Why we shut NewsTilt down

NewsTilt was a YCombinator-backed startup which aimed to provide services to help journalists become entrepreneurs and earn a living off their work online. It closed down in July 2010, a total of 8 months after it was founded.

A while back, we announced to our journalists that we shut NewsTilt down, only two months since we launched. I think people are interested in why it failed, and there are some interesting lessons in our demise, at least for me.

This piece focuses on my own view of why it failed, as opposed to answering the questions and comments which appeared following the publication of our announcement by Romenesko. I’ll probably take some time in the future to respond to critics in the future, and to get back to those who contacted me for comments.

Summary

Following the launch, everything started going to shit, and a huge number of challenges to the success of the company had arisen. The biggest of these were the lack of traction from launch, that we had lost the faith of our journalists, and because there were communication issues between Nathan (my co-founder) and me. This combination also killed our motivation.

As a result, I made a carefully thought out decision to shutdown the company, and return as much money as we had left (about 50%) to the investors. Nathan believed the best thing to do would be to pivot our company, and so I agreed to step down to allow him do that. After some work, he agreed that it was best to shut it down (hence the email above), and we are currently going through the steps of winding down the company, and returning the remaining money to investors.

Problem overview

NewsLabs failed because of internal problems and problems with the NewsTilt product. NewsTilt failed because:

  • journalists stopped posting content,
  • we never had a large number of readers,
  • we were very slow to produce the features we had promised,
  • we did not have the money to fix the issues with NewsTilt, and it would have been tough to raise more.

None of these problems should have been unassailable, which leads us to why NewsLabs failed as a company:

  • Nathan and I had major communication problems,
  • we weren’t intrinsically motivated by news and journalism,
  • making a new product required changes we could not make,
  • our motivation to make a successful company got destroyed by all of the above.

Overall, the most important of these are that Nathan and I had difficulty communicating in a way which would allow us save the company, and that this really drained out motivation.

NewsTilt problems

NewsTilt wasn’t a bad idea, but we certainly faced a ton of problems with it. Most of them could be overcome, but its instructive to go through what they were.

Who are you writing for?

NewsTilt was a news destination website, but we very quickly ran out of news. We relied on journalists posting news, and they stopped posting because they largely no longer believed that NewsTilt was good for them.

Journalists felt that they were writing for us, instead of writing for themselves, for their own brands. How could they feel anything else, since that’s the impression we gave them by the design of newstilt.com. The most important part of NewsTilt–that journalists would have their own brands and own domains–got cut from the minimum viable product in order to make launch date [1]. It was never re-added because of technical issues with Facebook.

As a result, it looked like NewsTilt was trying to be another Huffington Post [2], that is, a news company to compete with existing news organizations. As well as convincing the journalists that they were contributing to us for nothing in return, this had knock-on effects: other venues to which a journalist might sell their news refused to allow the same stories to be posted to NewsTilt. This is obviously the right thing to do if you perceive NewsTilt to be a competitor. If they had perceived each journalist’s NewsTilt site to essentially be a personal blog, as we perceived it, we wouldn’t have had this problem.

Since we weren’t making any money for them, and it appeared to them–correctly–that no-one was even reading their content, there was no earthly reason they would keep writing “for us”.

Worse is better

Somewhat surprisingly, the journalists we picked were too good. We made a big deal of only hiring the “best journalists”, something we spent a great deal of time getting right. We had a guy with a Pulitzer, one with an Emmy, and overall a great deal of talent writing for us [3].

In hindsight, this may have been a big mistake. The kind of writer we actually needed was one that was hungry to succeed. Someone who would write five pieces a day, and who wanted nothing more than to be a big-time journalist. We needed a young Perez Hilton or Michael Arrington, people who wrote for 18 hours a day in order to make their names.

Instead, we got journalists who were already successful in their day jobs, and who already had families and other commitments. They were checking out the latest thing in news, not hungry to make something of themselves. Why would they be, they already had made something of themselves. Unsurprisingly, they didn’t write 18 hours a day, instead just dipping their toes in to try NewsTilt out. They applied, and either never started or posted only a small number of articles.

