheaper, in more ways or available in more places, and when you pick your starting point, you work ahead a transition from the previous step seamlessly. So, where do you stop? You stop when the next step is done by a market leader that you’re not competing with.

So, if the next step is, well at this point, they normally transfer money money for the tickets they bought, that does not mean you go and build Paypal. It’s not a good idea. Similarly, if the next step is, well, at this point, Johnny would usually review the film and see if it’s any good before buying it. Well, don’t go and build IMDB, just work with these people, right. If the next steps completed in lots of different ways, buyer beware. The reason I said that, is like, paying salaries is done in 160 different ways in 160 worlds, or 160 different countries, and people already have a working solution that’s unlikely for 160 different integrations that you’re gonna manage to deliver any new value. More than likely, you’re gonna manage to deliver 160 new bugs.

Lastly, if it’s something you simply just can’t innovate on. So if they always phone their customer at this point, let them phone their customer. Don’t go building VOIP, it’s not gonna work. To illustrate that, if we take this workflow here, again, the default thinking is, alright, I got the problem, A to Z. We just need to get everyone these screens. And you might even go and, say, let’s just put this on Odesk and ask a designer to design it and boom, we’ll all be doing well, we’ll all be millionaires, but this is not one product typically. These are actually 3 products. Often done by different people at a company, often done at different points that are different mindsets or different workflows. One person does not usually sit down A to Z and do all this, so there are natural break points and you can find them and when you find them you say, well this is the piece we do, but then when you have that, you say well, happens before the add clients? How can we make the add clients seamless? What’s the event that precedes this. Maybe we should import their clients from their project management tool or CRM or something like that. That’s how we make this stitch neatly into the workflow. Similarly, if we know the next step is likely to be, send invoices in this case, then how can we like, T it up so it will be even easier as well. So, how can we make it super simple for them to carry on their workflow.

Once you stitch it really neatly in, you can then explain if you wish. Once you’re already well woven in and strongly woven in, but to come from the outside is always gonna be tough. I roughly called this the the Scopi-locks principle, right. No scope too big, ’cause no one can adopt this. No scope too small, ’cause then you’ll be dismissed as a feature not a product, but you find one that’s just right. That’s the sweet spot. So, when you do this, one interesting thing to think about is, what Peter Drucker calls the cake principle. So, if you sit down to build a cake, one way you can go about it would be, let’s go and build a beautiful cake base and then we’ll go and build some filling and then later one we’ll go and do some icing, and then on our wedding day, we’ll find out if it’s any good. Obviously you can tell, that’s not ideal, because it’s not a good time to find out that you hate the taste of salt and salmon in a cake, right. Don’t do that one again. So, the point is, at no point are you getting any valid feedback about the cohesive whole. You don’t, you’re not actually testing anything along the way.

Another way to do this would be to build something like a cupcake and at this point you realize, you know guys, the salt and salmon on a cake, it’s not actually a good idea. Also, the oven’s broken. And then you’re like, OK, we have a good cupcake, now we’re happy. Now we’ll go and build a cake and then we can step it all the way up to a wedding cake. It’s a slightly different way to think about it, but when you’re scoping small, this kinda stuff can actually shave weeks off, because when you actually think about these things in terms of, let’s get all the messy assumptions that we have out of the way. People often think that products have requirements. They don’t. Products just have assumptions, the things you assume about the world and you wanna test all these really early on before you go on to the monolithic time suck project.

So, let’s talk about feature creep for a second. So, to make a product simple, you wanna say no to anything that isn’t your core feature. In fact, there’s only a few reasons you actually should be saying yes. So, you probably remember, you might remember this from last year. I was slightly better dressed back then, I guess. But this slide here, I decided to slightly reorganize, and here it is again. Except for this time, last year I just skipped over it. I didn’t actually talk about it. This year we’ll have a quick chat about it. So, there are, in my opinion, like a good set of, an acid test that any new feature must score straight yesses on, to actually be allowed into your product.

The most obvious one for me is, does it fit your vision. Vision is a complex thing. Every product needs a vision. You need to know what it is that you’re actually trying to deliver into this world. Sometimes the vision will be tied to the CPO or the product officer or someone who’s the CEO or the founder or the original entrepreneur who had the idea, but you will make these margin calls that no one understands where you’re like, how the hell can you say that. The visualizations are included, but for some reason, predictive analytics aren’t, and it’s just like, look, it’s a tricky one. You will have this fight all the damn time, right, but as the product owner, only you will know what it is that you’re trying to build here. Only you have access to all of the data and yes, it’s quite possible you could be wrong, but you’re more likely to be wrong executing something you don’t believe in.

