In May last year I decided to take three months out to focus on a publishing startup project.

I’d been freelancing for about 18 months as a content strategy consultant, but I was finding it a little frustrating not technically being or feeling in charge of a publishing operation.

Often what I advised wasn’t carried out, and I felt like I wanted to better put what I was learning about the industry to better use.

While this is a blog about what I did and what happened, the lessons I learned in the period were invaluable, and I think anyone involved in content marketing could benefit from them.

The viral traffic surge in publishing

My research had led me to form quite a few hypotheses that I wanted to test, following a series of major traffic coups by US publishing startups.

A few articles on this can be seen below:

Headlines like this were making me eager to try it out for myself. 

I noticed that there was a fundamental difference between the publishing models of these companies and the larger organisations I had worked for in the past.

It was that these publishers managed extremely lean operations, often repackaging or curating content rather than creating it originally, yet they were bringing in huge traffic.

Made From History

I felt like there was an opportunity to build a site and experiment with building traffic quickly. I picked history as a vertical for six key reasons:

  1. There are unlimited stories to dig out and repackage.
  2. Most of the media in the space is very fragmented. Wikipedia dominates search, while historical magazines don’t offer a clear value proposition against Wikipedia content.
  3. I studied history to degree level and have an interest.
  4. The proposition could be popular amongst connected 18-25 year olds who are either looking to go to university or graduates.
  5. I would be able to get students to contribute fairly easily.
  6. The World War One centenary (4th August 2014) gave us a significant editorial hook, which was essential given the subject matter.

Hypotheses

At this stage, I wanted to test out three key hypotheses:

  1. Could we build a distribution network that was basically automated?
  2. Could we produce enough highly produced content that people would share?
  3. Could we create low cost video (I wanted to see how low I could go) that would act as a driver for the main site?

I launched Madefromhistory.com in late May 2014, employing a graduate intern from Kings College London. We also hired some office space.

First month / build up period

The first few months were way tougher than expected. To be honest, I hadn’t really foreseen the amount of frustration that comes in an early stage publishing operation.

While I had the romance of building rapid traffic success in my head (a curse of pretty much all marketing), I had to contend with a daily grind of creating content, building a network and manage a video process, with no feedback or advice.

Things were slow burning, and I remember being preoccupied with what came first: the content or distribution?

There are lots of historical Twitter feeds with significant followings. The largest, @historyinpics, didn’t seem to have a discernible strategy other than to publish retro pics (largely of celebrities).

Even so, I quickly realised that they had built that feed due to the longevity of a much wider network. Starting from rock bottom online is tough.

What happened in month one:

  • Our social media channels were sending dribbles of traffic, and it was costing about 20% of the intern’s time to maintain.
  • We didn’t have any authority on search engines, and subsequently didn’t rank for our brand name. No traffic there either.
  • Our content wasn’t being shared because we had no captive audience.
  • We hadn’t produced any video, despite an attempt to create an eight minute explanation to the 1964 Civil Rights Act, which had a significant anniversary. This ended up being a costly distraction.

But there were also a couple of minor successes. Some content did well on reddit, and drove us 5,000 or so visits.

We also created some larger set pieces that garnered some external interest, such as A Graphic History of Why Americans Celebrate 4th July, a section of which is below:

Our main style was to create large explainer pieces that were more reminiscent of scrollable Powerpoint presentations than standard articles. 

Second phase / viral period

I decided on a number of set pieces which I hoped would generate us a lot more viral traffic via social sharing in time for the World War One centenary on 4th August (for instance How the Start of #WW1 Would Play Out on Twitter).

I’d learnt a lot about reddit in month one, so I got a plan together about how to distribute on there.

I also continued to want to produce low cost video, so commissioned all of what can be seen on our YouTube channel (effectively 10 videos in various formats at approximately £1,750).

What happened in month two:

  • Our social traffic driving didn’t play out how we wanted on Reddit – for our most important pieces we got removed without explanation, but then we had three successes in a row for low cost traffic.
  • Facebook just turned out to be a surprisingly bad performer for growth – we simply couldn’t grow likes to our feed without paying, despite syndicating shareable content up to six times per day. Nevertheless, it still drove as much traffic as twitter, with largely better engagement.
  • I worked out a way to automate twitter linking back to our content, which relieved time pressure from production.
  • Video played out to be a big drain on attention – editing with narration is extremely time consuming. We just managed to publish 8 videos by the time the editor completed one month.
  • The hook of 4th August was okay – we made some traffic gains, but a lot of big publishers had built up lots of content for the World War One centenary and drowned us out. It wasn’t a big deal for us like I’d anticipated.
  • We finally ranked on Google for our own brand name.