I think it’s important to say that we really failed because of a lack of content. But that was a symptom of having the wrong kind of journalists. All the problems the journalists faced, not writing enough, their distrust of Facebook, their unwillingness to socially promote their work, would not have been problems for a young journalist eager to make a name for himself. If we had the sort of people who gave up everything to succeed at their dreams, these problems could have been blown past. But as established successes in their field, it was unreasonable to expect them to make giant changes for uncertain return.

We never made it clear how hard it was going to be to create an online presence, and so when articles went nowhere, there was little motivation to continue. Building a brand online is akin to doing a startup — it’ll take five years. But we failed to prepare them for this [4], and we failed to recruit the people with the kind of that kind of dedication to “making it”.

Wrong content

The actual content that the journalists wrote wasn’t what we needed either. Content on the site was a very high standard, but it tended to be very long pieces. Long pieces online are difficult to make a success of, as the online attention span is very low.

The problem was that I didn’t really know how the journalists should write their pieces, only a vague sense that it was wrong. I also didn’t really want to tell them what to write, since they were doing us a favour by writing at all. In fact, we actively told them to write what they wanted. This could have been fine, if they had taken their cue from their readers. But we didn’t really know who the readers were, and there weren’t that many of them anyway.

It took a while for me to work out that the first thing I did in the morning to sate my internet addiction wasn’t going to NewsTilt. I was still going to Reddit and Hacker News. When I did read the pieces, I wasn’t terribly interested in them; they were definitely better than what I read in the paper, but they didn’t trigger the dopamine receptors that made me want more and more, and I didn’t know how to fix it.

Who are the readers?

The fact that we didn’t know anything about our readers’ demographics underscores another problem: I don’t understand news readers. I certainly wasn’t one, and I didn’t know many people who really were. My customer development had largely consisted of talking to journalists and figuring out what they wanted. I never really–despite good intentions on lots of occasions–talked to people who loved news about why they loved it. So I was unable to say what was going wrong and why people weren’t sticking around.

I could possibly have fixed this at a certain level by giving a greater role to the editors, but I was very uneasy about having editors in the first place. I didn’t want to tell the journalists what to write, and I felt that greater input for the editors would have made each journalist’s brand less individual [5].

Traffic problems

The major reason the journalists bailed was that we failed them. We didn’t deliver the things that we said we would, and we wasted the content they provided.

One part of the service we offered was that we would get the journalists traffic. Whooops! Getting traffic is really really difficult. We completely underestimated how difficult it would be, largely because I’d never had a problem with it in the past. When I’ve needed to promote some pieces I’ve written, I simply submitted them to Hacker News and Proggit. However, that doesn’t generalize in any way.

In retrospect, it was foolish to offer to do promotion for the journalists. We should instead have built the tools to help them with promotion, and let them do it themselves. We had a couple of ideas of what these would be, but we never built them. We felt that we would understand promotion better if we started by doing it ourselves–standard practice for any company–but it sucked up massive amounts of time, and we never got anywhere with it.

We had no domain experience in promotion, and really had no idea what to do. Worse, we had no idea what to tell the journalists to do. We struck upon the idea that if we had fifty journalists, and they each cross-promoted each other to their social networks, then over time we would get more and more people to read each others’ content. Suffice to say that the journalists were not happy, and didn’t go along with it.

We had pseudo-hired a social media promoter, someone who had gotten bitten by the startup bug, and was interested in community management, social media promotion, etc. She was pretty good, way better than me, but was still relatively new at it. What we really needed was someone who knew this stuff inside out [6], rather than someone who was learning as they went.

We also missed a few opportunities for some traffic. We fluffed our TechCrunch launch by having the piece posted before the site was even launched. This happened because we were sending out old fashioned press releases in an effort to get old media to cover it, and they needed to go out a day early. Old media didn’t cover it, but TechCrunch did, a good 30 hours or so before we actually launched. We got 18,000 hits that day, nearly all of whom saw no news, and probably didn’t come back.

Technical promises

We basically over promised what we were going to do for the journalists. We showed them a list of things we were going to build quickly, then built things very slowly. Our technical ability was going to be our differentiator, but our technology showing was pretty poor.