A good test is, if you’ve never said no because of your product vision, then you don’t have a product vision. Then it is literally the force feeding of a product. Let’s just try one size fits all. It’s Joe’s Junkyard. All welcome. Will this feature matter in 5 years? Particularly in a world of new shiny and new jQuery effects and new products that you simply must integrate with and new Facebook widgets that you should support and new watches that you have to build on, it’s really, really tempting to basically chase your tail around in circles doing whatever matters today, not actually asking about what matters in 5 years time. The affect of all that is, you never deliver anything that customers value going forward.

Jeff Bezos has a great quote, which is, focus on the things that don’t change. It’s interesting, because most jobs, Tony talked about it earlier this morning, a lot of things actually don’t change over time. Here is 5 different people doing the same thing on 5 different generations of technology. All the exact same thing, right. And someone would tell this guy, don’t you know the book market’s being disrupted, or some shit. [laughter] Doesn’t matter, you know. He just wants to read. And of course, we see this and go, isn’t it so bad how technology’s made us all so uncivilized. Like as if this wasn’t a thing forever, right. [laughter] Another good question is, does it benefit all of our customers? There’s a couple of interesting things here. One of then is, the vocal minority of your customers are often the worse people to guide your product. I mean, you know, of course you can have that feature, I heard it twice already today. Twice already today, when you have a user base of 25,000 people, it’s not even approaching anything other than maybe an anecdote to tell in the bar sometime. It’s certainly not a reason to go build something. If you really wanna find out if something is useful, just because you heard it a few times, you still need to check if it’s useful. It still could be just a couple of people.

You can also ask other questions. What attributes do these people have in common? Are they all free users? Surprise, surprise, free users want more shit for free. Paying users want you to do what you’re doing, better. Free users want you to do more things. Only one of those things is gonna guide you to a successful product. I love this quote from Jason Fried, which is, we act not on the requests of your customers, but act on their behalf. To understand what it is, what’s behind their motivation, we search it, and then act on their behalf. Our customers occasionally point to the moon and we stick ourselves examining their fingers. We even go and build minimum viable fingers or pivot or whatever, but like, when you have users, you should have a formulized way for getting feedback regularly from your valuable customers and that’s what you should be guided by. Not the odd messy vocal minority of your user base. Even though they might seem to rhyme occasionally.

Another good question to ask when you’re considering a feature. This is great for avoiding, what I consider, valueless redesigns, where you just go and say we should redo this, why, ’cause I’ve been looking at it for 8 months and I’m kinda bored of it. That’s very rarely an actual good thing to do. A good question, does this improve on the current workflow? Does it compliment the current workflow. Does it let it be used in more ways. Or does it innovate upon the current workflow? If it does none of those things, it’s not worth shipping. Similarly, when you talk about engagement, one thing you always have to be careful for is, identify features that will actually get more engagement, not features that will fork engagement.

So, if you add something like, now you can select your group type, well, were any more groups created here just cause you can select your group type? No. Are you tying that into anything else? No. So have you actually changed anything? No.

Well, you have changed, you actually added a technical debt. Well done.

A good way to simplify this is, imagine in Atlanta, the Coca-cola company, there’s some project manager on diet Coke with lime, and he’s pointing at this chart saying, hey boss, total success, we started selling more, we sold none yesterday, we’re selling more today. Result. Somewhere else, another guys is like, why the hell is our sales down for diet Coke? And only the CEO has the vision of, well, we haven’t really done anything here except after buying all this fucking lime, and now we have all these juicers and now we’ve done all this advertising and nothing’s actually helped here, right.

Be careful you’re not doing that when you add features. It’s very, very easy. Like, people always push and say Facebook ought to make an opposite of a Like button, but it turns out if they do that, all that happens, is people don’t type, I’m sorry to hear that. So they actually get no new net benefit, right. Similarly, if a feature takes off, can you afford it? A real common thing to do is, we haven’t got a mobile app, let’s contract a mobile consultancy. Let’s get an Android app or let’s get a Window’s phone, actually scrap that, let’s get an Android app or an iPhone app or something and what you do then is you ship it. You’re like, hey and then your supporter crashes and you’re like, shit, we have to go on heroin now, can’t afford the heroin because there’s only small people using it.

