8 Mistakes From a Startup That Didn’t Quite Fly

In 1943, a Jewish-Hungarian mathematician called Abraham Wald told the US Military how to do their jobs properly. The Generals tasked with reducing the number of planes being lost to enemy fire were trying to decide which parts of the aircraft should be reinforced. The weight of armour plating meant that it couldn’t be applied everywhere, so they started analysing where planes were taking the most damage.

Images like the above seemed to show a consistent pattern, so they recommended that these areas should be reinforced — until Wald pointed out that they’d made an error. The aircraft being analysed were all ones that had survived: what the above image actually showed were the areas where a plane could take damage and still make it back home in order to be analysed. Rather than reinforce these areas, priority should be given to the areas not damaged in the above diagram — as the planes which had been hit in these places never made it back to be included in the sample.

Which is a roundabout way of saying: I’m Luke, and my startup crashed whilst taxiing to the runway due to my complete inability to run it properly. I did my best, reading everything I could from the titans of industry in whose steps I was boldly following, but it gradually dawned on me that I might be making the same kind of mistake that Wald had flagged 80 years ago.

All of these books had been written by people who had succeeded. The businesses I was trying to imitate were all ones who had flourished. Isn’t there more to be learned from the ones who didn’t? Surely those are the companies making the mistakes that you really don’t want to repeat — the fact that they’re no longer around would show just how dangerous those mistakes can be.

And so, 14 months after I first set sail on the good ship Filo to boldly revolutionise the world of UX Research Repositories, here’s my list of unbelievably catastrophic errors made whilst successfully building an app with zero users.

SPOILER ALERT: this is a long list. Apparently my ability to make mistakes exponentially exceeds my ability to run companies.

Mistake 1: I’ve worked in tech for 12 years and don’t know what an MVP is

I mean, I know what the letters stand for. Plus I’ve read Eric Reis, and I’ve helped build loads of products in my career. There’s no way I can get this wrong, right?

Wrong. A Minimum Viable Product should only ever be one thing: a button that does nothing except record that it’s been pressed. That’s it. If you have anything more than that, it’s not an MVP — it’s an iteration of your product.

An MVP only has one function: to test whether an idea is correct. If you’re a startup, the first idea you need to test in this way is the belief that people want the product you’re going to build. The easiest way to test that? Tell them what that product does, and ask them to click a button that says “yes, I want this” — preferably whilst also leaving you their contact details.

“But wait!”

You might now be saying, desperate to repeat the same mistake I did.

“Surely we need to build at least a bit of that product? Otherwise how will people know they want it? You’ve forgotten that it needs to be viable.

Well, maybe. If all you’re testing is an idea though, then “viable” simply means “viable enough to prove that the problem exists, and that you have a potential solution to it”. It doesn’t mean that you’ve actually solved this problem. That’s what your product is for, not your MVP.

Fail to make a distinction between these two things and you’ll be well on your way to repeating the same mistake I did. The problem with equating “viability” to “functionality” is that you’ll always be able to argue that just a little bit more of the latter is needed to ensure the former. I did this every day for about 9 months, consistently beckoning my long-suffering co-founder along a never ending path called “one more feature”.

I’d fallen into an Escher-esque nightmare of recurrent logic: if people weren’t using the app, this just proves that the next feature that I was planning all along anyway is going to be the one that finally tips the scales.

This is clearly absolute nonsense — I’ve torn my hair out in the past when companies I’ve consulted at have made the same argument to me — and yet I convinced myself it was true. How? Why? The answer lies in some of the pressures unique to the startup environment — and a whole bucketload of other mistakes which these caused me to make.

Mistake 2: Marketing is roughly 10,000 times more important than your product

This will sound like absolute heresy to some of you — it did to me too, which is kind of the point. I know how to build products, because that’s been my job for a long time. Comparatively, I know absolutely nothing about making people aware that these products exist. I’ve never done that job, and I’ve never had much contact with people who do.