One reason is that we misprioritized things, where the thing that really really needed to get built right now changed daily. Another is that building things take time, and we weren’t just building one product, we were building tons of products to be used in different ways. We needed Facebook integration and Twitter integration, to integrate the latest changes from the designer, to automatically admit journalists, and alert the editors upon application, and to actually build the product, to build the social promotion tools, etc, etc, etc. Individually each feature might only take between a few hours and a few days each. However, they build up very quickly, and soon we had features we promised would be ready weeks ago, still sitting in a low-priority slot on our TODO list.

We also suffered from a lack of technical resources. I spent nearly all my time doing CEO things [7], so Nathan was left to do the technical stuff alone. We needed to hire quickly to make up for that, but made a key error with our hiring.

We also made some bad technical decisions, such as failing to choose WordPress, and the whole Facebook thing. The latter also prevented us from moving journalists to their own domains, instead of being under the NewsTilt banner.

The end of NewsTilt

As a result of the errors above, we managed to alienate our journalists. By the time we ran out of content, it was already too late. The journalists were disillusioned and unhappy, and we did not have a product to prevent that.

I realized this when I found myself reluctant to suggest that people visit the site. Journalists were still excited by the concept, readers wanted to check it out, but I realized that I didn’t believe that NewsTilt was a good product for either set of our customers.

There were a few possible solutions:

  • fix NewsTilt,
  • make a different product, aka pivoting the company,
  • close down NewsLabs.

Fix NewsTilt

NewsTilt was definitely a good idea, just one that we hadn’t executed well on. The major flaw was that we couldn’t promote the stories well enough, and the journalists’ stories went into a black hole where no-one read them. The clear way to fix this would be to hire someone who knew what they were doing in this regard. We had a couple of people in mind who possibly could be brought in, and if it required stepping down as CEO so be it [8].

The second flaw was that Nathan and I hadn’t been working well together since about February. There probably isn’t any blame to be apportioned, except that we should have tried working together on something before building a company together. So one or the other of us should probably leave the company.

Presumably most other things could be fixed. We could shut down NewsTilt for a few weeks or months while we fixed everything, focusing instead on the personal sites of one or two journalists who were dedicated to making it big [9]. We’d fix our technical mistakes, build the tools we really needed, and relaunch later in the year to take in another 10 journalists. Over the next two years, we’d expand to 1000 journalists, and to 10000 over the next five years. At some point, we’d reopen NewsTilt as a news aggregator, when it actually made sense, instead of when we were too small to do anything valuable with it.

We’d need more money, but the new CEO could handle that. We’d need more tech, but the new money could fix that, and whichever one of us stayed would make it happen. Everything could be fixed.

Well, maybe. This assumes we could find a CEO, and that we hadn’t burnt our journalist bridges. It assumes that we could find anyone to give us more money, especially since we launched once and got no traction at all [10]. It assumes that we could do something with the first two journalists, especially considering making them money would be very difficult for the first year at least, probably longer, so we’d probably have to pay them ourselves.

Our motivation was sapped from not working well together, and so our ability to be optimistic was pretty sapped too, especially since one of us would be leaving. Working on NewsTilt had never been the fun that startups are supposed to be [11], and the stress was not being counterbalanced by anything positive. We also weren’t all that into news and journalism, so our desire to keep pushing the “mission” was extremely low.

I felt that it was probably possible, though extremely challenging, to make something of NewsTilt or some variation of it. When you combine a lack of interest and motivation with that extreme challenge, I think it’s clear that this wasn’t a good idea.

Make something else

We each had a few ideas of other products we wanted to make. I quite liked the idea of a Heroku for Python, a competitor to Google App Engine. I also had a few products that I knew I wanted to use, so this seemed like the best time to make them.

However, since we weren’t going to work together, whichever one of us stayed needed a new co-founder. To be fair to a co-founder, you have to be able to move to Silicon Valley, especially since my product ideas were all for early adopters [12]. I couldn’t move to California without my wife, and I couldn’t get a visa that would allow me bring my wife [13]. Take into account I’d essentially be starting from scratch, and that I’d been working 14 hour days for 8 months and had no motivation left, and you can see why I’m not sure I’d have made a success of this.

Close the company

When deciding how to proceed, we had to consider all the people in the company.

We had one employee, but she was working for free, and it now seemed unlikely that we’d be raising the money to pay her. Our co-founder relationship would have to end either way, so if we decided to part ways we wouldn’t have been screwing our partner. The journalists had already stopped posting, and they and we were not convinced that continuing was the best thing for them. The only thing left to consider was the investors.