Sometimes you can actually get features that might make sense but you simply can’t afford to support them yet. Another example would be like, concierge VIP onboard. Onboard will send you two free t-shirts. If you can’t afford it, it’s not a good idea. People often forget that that’s actually a requirement too. Can we design it so that the reward you get for using it is greater than the effort required? This is really important, ’cause, if you take this here, what you have is like, how hard is the task and how rewarding is the task. So, standard workflow is like, let’s say writing a project update in a project management tool. That’s a bit of work on behalf of the user and the reward is pretty high, cause you have to do that for your job, or whatever. The delightful features are often things like, say, filters on Instagram. It takes all of that to apply a filter, but the reward is pretty high. It outweighs it. What you want to avoid is the administrative and the gimmicky sort of stuff.

But there are other examples where this goes super wrong, for example, you can have a feature that makes perfect sense and delivers value, but you can fail to create it in such a way that’s actually worth using. One example of this for me is Google+ Circles. The idea is, you manage all your contacts and keep them grouped in circles. You set one of these up and then you have to maintain your circles. So, I meet a few people at this conference and I;m supposed to run back to my laptop and start dragging and dropping them into various circles. Yeah, he works in the same industry as me. No, he’s not a great friend but he’s kind of a friend. Is he a friend, I don’t know. But, the point is, the reward for doing that isn’t actually greater than the investment I have to do to keep these things up to date. And that’s really important because the concept itself makes sense to all of us, of course you update and share with certain types of people, but this does not make it worthwhile.

Lastly, if you can’t do it well, it’s not a good feature. This should be obvious. You’d be surprised and how often people go wrong with it. There are always gonna be areas where you have the main expertise and technical expertise and design expertise, and there will always be areas where you have no expertise, and occasionally it’s tempting in the goal to expand your workflow, to jump into pools that you literally have no sense belonging in. And you’ll do that and basically people will try the feature and tip toe around it and say, screw that, and now again, you just added a shit load of technical debt that no one uses. One interesting thing that falls out of all this is, a reason I hear a lot is something like this, But we talked about this feature forever, or, it’s month’s late, you know. Months late doesn’t mean anything. Deadlines you can’t afford to miss, result in product you can’t afford to ship. We can reduce this feature pretty quickly, like as if that matters, right. I talked about that last year. We’ve worked on this feature for so long. None of this matters. Literally, when you ship stuff that doesn’t work, none of that shit justifies it at all.

But, related to all this, you might feel like I’m arguing for saying no all the time, what I would say is, decisions are perishable, not final, never final, right. Every decision should have a best by date, right. You can decide, for the next 6 months I’m not gonna have the conversation about the iPad app, right. Doesn’t mean I’m saying no, it just means I’m saying, not now. There’s an important difference there because all the underlying things will change. Your business will change, your customer base will change, your feedback will change, all that sort of stuff. One thing to think about, is there are long and short term implications for every product decision you make. So be careful not to swap a long term ow for short term wow. The reason people smoke is they get a short term kick out of the expense of a long term cost. People are really bad at gauging that trade off. So if you’re someone like, Linkedin, you might think, hey, we can work out a way to boost engagement, for example, and it will result in something like this. That cartoon I love because it sums up a lot of where everything went wrong. I actually believe that in their metrics, it’s all looking good. They’ve managed to make a single click connect you to 16 people. Well done guys. But, hands up here if you’re connected to somebody they don’t know on Linkedin?

Right. This is your professional network, allegedly, OK.

You can now design and pitch your product saying, we’re like Linkedin, but for people who you actually know. How weird is that, right. [laughter]

The danger here is, you keep doing this, and you end up like one of these products that people have to take the crap out of just to be able to use. [laughter]

When I first saw this, I laughed like all of you. And then I gave it a while and said, that’s actually fucking useful. I’m actually gonna do that, cause everyday I pick up the remote control it’s like a foreign object in my hand that like, you know. It just makes sense. One thing to know is that, it’s hard to take a feature back. Undoing this is important, but it is hard, right. There’s this weird asymmetry in how people value features. If you take how a customer values a feature, before you ship it, they’re like, of course I’ll use that, I’d even upgrade for it at cents, well I can export the PDF right? And you’re like, OK, it sounds like a solid thing, then you ship it and they’re like, nah. You’re like, OK guys, we’re gonna end this life this feature and they’re like, WAAHH! Right. [laughter]

It’s hard to do this, but it is important, OK. People do kill products and features and it is important, right. Google has a whole, basically, graveyard of them, I guess. This here is an interesting one. Does anyone know what this is? This is the iPod hi-fi. It’s like the little dark secret that no one knows about in Apple’s history. It was alive for 11 months as a product line. They launched that feature at launch and it was absolute bomb. Basically, from the moment it went out the door, they went, let’s say nothing about this. So there’s two press releases that relate to it. Its launch and its shutdown 11 months later. That was literally it. I say he’s probably standing in front of one of the only four.