What then is more likely to prevent my startup getting off the ground? The thing I’m quite good at, and have a decade of experience in? Or the thing which Google Docs just informed me I can’t actually spell? Somebody with the opposite career path to me — lots of expertise in marketing (one “t”, apparently), but comparatively little in product development — could probably invert the statement above and the premise would still hold true. Your strengths will take care of themselves — it’s your weaknesses you need to worry about.

Fail to acknowledge this and you’ll end up frantically bolting ever more armour onto your plane’s fuselage, concerned that the minor scratch found by the last test is dissuading passengers from stepping aboard. All whilst a slightly befuddled mathematician stands behind you wondering when you plan on building the wings.

If you’re like me, then this all goes back to that button above — the “Yes, I like this” button that proves you’re onto something when pressed enough times. How are you going to get that in front of people? Put it on LinkedIn and hope for the best (marketing strategy ©Luke Kelly, 2020)? OK, let’s be optimistic and say 250 people like the post where you announce your idea. Because you’re actually the next Steve Jobs, an astonishing 100% of those people follow the link in the post to your website and register for the waitlist for your beta.

And now? What’s next? Where do the next 250 people come from? If you’re thinking “well obviously they love it so much that they tell their friends, family, and random people in the street all about it”, then you should probably pause for a moment here. When have you ever done that? When have you ever seen a post on social media, and got so unbelievably excited about registering for a mailing list that you immediately drop everything and start convincing other people to do the same?

If somebody had tried pointing this out to me, I’d have said “well that’s why I need to build the product right? So that they use it, get value from it, and tell other people about it?”. This might be true — but then what you’re building is no longer an MVP. Follow this logic and you’re well on the way to repeating Mistake Number 1: it will always be the next feature that gets people’s attention. Right up until the point where you run out of money.

Another problem with the above is that you’re committing from the outset to a rather bold marketing plan: deriving exponential growth from a fanatically devoted cohort of core users, who you’ll make so happy that they’ll consistently push more and more people to your glorious product. These new people will also be blown away by something you’ve built in 3 months with no money, and the process will repeat itself exponentially.

I mean, it could happen. Some companies manage it, after all. I’d hazard a guess though that such companies aren’t reliant on a dictionary to ensure that they can write the words “Marketing Plan”.

A more realistic strategy might be to look for less fanatical devotion but from a larger, more consistent audience. A constant, steady stream of people finding your potential product and saying “this actually doesn’t look terrible, here is my email address”. The key here is that if you can’t get people to look at an idea — just look at, and maybe put their details down — how are you going to get them to look at a product? How do these people know you exist?

I managed to successfully avoid addressing this issue by instead focusing on making Mistake Number 3.

Mistake 3: nobody is interested in your “vision”

Slightly harsh perhaps but if you’ve made it this far in the article then it’s probably because some of the above is starting to sound horribly familiar, so some tough love might be in order. Vision is perhaps important when talking to VCs. They’ll be slightly more indulgent of your claims that “I’m doing for UX Research what Microsoft Excel did for spreadsheets” (ahem). After all, they’re speaking to thousands of cranks like you — and even funding a few dozen of them — all in the belief that one actually is the next Steve Jobs.

Even this indulgence will end abruptly at the words “market traction” however. They might agree with every single world you say about where your product could be in three years — Sequoia definitely did when I cold-emailed them our pitch deck which contained the actual words “this changes everything” — but if you can’t show them something compelling from the last three months, then the conversation is over. And the only thing that’s compelling is numbers.

If people aren’t at the very least fighting like rats in a sack to breathlessly sign up for your waiting list then you have a huge problem, and your “vision” won’t solve it. The mistake I kept making was thinking that other people could see into the future. Because I could see what people could do with Filo in 2–3 years time, I thought they’d be queuing up to start using it. But all anyone else sees is what your product can do in the here and now. If it doesn’t solve a problem for them right now, they’re not going to use it in the hope that it might do so in the future.

As each new feature failed to drive up the number of active users we had, I kept thinking that it was because people couldn’t yet see how these features linked together to deliver something truly special. That’s not how it works.