We felt a great duty to two great sets of investors who had put up their personal money to finance our idea, even though they both kissed their money goodbye when they invested. They had largely invested because we had clicked as people, and we wanted to make sure we did the best thing for them too. We still had $20,000 of the original $50,000 they gave us, and we had the option to return that to them.

They too favored shutting down the company down and returning what money we had left [14]. They accepted that NewsTilt was not going to work, didn’t like some other ideas we ran past them, and everybody at that point agreed that returning what was left was the best outcome for them.

So far, everyone–founders, employees, journalists, investors–were better served by closing down. The final set of people to consider were YCombinator.

YC had consulted and advised us every step of the way. When we had co-founder problems, they gracefully refused to take sides. When we wanted to make a new product, they advised us not to proceed without co-founders, and that we’d need to move to Silicon Valley to be fair to those co-founders. And finally, they didn’t expect a cent back, telling us to give all the money back to our later investors. Not once in my whole time at YC did I believe that they valued their investment more than they valued us, and they were OK with us closing down. YC is a class act.

Given the options above, it was pretty clear that closing down NewsLabs was the best option.

Would the NewsTilt model actually work?

Despite everything that went wrong, I’m pretty sure that what we set out to do can be accomplished, though perhaps not by us. It is certainly the product that journalists want, but simply one we were unable to deliver.

The biggest thing it needs is a shit-ton of traffic, and that is not easy to bootstrap. Perhaps that’s why bootstrapped technology startups haven’t been very successful in media, and why most of the inroads into content startups have come from people with more money, ability to create traffic, or both: Demand Media, True/Slant [15], AOL’s Seed.com, or Pierre Omidyar’s Honolulu Civil Beat.

It also needs to ability to build great tech very quickly, and lots of it. There were tons of things we were dying to innovate on that media companies are still doing very badly, but we hadn’t the money to make them happen quickly enough.

Google could have made this work. I believe that if Google applied the same model they could probably succeed. They have the tech ability, they have the traffic, and they already have a massive news property. They also have have a big problem with whiny news organization, and an elegant solution would be to kill them off by enfranchising their journalists to be their own bosses.

One of the things we did right at the start was that the journalists trusted us. They may not have trusted us to do everything right or to be successful, but they felt we would do right by them. There are some companies who could probably try to replicate this model and not succeed because they wouldn’t be trusted by the journalists; Demand Media for example. AOL is still tinged in the scent of the content mill, but they’ve hired cleverly and are probably capable of pulling it off. Google has shown it is able to operate transparently and somewhat benevolently, but lots of people don’t trust it.

Things we did right

So far, I’ve tried to present why we failed, which focuses on our mistakes and is by definition pretty bleak. But we did lots of things right too.

For a start, we developed an idea that people wanted. I spoke to lots of journalists and came up with an idea that they were really interested in. Then I talked to more and more of them, developing the idea until it was something that lots of people got behind. To a certain extent, this was easier as an outsider looking in. We didn’t really know what the problem was with the industry, so we instead looked at what people wanted, and we really nailed the customer development aspect of this, at least initially.

The thing that the journalists wanted was to be in control of their own destinies. They don’t like how their newspapers are essentially fucking up their lives and possibly their pensions, and they don’t like the content mill alternatives. They really loved the model that we only made money if they did, and that a 20% cut was the way to go about it.

We were able to convince editors to come on board, and even to lend their names to the enterprise. Doug, Les and Jon helped NewsTilt no end, and I am very grateful to them. But they got behind us because they believed in the mission, and they believed in us, and getting this right was important early on.

Similarly, we got about 150 journalists to sign up to take part. Of those, 27 actually wrote something on the site. That’s no small number, and I was delighted in achieving that. We also launched, which is a lot more than many startups.

We did OK on the business side. We got into YC, and we ditched our first idea which was going nowhere fast. Between demo day and me going back to Ireland, 7 days, we managed to raise $50K from two small angels. I had a five minute video interview in the Wall Street Journal, which is going on my resume next to my Google Tech Talk. And when the chips were down, we somewhat bravely [16] decided to shut down, saving time and money for our unpaid employee, ourselves, the journalists, and investors.