Obviously, iTunes Ping is another example. It was ill conceived but kind of fell well outside Apple’s vision, that’s my take on it.

All of this is to say this, which is, when you have feature that’s on the ropes or that’s debatable, you have options. You can kill it like we just talked about and take the pain and take the lumps and move on with a simpler product and less code base. You can try to increase the adoption. Get more people using the feature or you can try to increase the frequency. Get people using it more. It’s worth saying, shipping a feature means nothing and you should be careful how you reward and how you speak about it because it’s not about getting it live, it’s about getting it used. Usage is what matters. And what comes out of that is that your product road map, you should be able to categorize into one of these four. Are you improving a feature, and if so, our metric or our measure of success will be, are customers more satisfied. Are we seeing a decrease in support burdens. Are we seeing less don’t contact, relating to this feature. Or, are we getting more people to use a feature, so we’re making it work for more circumstances or making it work with more file formats or making it work in more environments or with more languages. Or, are we getting people to use it more, in which case, we should see people interacting with this feature more often.

And lastly and for any mature product, this should be like the rarer case, right, are we building a whole new feature for a whole new workflow. Fine if we are but that’s like, you have to realize, number 4 should justifiably fight with numbers 1, 2 and 3. Your first screen, post log in, is the one that every single one of your users use every single day, every single time they log in, right. Or your launch screen for your product or your app. Any improvement there is infinitely more valuable because of the because of the throughput. So, it’s worth remembering that.

So, focusing a product. This will have a small degree of overlap with this morning’s presentation from Tony, but focusing a product, what I’ve learned, I am much more motivated, I guess. Less about dogma and more about customer engagement or dollars or whatever. What I’ve learned is, if you focus a product on the job it does much sooner than the category it sits in, or the personas that it’s designed for, you have much better results.

And people will tell you to categorize yourself. Oh, so you’re in the analytics base, you’re in help desk base, payments base, intranets base, and what they’re trying to do is get you to commit to a certain set, a certain piece, and I love this quote from Ryan Singer who is the park manager for Basecamp, he is absolutely excellent, he says, identifying with a product category is outsourcing your strategy to the past, right. Identifying with a product category is outsourcing your product to the past.

And it is, because when you take something like, the social analytics category or, there are over 2 to 300 of these where these didn’t last well. What they all end up doing is, they start looking at each other and they start chasing each other and what you end up in is a product strategy with like, let’s just fill in the feature matrix, right. And what’s annoying about this is that, it’s not valuing what the customer wants whatsoever. It’s actually valuing what your competitors have and how you rank against them, which is not at all what users care about most of the time.

So, what happens here is you have a few different companies represented by columns and features represented by rows and every company wants to do the same thing. We want X’s in every row and we want one row unique for us, right. Of course, they all end up adding each other to those features so they all are adding increasingly less logical features to their product just to differentiate. Even the guys who I pick my teeth with do this, by the way, which is just ridiculous, right. [laughter] Like, they literally expect me to analyze my buying decision based on this, right. I was shocked to see that. So, there are interesting categories where this plays out pretty quickly, right.

One of them is photo apps, right, and the other one that we’ll see soon is messaging. The messaging space is kinda blowing up right now. But photos are interesting cause they’re… I found a quote by Chamath Palihapitiya, who is one of the core team of Facebook back in the day, and he pointed out 1008 photo apps have been venture backed. Over a billion dollars have been invested in this category, right. Anyone wanna guess at how many successful, positive or creative exits to date, that’s where at least the same amount of money came back? 8 is the answer. 8 out of 1008. Well done guys. But what’s actually happening here is like, it’s pretty sad, ’cause they’re all adding these extra features that aren’t particularly useful. You got filters, we got filters. You can tilt rotate and crop, we can tilt rotate and crop. And then they’re like, let’s go for like, let’s go with private sharing, and we’re like, woah, what the hell is private sharing, well it’s sharing when you put it on Google+ and no one will ever see it. [laughter] Another area you’ll see this, it is, this is just such a thing, like, if you look to say, the to-do list space, right, it is literally full of people who even have the same icon. It’s like, this is literally, this is what categories get to. It does not have the user’s best interest in mind. Similarly people will tell you to focus on a customer. They’ll tell you you’e customer’s a designer, it’s growth hacker or a busy father or busy shopper or whatever. They’re always busy, whoever they are. [laughter]

