How we got a dozen launch-day write ups with a little JavaScript, Mechanical Turk, and the Google News API. Code included.

That said, lets hack some press…

Getting press is daunting for just about every founder. Who do you talk to? How do you find them? What do you pitch them? All questions I spend hours thinking about…for…every…project.

But with a little creativity, and a little hacking, getting press is easier than I thought. Easy enough to get a recent product covered by…

  

Background

Some friends and I were launching ThingsWeStart, a “Crowdfunding Aggregator” – imagine all of the projects on Kickstarter, Indiegogo, etc. displayed on a map, so you could support projects near you.

Something this badass deserved a press push…

Step #1 – Find the Right Reporters

The first step was to find reporters who might cover us. That started with a Google News search for “Kickstarter”.

This is a very different approach than listing all possible news outlets and submitting tips to them. Instead I was personally targeting individual reporters, who already demonstrated interest in our subject matter.

Want to follow along? Search Google News for articles related to your company’s value proposition.

Step #2 – Download their Articles

Now I had some articles, but what I really needed was the reporters’ contact info. Instead of searching through each story on my own, which would take forever, I wanted to created a spreadsheet of them so I could farm out the work.

Luckily, Google News has an API which makes that relatively easy – but there’s no sense in you reinventing the wheel. Please, use my code:

Step #3 – Collect their Contact Info

With the list of articles in hand, I wanted to collect the names and email addresses of each reporter – so I could personally tell them about our launch.

Thing is, I didn’t want to do that work myself – it’s boring. Instead, I asked the good folks on Mechanical Turk (mTurk) to do it for me…for $.03 per email address! If you’re not familiar with mTurk, I can’t recommend it enough.

To get started, I created a HIT template:

Then I uploaded a CSV (the spreadsheet you downloaded in Step #2) to mTurk:

…and mTurk automatically combined the two to create a HIT for each article in the CSV:

Rather than investing two weeks digging for contact info for 700 articles, dozens of people delivered me results within a couple days.

Important: I suggest you turn off requiring Master Turkers. You’ll have to babysit the results, and block Turkers who aren’t giving you valid results, but you’ll likely find more people willing to help.

Following along? Create a new HIT on mTurk and use this HTML as the basis for your template.

Step #4 – Progress Monitoring

When I did this, I saw a large number of HITs completed within the first few days. After that, results started to drop off.  I think my HITs eventually got pushed to the bottom of the stack.

To counter that, every couple days I would cancel my incomplete HITs, and relaunch any articles I didn’t have contact information for.

Within a week, I had contact information for 350 reporters who had written articles about Kickstarter.

Step #5 – Build a Press Kit

The most important rule I’ve found when working with reporters is:

The more you help reporters, the more they’ll help you.

These poor folks are under crazy deadlines, have to produce multiple stories per day, and are inundated with requests for coverage.  The more we can help them out by writing our story for them, the better off we’ll both be. Win-win.

Here’s the press kit we prepared for our launch.  It includes:

Step #6 – Prime the Pump

With contact information in hand, and a killer press kit, it was time to reach out to the reporters. 5 days before launch, I gave them a heads up we were coming, but no access to our press kit:

A couple days later, I gave them access to our full press kit, but told them not to write about us until launch day:

On launch day, I told them to have at it:

Important: embargoes are essential. 

It’s important to tell reporters that your content is “embargoed” until you officially launch, which means they’re not allowed to write about you until your launch date.  If you don’t mention this, one of these folks is going to jump the gun and publish your story early. That in turn, will make other reporters less interested in covering you, since it’ll be old news by then.

Some news outlets/reporters hate embargoes. If you really want their press, only tell them about your launch the day before you go live.

Step #7 – Watch the Ink Roll In!

Once we launched we simply setup some Google Alerts and Twitter searches, and watched the press roll in.

Hack Metrics

• Hours spent: 40 (includes time coming up with the hack, writing the code, etc. Should take you less.)
• Money spent: 350 email addresses x $0.03/address = $10.50
• PR pros hired: 0
• Total write-ups: 13
• Traffic generated: ??? Remember when GoDaddy’s DNS service went down? That was the day we launched :)

Following along? I’d love to hear if you’ve tried this, or other press hacks!

Final Tips:

• Play around with pricing. If you’re in a hurry, you’ll probably want to pay more than $.03/article
• Block turkers who submit obviously fraudulent contact info (you won’t have to pay for those)
• I’m stoked, quite a few people are finding this article useful! Unfortunately, that means there probably lots of HITs on MTurk right now trying to get contact info for journalists. That means, you may need to either pay more, or wait a little while before trying it yourself.

