It all started with a single tweet…

Before we knew it, our traffic was redlined, our site was down, and we were frantically calling our hosting company to get a hold of someone – anyone – who could help.

Sigh – and things had been going so well…

Hold on, woah, back up a sec

We’d been planning a big PR push around our work for the New York Times’ bestselling author Tim Ferris and his book, The 4-Hour Body. Everything was going according to plan: the case study was engaging and written in a brisk, easily-digested style; the design was awesome; we had all the right calls-to-action; and a solid viral loop built into the page. In a major coup, we’d also secured help from the “Superman of Silicon Valley” himself, Tim Ferriss, to help us promote the article.

So everything was all set, right? Wrong.

Our first inkling that there was a problem came via our server monitoring tools. Out of nowhere, Chartbeat and New Relic went crazy, dwarfing our 30-day record number of visitors in a matter of seconds. The next thing to go was our Twitter stream. @-mentions started to pour in, mentioning our case study, and the words “doesn’t work”, and “broken”. It didn’t take a ton of detective work to figure out what had happened – we’d been Ferriss’ed.

Tim had tweeted out a link to our “4-Hour Website” case study, and posted it to his Facebook wall. Although we’d known he was onboard to to provide promotional support, we hadn’t been too specific as to exactly when he should do it. Turns out our man Tim knows how to sling some serious traffic – we were getting something like 600 page requests per second. Within short order, our poor server began to foam at the mouth, before finally choking on the virtual firehose of requests being flung its way. Although our setup is normally pretty solid, we had simply underestimated the volume of people that’d arrive at our front door all at once.

Server overload

The upshot was that our main site at dtelepathy.com slowed to a crawl for everyone – and several thousands of visitors didn’t even make it to the page, instead seeing a white screen as their browser requests struggled to be heard in the overwhelming din.

What Happened Next

Honestly, our immediate reaction was to toss out a few choice cusswords (that we’ll decline to repeat here).

Our next move was twofold:

  1. We called our hosting company, Media Temple at multiple different numbers, trying everyone there from their support line to our personal contacts, looking to get our servers beefed up to handle the traffic load
  2. We hopped onto Twitter and Facebook and kept the lines of communication with our suddenly-swelled audience open

But there was bad news – in order to get more capacity on our servers, we were going to need a physical change in our server hardware. This meant that our IP address needed to be changed, and we were going to have to wait for the new address to propagate across the DNS, and redirect visitors to the new server’s location.

This would take a while. Rock, meet Hard Place.

We were stuck between a rock and a hard place

The Aftermath

And so we waited. And waited. And waited.

Actually, that’s not all we did. As the rush subsided, more visitors were gradually able to view the site, so we immediately flipped on our trusty surveying tool, Qualaroo (formerly KISSinsights), and our onsite live chat tool Olark, in order to monitor for any additional issues they might be encountering.

On the social media side, we were heavily engaged in clean up – letting people know that we’d been experiencing some issues, and to give us another try later on.

Meanwhile, our developers used the time to assess our current server setup. They then provisioned and load-tested a more efficient system, consisting of an improved software stack, plus better protection against sudden traffic spikes, courtesy of CloudFlare.

Most importantly: we didn’t stop. Despite the initial setback, we still had a story to tell, and we were going to make sure that as many people heard it as possible.

How not to mess up a PR launch

“Why do we fall, sir? So that we can learn to pick ourselves up” -Alfred, Batman Begins

Although we’ve done a number of product launches in the past, we’re still learning new things about these tricksy beasts.

This time around, we gleaned some particularly valuable learnings that we’re seeking to apply to our preparations for the next big PR push. But of course, we’re not ones to keep this stuff to ourselves, so here are the major takeaways.

Pre-flight

Pre-flight

From Shutterstock

Takeoff

Takeoff

From Shutterstock

Houston, we have a problem

Ejector seat

Looking back

Running a big PR launch is a lot like being a teenager – it’s an exciting, chaotic time where everything seems of hugely magnified importance and dire consequences. But unlike your pubescent years, you can actually take simple steps to prepare for them, and mitigate at least some of the instability and uncertainty.