It’s true, like, here’s a busy CEO, oh as opposed to all those relaxed ones, you know. And they’ll give you demographics and psychographics and all that sort of stuff that will kind of pack in with as much content as you want, and they’ll give you these target things, like our customer’s a young professional. This is the face of young professionalism according to Google image search, by the way. But like all of these things correlate with your customers, maybe, or they correlate with a lot of your customers, but they’re actually, none of them are reasons why any of them actually are your customers, and you always have to be careful not to confuse those two. Wherever I see firemen, I see fire, therefore, obviously firemen start fires, right? And in fact, the more firemen, the bigger the fire, so it must be those guys right. [laughter] Of course it is right.

That’s literally the log in you apply, when you look at your customers and say, OK well, seems to be popular with 18 to 25 year-olds, so anyone who’s 18 to 25 must be compelled to just wake up and buy this product. It’s not actually true. If we take this, say, young professional we were targeting earlier and we say, let’s say that person’s hungry and I’ll give you like, 5 pieces of information, they’re an iPhoner, their salary is 30 to 40 grand, they’re left handed, they have small family, they drive a Vauxhall Astra, you know, I can keep going, and I say the person’s hungry and so is this the right thing for the person?

Well, who knows, right. I mean, it looks nice, but is it right? I don’t know.

What about some, like, Indian food? Again, it looks nice, but I don’t know. What about a salad? Always bright and fresh. I know I’m showing you food and you’re all still full, but deal. [laughs] Or maybe something like this, right, everyone’s favorite. The point is that, you can’t answer that question for me because I…(that’s a burger…oh… welcome to America Mark) [laughter, applause] (the land of the free)

So, the reason you can’t tell me which of these is correct is simply because, products match problems, they don’t match people. The more you know about the problem, the more you know about the solution. The more you know about the person, go have coffee with them, you know. It doesn’t actually help you design a solution. He produced this diagram that I really like. It says, adding more context refines the situation, the more you can refine the situation, the better you know the solution. So if all you know is something like, I have, I want something to eat, right, and it’s a very blurry concept up here. Well, what if you know you’re in a rush, well, immediately, so I gave you 5 pieces of information, didn’t help you at all, I told you the person’s in a rush, we can start cutting stuff off the list, right. Michelin star restaurants are gone, for example. Anywhere with a long queue, gone. But, we can go further, I’m in a rush, I’m starving and I want something to eat. We can probably take that salad off the table, let’s be honest. No one makes friends with salad. I’m in a rush, starving, need something to eat with one hand on the go and I’m not sure when I’m able to eat again. And then, as you can see, they progressively, the understanding of the situation gets better and the situation defines the solution, which in one case could be pizza, right.

And what’s interesting is, this has nothing to do with a young professional, because, you know, even President Obama eats pizza, right. And it looks like kind of a bad ass dinner too. [laughter] But, more than President Obama, we also have, like, J-lo, we have Hillary Clinton, we have Gweneth Paltrow, Ryan Gossling, Jay-Z, Madonna, everyone eats pizza, right, because they all have this problem at some point. Notably, you can only look good eating pizza if you’re Ryan Gossling. [laughter] Pro tip. Seriously, like that must have been a photo shoot, that’s, I can’t imagine, like.

The key point there is that people who have the job is always bigger than the market that you think you’re trying to turn, so these are not all in the young professionals category, but they all have that problem. The problem is bigger than the demographic. Therefore, the job is the unit of analysis, not the person. So, to play this back to software for a second, I’ll give you an example, and also, I guess, to show that I mean what I say, and example from Intercom, which is a company of which I’m a co-founder, one of the features we added to Intercom in the sort of, wouldn’t it be cool category was a map, right. We thought this would be cool. We could see all the users all around the world on a map and you can filter them down and have a look and you can get to know who the users are. And we shipped our feature because we thought, wouldn’t that be cool, and you’re all probably thinking, that’s pretty cool. But we’re wondering why the hell do people use it. And it turns out people use it if they want to impress somebody at an expo to say, hey, check out our global business. Or if they’re pitching it on incubator, they show us the user map and try and show, hey we’ve got customers all over the world already. Or if they wanna show, hey look, our product is Freckle and we’re super popular. They realize, well, we realize that this tool is used in all of those cases.