Third phase / shut down and revelations

We managed about 30,000 sessions in August. Given we were at zero six weeks before, that felt like some sort of progress, but it had almost entirely been driven from Reddit.

We had no captive audience to speak of and continued to perform wretchedly on search engines. Our video was also getting nowhere.

I decided by the end of August that the experiment was no longer viable and I had to go back into consulting.

One last chance…

I did, however, find the energy for one massive set piece. I had an old video script about the Vietnam War that I decided to turn into a large graphic explainer.

It took ages to produce, but it was definitely our best piece. However, we didn’t really have any editorial hook – there was no anniversary to my knowledge, so I had no idea why anyone would care.

Title section for The Vietnam War: A Visual History of ‘America’s Greatest Shame’, which was by far our largest piece.

But on the day I finished it, a friend on Facebook said they were in Hanoi for Vietnamese Independence day – which had occurred exactly 69 years before.

Our piece was about Vietnam, so we submitted it to Reddit with a strong emotional headline. It stayed top of the history subreddit for 2 days, had over 1,000 comments and drove 30,000 visits to the site. I put this down to complete luck.

We would never had this win if I hadn’t chanced upon my friend’s status update.

Traffic starts coming in

Expecting reddit to be your main traffic source is a bad idea. We had some successes there, but nowhere else, so we were unable to capitalise on rapid viral spikes that had little engagement.

I decided to wrap up proceedings almost totally by mid-September. The site was still live and we still tweeted content, but that was it.

We had published about 220 articles in three months. Not a great deal by professional publishing standards, although we had also maintained Pinterest, Twitter, Facebook and published 8 videos on YouTube.

Our most successful video from our YouTube channel – so much harder than it looks!

But when I tracked traffic over the next few months, traffic was going up. In fact, from the time we stopped working on the project, traffic from search rose pretty much every day.

By mid-November, this was up to 1,000 sessions per day. I was doing less than I had in the summer but now had more traffic. I considered there might be something in it after all.

Project outcomes:  hypotheses tested

The whole project showed that most my initial hypotheses weren’t really wrong or right:

  1. Could we build a distribution network that was basically automated?

Yes, of sorts, but it was going to be a very long time before it got any noticeable numbers without a significant spend on social advertising.

  1. Could we produce enough highly produced content that people would share?

We certainly produced some stuff I was really proud of, and we got quite a bit of kudos from teachers and the like, which was nice. But with a team of two it was impossible to create enough of it.

  1. Could we create low cost video (I wanted to see how low I could go) that would act as a driver for the main site?

No. This just didn’t work and was a big distraction. If anything, I wish I never pursued this option and focused on our strengths in graphic and article production. But there were some other things I’d totally overlooked that only became apparent once the project was over.

Five outcomes I’d never really thought about

  1. Reddit could drive significant amounts of traffic and it was possible to have some sort of submission plan. Relying on it, however, was folly.
  2. You need a lot to spend a lot of time or money on social media to get a reach big enough to drive virality. We simply had neither.
  3. Good old search took a long time to get going, but when it did, it was our best traffic driver. Given other startup successes in publishing, this surprised me. If I’d have initially focused on search traffic and stock content, we would have seen far higher long term returns.
  4. No matter how good your content is, you always need a good trending hook to get the best returns. Spending ages on a piece of stock content without a wider public interest in the subject matter is a wasted effort.
  5. We got drowned out by other publishers when our ‘viral period’ occurred (the week of the World War One centenary). With so many big publishers upping their game, we were largely ignored.

Conclusion

So for this initial phase of the project, I tried a lot of stuff, and some didn’t really work, but then some did.

The project didn’t match my initial ambitions, but the learning experience was quite incredible – far higher than I could have learnt if I’d continued consulting. The main thing is I was able to take this experience, and the data that went with it, to create a new phase for the project (it is being continued), or reuse in lots of things related to my career.

Disclaimer: This post might seem like flagrant self-promotion, but hopefully it gives a unique take compared to the best practice successes the tech media like to hype up. In short, it’s a story about things not turning out as first reckoned.