Personally, I’ve learned so much from working on NewsLabs. The largest, for me anyway, was that before I started I still had a bit of that shyness that lots of geeks have; doing the CEO job quickly cures you of that. I don’t want anyone to come away from this essay with the idea that they shouldn’t do a startup because it might fail. Mine failed, and I still learned more and improved myself more than I probably could have in any other way.

Personal Lessons from NewsTilt

I’ve made a list of what I’ve personally learnt from working on NewsLabs. Not every one of these will generalize, but I hope my mistakes are instructive for other founders.

In no particular order:

Lesson: Deeply care about what you’re working on

I think it’s fair to say we didn’t really care about journalism. We started by building a commenting product which came from my desire for the perfect commenting system for my blog [17]. This turned into designing the best damn commenting system ever, which led to figuring out an ideal customer: newspapers. While there, we figured they were never going to buy, and we figured out a product that people were dying to use if it existed.

But we didn’t really care about journalism, and weren’t even avid news readers. If the first thing we did every day was go to news.bbc.co.uk, we should have been making this product. But even when we had NewsTilt, it wasn’t my go-to place to be entertained, that was still Hacker News and Reddit. And how could we build a product that we were only interested in from a business perspective.

This compounded when we didn’t really know anything about the industry, or what readers wanted.

Lesson: Don’t be too ambitious

NewsTilt would be a great thing to succeed at. If you assume that newspapers were going out of business [18], all the journalists would become their own bosses, and need something exactly like NewsTilt to help them. As such, there was the potential to be the sole source of news online. Ridiculous, massively ambitious, and very unlikely as a result, but if it worked we’d be billionaires.

Next time, I’m going to make a product which will make me comfortably rich, rather than one with a tiny chance of going supernova.

Lesson: Communicate your idea (and manage it)

Our idea kept changing. The more journalists I talked to, the more I understood what our product should be, and what people needed. Unfortunately, that means I changed [19] our idea all the time.

Worse, I failed to communicate effectively what changed and why. I communicated this badly to Nathan, and badly to the journalists. The latter was difficult to manage–no-one is going to listen to everything I say when it changes regularly–but the former was very important. The result was that Nathan and I never shared a vision of where the product was going, which was one of our biggest problems.

Lesson: Make sure your minimal viable product is viable

We were greatly influenced by the idea of a minimal viable product. Build less, launch, then iterate when you have customers. This is a great idea, but judging when your product is viable is always a tough challenge.

The conventional wisdom is to cut any feature which isn’t essential. Ultimately though, if you cut features which make your users feel differently about your product, that’s a problem. We cut multiple domains from our MVP, meaning that journalists were publishing under our masthead, which substantially changed how they felt about NewsTilt. They were writing for us, whereas they should have been writing for themselves.

Lesson: Be careful about cool ideas

One of the reasons that switching to domains was cut from our MVP was that it didn’t work with our Facebook integration. I was married to this idea that Facebook integration was really important. It was the only way that we would allow people to comment, because it forced them to use their real names. This would mean high quality comments, and great community interaction, and I was convinced that this was essential for our success.

And we could only get real names by making everybody use Facebook to sign in. Absolutely everybody worried about this, but I was convinced. I was totally wrong. It alienated people who didn’t like Facebook, including some of our journalists. Worse, it caused people to just not comment, meaning they didn’t come back, they didn’t engage with the journalists, and they didn’t start to frequent the site.

This was at the time of renewed interest in Facebook’s privacy policy–they had just changed it and people weren’t happy. Every day there were articles in newspapers about how Facebook was doing a terrible thing; there was massive backlash [20]. Our own John Graham-Cumming even wrote a piece called ‘the Facebook Cull‘, and told us the only thing keeping him on Facebook at all was that we required it.

Man was I stupid. When people asked to signup without it, I told them no. When the people who did sign up were worried about things being posted to their walls, I didn’t understand the problem. When readers said they wouldn’t signup to comment, I thought it was just a small minority.

After a few weeks, I realized I had made a mistake, and put it on our “nice to have” list. There were a million other priorities, and how important was this, really. So that was another massive mistake. I should instead have moved it to the top priority, in particular because it held back domains which should (um, also) have been our top priority.