It’s not a map, it’s a tool for impressing people, OK. If we think it’s a map, we go and build geographical accuracy. We build precise filtering, clustering of segments. We build drag to zoom, we build all that sort of stuff. Accurate scales and distances. We really go deep on cartography, right. But when we know what the job is, we’re like, well what actually does this job better. Well, a more beautiful map, maybe an animated map, maybe a full screen map. Maybe let you download the map as a keynote slide. Maybe those things would be useful, right, ’cause we know what the people are trying to do, and what you realize is, a worse map does a better job, and that’s really interesting to me.

Never confuse the category you’re in with the value you deliver. Customers only care about the value you deliver.

Weather websites get this wrong. They’re obsessed about being in the meteorological category, right, so they give you all this figure like, precipitation and humidity and wind and pressure, but that’s not what people actually want to know. People want to know, like, is it raining? Do I need an umbrella? What should I wear? [laughter] Surprisingly, no one knows, no one knows how to read humidity figures predicted over the next 14 days, if that ever was useful. So, what you realize is, like we did with the map, like these guys realized when they looked at how bad weather maps were, is that before you release, you design for the behavior that you expect. Once it’s out there, you design for the behavior that you get.

Peter Drucker has a wonderful quote on this. To reborrow the drill example for one second, the customer rarely buys what the company thinks it’s selling.

Last up is, delivering magic.

Homestretch now. So, you’re probably wondering why there’s beans here. That’s fair enough. Such a discerning crowd, business and software people. So, the reason there’s beans there is ’cause, here’s an interesting thing to think about, right. In 1901, the year Heinz launch, what did they do? Put beans in a can and they sold them. In 1955, what did Heinz do? Put beans in a can and they sold them. 2014, guess what they’re doing, right. The core output of a beans company, is beans. We do not make beans in this industry. We are not a beans company. The core output of a technology company, is innovation, and there’s a big difference, and that’s attributable to Marc Andreesson and I slightly paraphrased it.

The point here, is that, you have to continue to innovate. You’re actually delivering innovation and letting customers do a job. Your product will change, it must change, it has to change, that’s what you do. Your product is not as simple as a tin of beans. And the entire underlying landscape, if you like the beans architecture, is all shifting, what you do, tectonic shifts, mobile, touchscreen, etcetera, but it will just keep going. So, you need to be continuously innovating on the ways in which you can deliver magical experiences for your customers. I just have a few points in this. The reason I say magic is ’cause it’s this Arthur C. Clark quote, which I’m sure you’re all familiar with it, any sufficiently advanced tech, is indistinguishable from magic. The old magic of our world, magic where a lot of us grew up, is now basic expectations. That’s simple Kano model stuff. Everything that was once novel, becomes standard, sort of table stakes, the least you could do. Saying your product is easy to use in 2014 is like a chef saying our food is edible. [laughter]

Saying your product is mobile, so, if it’s not it wouldn’t have taken off. Your product’s cloud based. You can access your files from anywhere. People are like, are you serious. All this stuff, if you go back to like, archives that are, look at your favorite web products and look at their marketing pages in 2005. I guarantee you, this is all the stuff they said, except for mobile. But everyone just pitches against that, ’cause that’s what the magic was at the time. The new magic that I’m seeing, the one that’s really underpinning a lot of success of products these days, are these three things here.

What we’re seeing is the uber-fication of interaction, automated input and ambient insight and we’ll step through these. So, uberfication is, a phrase coined by David Sacks who is the founder of Yammer, and he calls it like, the slimming down of application interfaces into a simple push button experience and does one thing. He calls it the next consumerization, which is no small thing, right. So, it’s interesting to think about that, because it can pass you by a little bit, but here’s what uber would’ve looked like in the old world. This is uber in the old world. It’s a web form. Select your type of car, your location, cross street, phone number, pick-up, how do you wanna pay, you know. That’s like, that was table stakes, and I guarantee you a lot of your apps still look like that today, right. But here’s what uber actually looks like in this world. Press a button, right. Actually has a car arrive but it’s basically press a button.