A truly successful startup solves a big problem, which it breaks down into lots of smaller problems. That’s what VC’s like Sequoia are looking for, and that’s why they’re definitely about to start returning my phonecalls. But partially solving each of those small problems does not equate to having partially solved the big problem. It equates to an app with loads of features which people don’t use because they have no idea what you’re building.

The scales won’t fall from their eyes the moment the Next Big Feature lands. They won’t look past the compromises and hacks you made on the last release purely because they’re so excited by the next one. They won’t suddenly tune into your Big Brain vision for the future of all qualitative data because you’ve built a Kanban board. The “vision” means nothing. It only has value in one, very specific circumstance — a situation which will only arise if you avoid Mistake 4.

Mistake 4: solve one problem perfectly before you even think about moving onto the next one

This might sound obvious, but the word “perfectly” is the one which should scream at you in the above sentence. What was the last product you used which you’d describe as “perfect”? It’s probably the last one you recommended to somebody else, which in my instance means an app called Pastel.

I love Pastel. I tell everyone about Pastel. At one point I tried to convince my dev team to replicate the functionality of Pastel because I’m relentlessly committed to making Mistake Number 1. All Pastel does is let you annotate a webpage. That’s it. Put a URL in, and you can make notes on different parts of that webpage that other people can read.

Whilst in the process of completely remaking our website for the fourth time in the space of a year — not a mistake by the way, that’s clearly next-level startuping right there — the designer who was helping me build it asked me to use Pastel for the revisions. I did, because it completely solved this small problem we were having: how best to communicate several distinct pieces of information about a website.

By contrast, the app I was building was proceeding to address the slightly more ambitious goal of “fixing all human communication”. I’d grudgingly been pulled back by my despairing co-founder to the slightly less grandiose “completely change how the field of UX Research operates”, but — weirdly — this too always seemed slightly beyond our reach.

Personally I think it’s because we haven’t launched enough features yet, but there’s also an argument that we could perhaps have followed Pastel’s example: solve one problem so well that people recommend you to others.

This means that however grand your vision is, you need to try and find the most incomprehensibly small element of it that exists in its own right and solve that first. Even if most of your users have no idea what you’re planning long-term. They don’t care — they just want you to solve that problem.

For us, that should probably have been something like “shared drives don’t tell you enough about the contents of what’s in them”. A problem many UX Researchers face is that their work is locked inside various different reports. Putting these reports in a shared drive so that people can access them — itself a solution to the problem “people don’t know what research we have” — doesn’t help that much, as you either need to describe the contents in so much detail that you’re effectively writing a new report, or people have to manually download and check each one to see what’s inside.

That’s a problem. It has nothing to do with conducting research; analysing research; making research more useful in strategic settings; making research more efficient; or any of the 10,000 other problems I’m utterly convinced Filo can solve — but it’s a great candidate for an MVP. Show people an image of the most immaculate looking interface ever, and see if they click the button that says “register”. If so then you’re away! Apologise that the button did nothing, but tell them you’ll let them know in two weeks when it works — they’ll forgive you when you solve their problem.

Can’t get enough views to even test it? Then you made Mistake 2. Do people look but less than 50% click? Then it’s a good job you didn’t waste time building it — change the design and try again.

This, finally, is the actual definition of an MVP: something that you can throw away without a second thought if it’s not working. An iota more effort than is absolutely necessary will see you torpedoed by the “sunk costs fallacy”: if results aren’t quite what you expected, you’ll convince yourself that a little more effort is needed rather than a change of focus or approach. Confusing the need for more effort and the need for a different approach is disastrous — and I made this mistake daily for almost a year.

Tim Jackson makes this point well: when you’re just getting started, you should measure your runway in terms of how many experiments you can run — not how much of a product you can build. If you can run experiments more quickly, you increase your chances of hitting gold. And the best way to do this is to keep your MVP as simple as humanly possible so that you never feel bad about throwing it away.