From a technical perspective, it also prevented us from rolling out one-domain-per-journalist a lot sooner. There is an issue with Facebook where single-sign on doesn’t work across domains, so readers would have to approve each domain separately. As a result, we didn’t introduce what was probably the most important feature for actually making the journalists feel like they were writing for themselves.

Lesson: If you think you should build it, not buy it, you’re wrong

We built our whole platform ourselves. Now, we used lots of scaffolding, built on Rails, hosted at Heroku, using every plugin we could find. I reasoned that the platform was the core of our technology, and we were a technology company, and smart technology companies needed the flexibility that comes from writing the core of their platform themselves. In retrospect, this could only be considered premature optimization.

The natural thing to do was build on WordPress instead, but I wasn’t having any of it. The major problem with WordPress is that it’s written in PHP [21]. I hated PHP with a passion, and couldn’t fathom building my company on it. How would we attract good developers? How could we live with ourselves?

Really, I shouldn’t have worried. It was far more important to just get it built, and nothing could have helped that more than just using WordPress. We could easily have given journalists distinctive styles, so they didn’t feel like they were writing for us, and we could have built things really quickly by just plugging them together.

If I was worried about how productive we’d be with PHP, well, it’s not like we had to build everything in PHP. We could just have done data collection in Python, and made it available to the rest of our app through either a web service, the database, or some other way. We might not have been happy, but we would have stood a much better chance of being successful.

Lesson: Build quickly, little company

The biggest fallout of building our platform ourselves was that we couldn’t build quickly enough. When you roll your own infrastructure, everything takes time, more time than you can afford. And we had promised the journalists that we would very quickly build a large list of features, none of which were produced nearly quickly enough. This was the major cause of disillusionment–we over-promised and under-delivered–and this was an important reason why.

Lesson: Hire well

Since we needed to build so quickly, as soon as we got some money we wanted to hire another technical person. Nathan had a friend he wanted to hire, who was exactly the kind of great programmer he could work well with. However, it took some convincing to get him to try working on a news website, and he wasn’t sure he’d stick with it. We were sure we’d be able to convince him to stay, and we even waited two weeks for him to move to work with us.

Unfortunately, we were never able to excite him about the project, and we quickly realized he was never going to be intrinsically motivated the way we need for a first employee. There was a point when I was over in Cambridge with Nathan and the other developer, and I noticed that the developer wasn’t working on a Sunday. Now, we aren’t the kind of people who think our employees owe us 90 hours a week, but startups need that kind of work ethic from very early employees–exactly the reason that intrinsic motivation is so important. If your first employee doesn’t love what you do, doesn’t wake up each morning dying to work on HIS product, you have likely chosen poorly, and that’s exactly what we did.

Similarly, we hired someone who wanted to learn how to do community management and social media promotion, instead of someone who knew how already. This is a pretty tough area, and I think we made a mistake in not hiring someone with much more experienced for such an important role.

Update: The comment about working on a Sunday probably raised more ire than anything else in the piece, largely from developers, so I think I should clarify some things.

Firstly, we’re not talking about death marches or prolonged periods of 100 hour weeks. It was early days in his employment, maybe the first or second week, and we had either just launched, or were coming up to launch, and had a great deal of things to be done.

Secondly, early stage startups are not normal jobs, and early stage employees are not normal employees. Your first employee is almost a founder. While they get less reward as a result of taking less risk, the success of the company depends on them a great deal.

Thirdly, it’s nothing that Nathan and I didn’t do ourselves. We worked 80 hour weeks basically since December. And you can get more done in 80 hours than you can in 40, so long as you don’t prolong it.

Finally, I’m not speaking ill of the developer. The problem is that we tried to convince someone to join a startup who wasn’t really interested. The fault was ours.

Lesson: Distributed teams are hard

At the end of YCombinator, I moved to back to Dublin, and Nathan moved back to Cambridge. Neither of us had US visas, and we both had things to keep us here: Nathan’s girlfriend, and my soon-to-be wife. Both of us prioritized our significant others over the company, and I stand by that.

However, that meant a number of sacrifices, including that we were in different cities. We already had a communication problem when this happened, but this made it far worse. Our social media optimizer was in San Francisco, our designer was in York, and we struggled to make the whole thing work.