Here’s what ordering a pizza looks like in the old world, configure what you like, multi-step wizard to Dominoes. Here’s the latest in pizza technology, Push for Pizza, anyone heard of it? It’s amazing, right. It’s one button, you push it and you get pizza. [laughter] If you don’t want pizza, you don’t push it. And lastly, and I’m sure you’re all getting the pattern here, what do you do if you want two pizzas? You push it twice. Right. Done.

Next problem, right. Another favorite of mine is, you probably heard of this one too, Yo, right. Yo is a single digit, single piece communication where you can send somebody a thing and all it does is say, Yo. That’s it, yo, right. And the weirdest thing is, somebody yo’ed me at 1:44 p.m., and I looked, it was one of my friends and I knew exactly what it meant. It meant, best of luck. And I’m gonna get a yo from people when I get off the stage and I know exactly what they mean too. You think I’m fucking mental right now, I get that OK, but there’s a lot in this. It’s single piece communication, fully contextual, right. Any managers here? Don’t be afraid, c’mon. You all think this sounds ludicrous, right. Aaron Levie had a wonderful Tweet about what enterprise Yo would look like. Yo for the enterprise is just saying, hey, did you do that thing? [laughter]

Now, hands up if you spend most of your working day doing that, right. That’s the point, right, there is more to this than meets the eye. So, the fact that it’s a push button experience is not to be sniffed at. Obviously there’s been some good gags that have fallen out of this, for example, this is the future of Yo. [laughter] But, for real though, these single swipes, taps or clicks, whatever, they can get us a date, they can get us a car, they can get us a restaurant, they can get us a movie, they can get us a hotel, they can get us a job. You can swipe to pick your apartment, your roommate, even find yourself a dog. This is your life and you’re living it one swipe at a time. This is the world we’re creating right now, and get with it, sort of thing, is a, what I realize is that, any design isn’t really complete until you’ve reduced frequently recurring problem into single swipes, taps or clicks.

That’s literally what we’re doing right now. And if you don’t, basically, the future is going to be built by the people who get that, because the people who say push buttons say, well my product could never be reduced to that. Uber just doesn’t give a shit. With uber, it’s just, you’re already getting it or you’re not.

The trick to this is, be really, really clever about your defaults. Be really, really clever about your onboarding, get it all set up so that one push literally does get you a pizza. The second key part of magic for me is automated inputs. So, the Web 2.0 era, which I’m guessing a lot of you kinda moved through and struggled with, as did I, was a world where, you basically enter data into novelty sized text fields and then on the next screen, you saw the same data you just entered, except for now it’s back at you. Sometimes they put a rounded font on it. Like they sort of take and expense tracker 2.0 and it’ll be something like, put in all your expenses, guess what, here’s all your expenses, you are like, got it. Already had those, so thanks. The new world has you authorizing the data source, and there’s a real difference, cause this is push button magic, right. This let’s you create a live connection between the stuff you need.

So, Tripit, for example, say, connect your Gmail once, and you never have to tell me about a single trip again, I will know and I will tell you. That’s magical, right. The value you get out of that one click is huge. Barometrics is sort of revenue analytics for Stripe, you literally click and authorize Stripe button, and then you get revenue analytics going backwards and forwards for your company’s history, from one click. ‘Cause you authorize the source. You never have to update it. You don’t log in to give it the latest figures. We follow this principle at Intercom. You install Intercom once, and that’s it. You get this live user list forevermore. Anywhere you’re asking users for input, first ask yourself, does this data or this content exist anywhere else, and if you’re not pulling it from there, one of your competitors will, and that’s how you do it.

Lastly is ambient insight.

So, real quick, when I was growing up, one of my friends lived next to a library and I was really jealous ’cause I would’ve loved to live next door to a library. It was cool, you had access to all the books, computers there and all that sort of stuff. One day I called over, he was not the sharpest kid, I called over and he was like, you know, drinking paint or licking hammers or something, whatever it is that not so sharp kids do, and I realized that basically, living next to a library doesn’t make you smart by default, right. Just ’cause it’s all there, doesn’t mean you’re in some way, smart, OK. And yet, when we build our products as being these siloed shells full of data that you have to come to and root through to find value, we’re assuming living nex- or we’re selling the value of living next to a library, and there’s a big difference. The old world had you going to each of these tools everyday. If you wanna get any value out of them, you have to literally spend time, invest time going into the tool, root around, what’s here, what’s there, and come all the way back, and you do that for every tool. Guess what, people don’t do that, ’cause they have other things to be doing, like staying employed, and because of that, their appreciation of the value of this tool diminishes very quickly.