Imagine putting a different image of a potential product in front of 100 people once a week for 6 months, and iterating it each time based on the results. Now imagine the product you would build at the end of that process having tried so many things and collected so much feedback — even if you only had a month’s worth of funding left to build it. My guess is it would look a bit like Pastel.

Now imagine trying to do the same thing, but with a product people can actually use. How many could you have built in the same time? Maybe three? And would you really have thrown each of these out and started again just because one of the tests wasn’t that promising? Or would you have convinced yourself that a small tweak to what you already have was basically the same thing and persevered?

Note that even in this scenario where you’ve properly defined the scope of what you’re working on, marketing is still exponentially more important than product. The image of the interface could be bashed out by a half-decent designer in no time — especially if you’ve done your research and properly defined the problem you’re asking them to solve. If you can’t get that image in front of enough people to meaningfully test it however, then you’ve made Mistake Number 2 again and should drop everything in order to resolve this.

Or you could follow my lead and distract yourself with Mistake Number 5.

Mistake 5: people don’t give feedback on a product that they’re not using

When was the last time you took the time to send some feedback to a company? For me it was when the world’s greatest wrestling-themed Yoga app stopped working and I wanted to know how to make Diamond Dallas Page appear on my phone and start shouting at me about stretching again.

People give feedback on things they’re actively using. If they’re not using it — for whatever reason — they’re very unlikely to take the time to tell you why. When was the last time you signed up for an app, hated it, then took the time to explain to the people who made it the exact ways in which they had disappointed you?

It’s all well and good to say “we need to test this with our users”, but if you don’t have any users, what are you going to do then? This is yet another way in which your ability to build a product is exponentially less useful than your ability to build a prospective audience. 100 people willing to look at an image and say “yay” or “nay” is much more useful to you early on than an ability to build something those people might like if only they knew you existed.

We found this out the hard way when we started applying for funding. We were quite proud of what we’d built. We’ve both been working in tech for a long time, and knew from experience that the amount we’d got done was pretty gosh darn impressive for two people. We thought this would be enough to establish what trajectory we were on — especially as we could show how each of the pieces fit together into something truly revolutionary a bit further down the line (see? Mistake 3 can sound awfully logical at times).

We got quite far with quite a few different investors. Each thought we had a strong team, a decent long-term vision, the makings of a good product… and absolutely no path to market whatsoever. Personally I was quite relaxed about the fact that we had no users — I mean they’d surely arrive once the next 17 features were launched next year — but even I started to see their point after a while.

Imagine two people come to you with the exact same idea. One is great at building things, and they’ve built half of the product they need already. It’s obviously a bit rough around the edges, but you can see that this is a Steve Wozniak level genius here: this person could absolutely finish building the product they’re showing you given enough time and resources. They can do more in a week than a team of 10 do in a month — they’re that good at building things.

The second person who comes in with the same idea is Steve Jobs. Steve Jobs has 100 million Twitter followers, and answers any questions about his product by saying “I am Steve Jobs”. He then shows you a tweet he made last week that is nothing except a giant question mark and the word “SOON” which has been liked 4.5 billion times.

Who do you give your money to? You can back first Steve, and hope he uses the money to hire the expertise that will somehow turn him into Second Steve so that enough people can hear about his product. Or you can back Second Steve, and trust that he’ll hire someone to build at least something before opening up pre-orders to those 4.5 billion people.

This is every conversation you will ever have with an investor. If you can’t show them that you have a high profile in your industry and a way to turn that into a large amount of attention on your product, then you’re facing an uphill struggle. Conversely, if you can show that you have that level of attention already, this in itself gives people confidence that you’re the sort of person who can deliver a great product (if you weren’t, why would all those people follow you?).

Maybe by this point you agree with me, and you’re rolling your eyes wondering when the person who at the start of this article explicitly stated that they’d failed to solve any of these problems is going to tell you what the solution is. We’ll get to that in a moment, but assuming you’re now totally on board the marketing hype train and are convinced that this is where you need to focus, here are a few top tips to stop you careering off the rails faster than you can say “I hate Twitter”.