While there is nothing wrong with remote teams, I think your company has to be at a certain point for it to work. Everyone has to be on the same page, everyone’s roles have to be certain, and the communication has to be constant. We had none of these, and we never had the time to implement them.

Lesson: Work with co-founders before starting a company together

You need a co-founder who gets you, and who you work together well with. When Nathan and I signed up together, we had not spent any time working together, and that was a big mistake. Nathan is certainly a great coder, but when we didn’t share a vision, and we found it so difficult to communicate, there was no way we were going to get this built.

When I get another co-founder, I’m going to make sure that we spend a lot more time working together on other things before we start a company together.

Lesson: Transparency is tough

It was important to the journalists that we were a very open and transparent company. From the start, we tried to put as much information out there as we possibly could, and the most efficient way was to put every journalist we accepted onto a mailing list. However, this meant that our blunders and critical feedback were visible for all those journalists to see. Lots of them hadn’t started writing, we didn’t know them, and they had simply signed up, so we were always aware that our emails were semi-public. As a result, when we decided to close up shop, our closing down email was “leaked” to Poynter, leading to all sorts of speculation.

It takes a lot of time to be open like this, and a lot of effort to communicate effectively. The lesson here isn’t so much that we did it wrong, but that it’s difficult to do well.

Lesson: Don’t do too much at once

I finished and submitted my PhD thesis a week before the YCombinator application deadline. Three days later I gave a talk to 900 people at StackOverflow London. When I moved to California in January to do YCombinator, I had still to organize my wedding in May, and I had a paper to write in between. My PhD defense in April was 4 hours after we launched NewsTilt. In May I got married and went on honeymoon.

Basically, life happens. There is never a good time to start a company, just like there is never a good time to have kids. Certainly entrepreneurship favor those without other commitments, but it seems like nonsense that people with other commitments shouldn’t start companies.

While I don’t regret doing all those things, I need to stop feeling like I can do everything at once. Everything takes time, and that’s time which could be spent on other things which are really important too. In retrospect, I should have delayed either the wedding or YCombinator.

Lesson: Be very careful how you are presented to the press

When I gave my demo day speech to investors, I explained that there were tons of customers out there; in 2008–2009, 30000 journalist had been laid off. When I gave an interview to AllThingsD a few minutes later, Peter Kafka focused heavily on the unemployed part of this. I didn’t quite realize the problem–it seemed like a minor detail that he was focusing on a bit heavily–until potential customers kept asking “what about solutions for journalists not laid off”. Even though our product was for all journalists, it had effectively been maligned by what I thought was a minor detail.

This also led to people thinking we were going to take advantage of them, and that we were just another content mill like Demand Media. Even when we made it clear that we were only making money if they did–taking a 20% cut–this kept coming up, even with journalists who we had signed up and were using our service.

Update: Peter Kafka spent some time defending himself when he wrote about this post. I feel there is no need for this. It was my pitch that put unemployed journalists in his mind, and I had the opportunity to correct him afterwards. Clearly, the error was mine.

Lesson: You have the greatest product on earth, and everyone should be lucky to talk to you

My natural way to network is to chat to someone, develop a rapport, and to set up another chat to talk about the world, current events, and (given time) some business [22]. But even to chat to someone, you need an in: what do I know about that person, what could I say to get chatting. So this is what I did on Demo day.

The way it should be done it in to boldly walk up to them and ask them what they thought of your speech. After a few minutes discussing it, they have to talk to someone else, so do you, that’s fine, here’s a card, we’ll chat later.

Irish people tend to self deprecate. They also don’t like successful people, and certainly not people who talk like they’re the greatest thing in the world. Silicon Valley is the complete and utter polar opposite. Self deprecation is out. No-one invests in a company who isn’t convinced that they are the greatest thing to ever happen. I’m thinking “my company has a great idea, but most companies fail, probably mine too, but we’ll certainly try as hard as we can to make it work”. Great entrepreneurs never concede that they might fail, and tell everyone how lucky they are to be able to invest in their company.

I was certainly told early on to present ourselves as the greatest thing ever, but I didn’t properly internalize it until demo day was over, which was probably too late.