In the new world, tools bring value directly to you. The reports and insights all come to you. And, we follow this principle, I also have a lot of products of late. For example, every morning, we tell you who signed up for your product, ’cause people need to know that, so we tell you. Everyday mention tells you, who’s talking about your product and what they’re saying. Everyday, gauges tells you about your homepage traffic and your most refers. You’re always in touch. They’re always delivering value, regardless of whether you used them. Your product is only ever a small part of someone’s job. So you design it as a system that shares valuable information, not as a destination that you have to visit just to get value. And when you see things like card based design, you see things like notifications and interactive notifications being built into new operating systems, you see things like Apple’s watch and all of the various Android watches, and Google Now, you realize that real value is created when value, when it comes to the user, not when they have to go hunting. And that was delivering magic, which means we’re on the final slide.

My name is Des Traynor, my company is Intercom, my blog is insideintercom.io. If you think this talk was useful, the blog is better. If you want the slides, drop me an email. Thank you very much. [applause]

Q&A:

Bernard Gunther, looking for the next company, between companies at this point. You say, push notifications. Everybody’s trying to push everything to you and their stuff is the most important so they tend to flood you. How, well user, how do you cope with that?

A: So, is this in relation to the last point about bringing data to the user, right? So, I think that’s a, it’s a really good question, but it actually asks a deeper question, which is, if you push junk to people, yeah, they would disallow permission and they’ll unsubscribe from your emails and they’ll stop using your product. About the way I think about it is, when you set out to build a product, what is the most valuable thing you can deliver to a user? And you start, like, I really believe in my colleague, Paul Adams, is doing a lot of work on this around, but I really believe the future of design will start with notifications. Most people who grew up in the world where I grew up are like, when you say we’re building it up, what we do is we draw a big rectangle cause we’re already thinking web app and we’re thinking tabs and we’re thinking left bar, we’re thinking search, and we’re thinking all this sort of stuff, the actual, in my opinion, the smart way to think about these things is, what’s the touch point first. So, what is this little card that I would want on my phone everyday. So if it’s an accounting app it might be, my bank balance, my spending trajectory, the fact that I’m running out of money or whatever. But you want to start from that piece backwards so, absolutely if you’re pushing junk you’ll fall like all the rest of them do. But you actually have to start your design from what is the update first and work back from that. I hope that’s, I know it’s not super actionable but that’s the aspirational answer I’d give.

Q: Dan Flannigen from Percussusion Software. One of the things I’m a big fan of, like the jobs, framework as well, but where I’ve struggled is with marketing organizations that love their personas and the motion based and personality based type of sale. How have you had experience kind of translating the one into the other?

A: Well, I think, like, I’m very much sympathetic to the situation you’re in, but a, what I would say is that, jobs does not necessarily tell you how to reach your target customers, I mean, guys here can jump in if they want, but like, it’s useful to know, that 78% f our customers are, found us via an advertisement on a tube station, or something like that. That’s like, good data, and, so you could build a persona that embodies that characteristic and that’s really useful to know how to reach customers. Like, marketing does not build product, marketing is there to like, it’s actually there to get a product out there to the world. And it is really useful to know attributes. I’m not saying personas are useless, I’m just saying that I wouldn’t focus my product around them. So, I think they hold a lot of value when it comes to identifying, certainly how to reach your audience but often times, it can also help you find things like, does the high end of this audience exist, so, for example, are you trying to sell high end leather six finger gloves, right, the high end of that does not exist, OK. Maybe there’s a market for six finger- but the certainly not an enterprise market out of it, right. So I think that personas carry a lot of useful information for the marketing department, who I think are the people who push for it. I don’t think it’s actually at odds with the jobs idea, I just think that products build that jobs and products market according to personas, I’m happy with that world, personally.

Next Events

BoS Europe 2025 🇬🇧

🗓️ 31 March – 1 April 2025
📍 Cambridge, UK

Spend time with other smart people in a supportive community of SaaS & software entrepreneurs who want to build great products and companies.

Save your Spot

BoS USA 2025 🇺🇸

🗓️ To be announced soon
📍 Raleigh, NC

Learn how great software companies are built at an extraordinary conference run since 2007 to help you build long term, profitable, sustainable businesses.

More Details