Mistake 6: anodyne listicles about nothing are not the inbound marketing triumph that I hoped they would be

Shocking, right? I did at one point realise that the fact that nobody knew our app existed could potentially prove to be a problem. Luckily I had the perfect solution: I’m a fairly good researcher, so if I bash out a few articles about things I’ve learned along the way then we’ll be drowning in signups. People will be blown away by my purple prose as I Pied Piper them along the breadcrumbs of mixed metaphors towards our gingerbread house solution.


Two of those stories were featured by the UX Collective. One was chosen by Medium for the front page. And still nobody read them. They’re not bad — they were good enough to get picked up and promoted after all — they just don’t really have all that much to say. They’re certainly not something people would go out of their way to share, so their ability to drive traffic towards our app was exceptionally limited.

When the stats proved so underwhelming however I was at least able to diagnose the problem: what we needed was more content! I could write an entire guide to conducting research, then this would become the new go-to bible for anyone looking to get into the industry! Then I remembered what Mistakes 2–4 were, and went off to have a big cry instead.

The story with the most reads and views I’ve written is actually the only one I’ve publicised myself on LinkedIn, which shows the importance of personal networks. Happily for me I’m a Social Media Maestro, and not at all somebody liable to make Mistake 7.

Mistake 7: People liable to turn into deeply neurotic teenagers when doing things in public do not make excellent founders

I don’t have a Twitter account. I don’t have an Instagram account. I didn’t even have a Facebook or a MySpace account at school. Sometimes I pretend that this is because I’m heroically resisting the corrosive tides of modern culture, standing on the beach of our collective shattered dreams like King Canute, bellowing at the tide “No! Back! There is a better way!”.

The actual reason is because I don’t have many friends. The idea of a public — public! — counter showing exactly how many friends I have was utterly terrifying to me as a teenager, and is only mildly less anxiety-inducing now. Don’t get me wrong, I do have friends (marketing slogan ©Luke Kelly, 2020), I’ve just always hated the idea of having to publicly display them. Social Media still feels like a giant popularity contest to me, and I stopped caring about those around the same time I stopped listening to Nu Metal.

Which is all perfectly reasonable — just so long as you’re not planning to do something like launch a company that you need to somehow make the world aware of. Out of all the mistakes listed here I actually think this one is the biggest: you simply cannot be shy if you’re trying to start a business. Either you, or your co-founder, needs to make Jordan Belfort look like Alan Bennett.

A case in point: there’s a Slack channel about UX Research Repositories — the exact thing I’m building — which a few people have asked me if I’m part of. I’m not, because the idea of walking into a room full of strangers (even a virtual one) and saying “hi im luke i built this do you like it if not then dont worry its fine” felt too intimidating. Partly as I knew that what we had was still at such an early stage (“Let’s just build that one more feature — then it’s good enough to share!”), and partly because such a naked attempt at selling a product felt gross to me.

This is not a feeling you can indulge if you want to succeed. The one network I have is LinkedIn, but I’ve been wary of putting too much about Filo on there as I was worried it would look like I was trying to sell things to people. Which is, like, literally my job as the founder of a startup.

The truth is I’m always selling Filo, to everyone. I have to be. They know that — I should have just outright asked everyone I know “can we build a custom repository for you for free, just to get the case study”. The worst they would have said was “no” — to this day I’m not sure what I was afraid of.

At one point I’d lost the plot so completely that I “liked” a post I saw by someone who I’d worked with in the past (and have even met socially several times) that had just been promoted to Head of Design and Research at a new company. I then thought I should follow up somehow, so sent a rambling DM about how much I trusted their “eye for design” and asked if they could spare 5 minutes to help me review some designs for Filo.