As for the rest of it, gird your sense of humor, and keep communicating – as unbelievable as it may seem when everything’s falling apart, you’ll be fine.

What PR damage control tips do you have? Share them with us in the comments!

Comments
  • There’s a delicate balance between having too much capacity and over-paying, and not having enough but saving money. In the end, there are key hosting points that can help. Caching and content delivery networks do wonders, so I always use Cloudflare (as you ended up using), and WP Supercache combined. Then, I’d make sure you’re on a VPS (Virtual Private Server), one that can be scaled instantly, adding more RAM or CPU on demand, instead of those old-fashioned dedicated hardware servers (so 1993 lol). You could go the Amazon EC2 route too, but that’s a more complicated layer to deploy. With my WordPress services, I offer a mix of shared and VPS, fitted for the client’s needs like a finely tailored suit. 😉

    • Jason Amunwa

      Agreed, Rob – there’s no one-size-fits-all when it comes to the infrastructure. We learned that it pays to regularly evaluate your setup to ensure it fits your current, and anticipated needs.

  • Hey Jason! Very helpful article and all-around awesome site you have here, from content to design. The type of read you want to bookmark and refer back to later – well done sir. 🙂

    Drew J
    (mt) Media Temple
    @MediaTemple

    • Jason Amunwa

      Thanks Drew, and thanks to your team for the much-needed assistance! =)

  • Hi Jason. There’s some invaluable advice in this post! 600 page load request per second via social media! That is some serious Klout (excuse the pun…)

    Best wishes, Alex

    • dtelepathy

      Hey Alex – yep, hard-won experience indeed. Tim’s a traffic beast! =)

  • Great article. One highly techy thing we always do to protect ourselves (and our clients) when we do a new launch is to lower the DNS TTL value to its lowest setting. This way if there is a problem you can transition to a new server/ip a lot faster.

    This could apply to potential upcoming PR opportunities. If you know something big may be coming down the pipe, lower the value as a last resort – so if you have to do something drastic the propagation time is really quick.

    Love your blog btw.

    /mike

    • Thanks Mike! Yeah, the TTL value did factor into the proceedings – definitely a great lesson learned all round.

  • Nice share!, I really agree that we had to stress test and implement a caching plugins also had a hosting company that has a 100% uptime guarantee SLA to avoid ‘down disaster’ when Search engines crawling our sites upon PR update, PageSpeed is a crucial thing next to Content Quality.

    • dtelepathy

      Thanks, Axel. a 100% uptime guarantee, huh? I personally view that with a pinch of salt, given that even Google isn’t immune to the occasional downtime, but the gist is that you want to have someone on the other end of the batphone when you need to pick it up.

  • Great tips, Jason. I always find swearing helps! But seriously, my favorite tip is “Your Hosting Company is Your Friend.” A lot of people – and companies – view their host as the devil because their site is slow, down, etc. If you look at them as a friend and partner, things will go smoothly.

    • dtelepathy

      Hey, thanks Michael – glad it was helpful. And yeah, there was definitely some blue vocab tossed around =) Identifying and nurturing close vendor partnerships like your hosting company can really pay dividends when the sky is falling.

  • Cool case study Jason. If you’re feeling bad about it, just remember; some of us dream of a PR launch that crashes a server 😉

    • Jason Amunwa

      True words =) Although it never feels that way at the time!

    • i noticed while dealing with clients that they do rely only on their teams! But I used to suggest that your executive team isn’t enough. A lot of start-ups have a great executive team who have done amazing things for other organizations. Unless your team is made up of Sean Parker or Andrew Mason or Elon Musk, no one will care. Well, that’s not entirely true. People will care and it might get your foot in the door, but it won’t be enough to get you the stories you need.

  • In my opinion every SEO should launch their PR ASAP of new businesses. It will help you to increase digital marketing.
    Visit Here