Lesson: Two sets of customers is hard

NewsTilt was a product designed to connect journalists with readers. As such, we had two sets of customers, which means we need to do customer development twice. I spent a great deal of time designing the ultimate solution for journalists, and almost no time on what readers wanted. As such, I didn’t really know what to make, or what to say to the journalists about what they should write.

It’s tempting to look at the lesson as “don’t forget to do both sets of customer development”, but I think its many times more difficult to do it twice than to do it once. In the future, I’ll certainly be aiming for tools that only need to appeal to one set of customers.

Epilogue:

Update: A final point that should be made is that this is not an attempt to blame anyone. The journalists aren’t to blame: we didn’t make a sufficiently good product for them. The developer isn’t to blame; we tried to hire someone for a startup role who had no interest in startups. No, the only people to blame is us, and more specifically me, since I was at the helm when it all went down.

What’s next?

While I’m still tying up loose ends with NewsLabs, I’ve gone and gotten a real job! It’s great to take a break from the stress of startup life, and I’m loving working on compilers again.

I’ve just started a job with Mozilla, and I work on the Javascript engine in Firefox. Its a great job, working with smart people, on product used daily by about 400 million people. Its the sort of job I was looking for when I decided to do a PhD 6 years ago, and the perfect place for a geek to end up.

Finally, I’m looking forward to working on side projects. All projects get put on hold when working on a startup, and most were on hold during the final two years of my PhD. I have to write a scripting language, learn Haskell, read SICP and Concrete Mathematics, fix the mess that is Autotools, and am currently writing a little language for an itch that needs scratching. I’m also going to start writing a lot more. There are a few NewsTilt observations yet to be made, some lessons from my PhD that were a bit informal to make it into the thesis, and about 15 half-written pieces which I hope haven’t bit-rotten since I sketched them.

If you’re interested, all things will be posted on my twitter, as soon as I get a chance.

Updates

[1] We didn’t consider delaying the launch, as news is rather timely, and journalists had prepared news for our launch that started going stale as we stalled.

[2] Many of the people we spoke to felt that the Huffington Post was no better than Demand Media in that they exploited journalists. And we looked the same because we also didn’t pay journalists.

[3] Did I say “for us”? I meant “on our platform”. Easy mistake to make…

[4] We weren’t hiding this information from them. It just took us a long time to realise it.

[5] Here’s another problem. We thought way way too far ahead of ourselves. We were worrying about brands when there weren’t any readers.

[6]which we didn’t have the money for, etc.

[7] Incidentally, I used to wonder why tech startups had people who weren’t coders. I have a new-found respect for these people.

[8] I was generally conscious to call myself “co-founder” rather than “CEO”.

[9] More, um, dedicated founders might have just brought on new journalists, and pretended there wasn’t a problem. No points for guessing why we didn’t do this.

[10] If we did, we probably wouldn’t get very favourable terms, which isn’t the end of the world, but it does lower the potential reward for everyone involved.

[11] When PG lures in a new batch, he always talks about how much fun a startup is. And looking around, all the other startups were having fun as they went. This should have been a bigger clue earlier on.

[12] Without going too much into it, I used to believe that a company didn’t need to be in Silicon Valley to succeed. I still believe that, but not being there is a large disadvantage at the least.

[13] If you support the Startup visa take note: if the startup visa does not allow a founder’s significant other to work, then many founders won’t move. I can support my wife on a H1B because it comes with a high salary, but good luck on a founder’s salary, no matter how good the funding is.

[14] Technically, neither set of investors had any say in what we did. But we had to consider them because we felt a duty to them.

[15] Which, despite being a much larger success than us, also shut down, in circumstances which are regarded to be largely unfavourable.

[16] If I do say so myself.

[17] Despite the fact that 90% of what I write goes unfinished.

[18] My money is still on the newsapocolypse.

[19] There is a challenge here to iterate your idea and change and refine it to be what people want, without making them feel like they haven’t a clue what you do.

[20] Ah, how soon people forget. Haven’t seen anyone care about Facebook’s massive privacy problems since Google’s Net Neutrality thing.

[21] Update: deleted cause it wasn’t a nice thing to say.

[22] I’m going to say this is because I’m Irish, but no doubt real Irish businessmen know how to do this better.


Originally published at blog.paulbiggar.com on September 16, 2010, last updated October 7th, 2010