I then started panicking that I’d somehow overstepped some non-existent social boundary, went back online an hour later, deleted the DM hoping that they hadn’t had time to read it yet, and unliked the original post (!). My completely addled brain had decided that even this token celebration of an ex-colleague’s achievements was a slippery slope towards hard-selling myself into ostracization by my professional community. I’d be shunned at all of the industry networking events I don’t go to, as people disdainfully looked down on my outrageous attempts to build a digital tool that might help them do their jobs a bit better.

Absolute madness. And also clearly a fantastic anecdote to trot out on a professional networking site. It’s such a good lead-in to the final (and biggest) mistake though that I thought it was worth the risk:

Mistake 8: get over the embarrassment of failing before you even start

Everybody knows the stats: 90% of startups fail. Whatever. As soon as Sequoia answer the letters I’ve been leaving under the windscreen wipers of their investors’ cars I’ll be away.

Whilst the vast majority of founders undoubtedly know going in that the odds are stacked against them, one thing that’s rarely mentioned is how embarrassing it is to fail. I think this embarrassment was the root cause of a lot of the mistakes I listed above. I’ve been working in tech for more than a decade, and I’m pretty good at what I do. I had a good idea, a great team, and 14 months to make it work. Why did I make so many avoidable mistakes?

Partly it’s experience of course — I’m really hoping that me listing these mistakes will help a few other people avoid making them — but looking back I can see that on too many occasions I’d let anxiety cloud my judgement. It stopped me from chasing down every even remotely plausible lead until people were sick of the sight of me, because that would be admitting that I was struggling. It stopped me from ripping things up and starting again, because that would be admitting I’d got things wrong. It stopped me from writing honest articles about what I’d learned, because anodyne listicles about nothing are clearly the key to being given a ceremonial black turtleneck and a ticket to the big leagues.

Above all, it stopped me asking for help. It stopped me from being able to see that I needed help. Which, for the three of you who have made it 4500 words into this article, finally leads us to the finishing line: because this is one mistake which I’m not going to make.

(Not) Mistake 1: can you help? I don’t really know what I’m doing

I’ve loved working on a startup. Despite everything that I’ve described here, I’d recommend it to anyone. And personally I don’t see why the complete absence of any signs of actual success should be taken as a reason to stop.

My brilliant, amazing, inspiring co-founder is still strapped into the passenger seat alongside me, muttering under his breath as he puts out the latest fire I caused as our plane splutters round in circles on the runway. And we’re both keen to keep going, as we’re pretty sure that there are a whole boatload more mistakes out there just waiting for someone to come along and make them.

We both have full-time jobs now (which we’re much better at than we are at this) so Filo is very much a part-time project for us. But you know what? We genuinely don’t care. Even if what started as a chance to change the world becomes nothing more than a way to work on something I enjoy with someone I respect, then that’s still something I’ll always feel grateful for.

And as for the one mistake I’m not going to make, the great man himself said it much better than me:


It’s never too early to ask for help — and I don’t think it’s ever too late either.

If you’ve read this article and it’s made you think of something — anything — that I’ve missed, you can reach me at luke@filo.io. If you’ve taken a look at Filo and think it’s horrendous, then I’d be hugely grateful if you tell me why. If you think it has potential, or you have an idea to improve it, then I’d love to hear that too.

If you’d like to read the pitchdeck that changed Seqouia forever meanwhile — even just out of curiosity about what a deck looks like — then let me know and I’ll send it to you. If you want to know about the 37 mistakes that missed the 5,000 word limit of this article, then drop me a line and I’ll send you a 4 hour voicemail of me crying down the phone.

I don’t think I am the next Steve Jobs you know, but I don’t think anyone should ever let that stop them. If you ever have the chance to try and build something of your own — even if it’s just a hobby — then take it. You’ll learn more in trying and falling short then you could ever imagine.

And even more than that, you’d be proving to other people that it’s worth giving it a go.

Luke Kelly

Founder, Filo

(still not on twitter)

P.S. Sequoia did actually get back to me by the way. I mean they said no, but still. Once that next feature launches…

Founder of Filo.io, the UX Research Repository which lets you build a custom repository on top of our platform.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store