TL; DR

The Google News API and Mechanical Turk can be combined to get lots of press for your launch.

The hackers’s guide to getting press 

What’s Next?

Subscribe via Email or RSS for more hacks and in-depth discussions on getting press, validating marketing channels, and reaching your customers.

First time to Customer Development Labs?

Check out our most popular experiments:

1. Interviewing 100 customers in 4 hours with MTurk
2. You’ve interviewed customers. Now what?
3. Lean Recruiting: Testing a 2-sided market

Customer Development Made Easy…

What to test
+
How to test =

Now offering the...

FOCUS Framework

This post is brought to you in part by this week’s Startup Edition question: How do you get a job in startups?

I am Not a Recruiter

I was playing ping pong with a friend recently when he tried to distract himself from the beating he was receiving…

Friend: I need to hire 15 people this quarter. Do you want to help?
Me: Of course not…I’m a dev, not a recruiter. Don’t be rude just because you’re losing.* 17 serving 3.

* Engineers typically despise recruiters. 

But he got me thinking – companies are so hard up for senior dev talent, they’re paying recruiters an extra 20% – 25% of a new employee’s 1st year’s salary (e.g. for a $100k/year hire, they’re shelling out an extra $20k to $25k to the recruiter who found her).

Curiosity piqued, I wondered…what would it take to solve this problem? And how would I test it?

Except when I’m Testing

Recruiting is a classic 2-sided market. You’ve got candidates and employers – neither of which will talk to you if you don’t have buy-in from the other (e.g. Craigslist is useless w/o buyers and sellers, Groupon is useless w/o companies and consumers, etc.)

When I mentor startups in 2-sided markets, the question is always, “Which side do I test first?” To which I answer, “The harder one.”

In this case, finding great developers is harder than finding people who want to hire them, so…

Of course, that begs the question, “How do you test one side of the market, without the other on board?” In other words, how can I get great candidates (e.g. amazing devs who currently have jobs) to talk with me, if I’m not working with great companies they want to work for? The answer…

Forecasting

To crack the chicken & egg problem of 2-sided markets, we’re going to “forecast” to one side of the market that the other side is on board. Let me be clear, I make a big distinction between Forecasting, Predicting and outright Lying:

Forecasting = Statements made with > 90% certainty
Predicting = Statements made with < 90% certainty
Lying = Statements made with < 50% certainty

You can see what I mean with a screenshot from my MVP:

Predicting would have been including logos of friends’ companies on this page, without talking to them first. It’s likely they would have been on board, but since I wasn’t > 90% sure, I didn’t include them.

Lying would have been including logos of companies I had no relationships with.

The flier above is an example of forecasting. I haven’t nailed down everything 100%, but if the candidates come, I’m > 90% sure I can deliver what’s written.

MVP Time

Using Unbounce, I whipped up an MVP and posted it on Hacker News. 8 hours later, I had 200+ candidates:

Next I spent 3 weeks researching and interviewing each of the candidates to confirm these folks were high quality. Sounds tedious but I had a couple tricks up my sleeve:

• I used to be a hiring manager for a large software company – that made screening faster.
• It’s actually fun. Nothing better than 1-on-1 conversations with smart people who I can help take the next step on their path. It’s proving to be a privilege.

Within a month, I had validated problem market fit for the “hard” side of my 2-sided market.

Validating the “Less Hard” Side

Once I knew I had solid candidates, it was time to validate I had opportunities to place them in. Of course I had friends who were hiring, but I wanted to see if there was a more scalable way to make connections:

Since my customers are my friends, Facebook did the trick and got me 4 intros to recruiters – 3 of which wanted to collaborate (i.e. I would source candidates, they would source open positions, and we’d share referral fees).

As a meta-cruiter(?) I now have access to a couple dozen open positions.

Problem Market Fit isn’t Product Market Fit

At this point, I’ve validated Problem Market Fit for both sides of the market, but it remains to be seen whether or not I’ll successfully place folks. If I can do that with consistency, we’ll know we’ll have achieved Product Market Fit as well.

Want to know how this story ends? So do I. I’ll share the results as soon as they’re in. Subscribe via Email or RSS to get them, plus I’m really excited about our next post – Hacking the Google News API & Mechanical Turk to get press!

TL; DR

3 Steps to Testing a 2-sided market:

1. Define the “hard” side (e.g. candidates) and the “less hard” side (e.g. employers)
2. Validate buy-in from the “hard” side by forecasting buy-in from the “less hard” side
3. Use validation from the “hard side” to validate the “less hard” side

Step 4. Profit.

First time to Customer Development Labs?

Check out our most popular experiments:

1. Interviewing 100 customers in 4 hours with MTurk
2. Testing your Domain Name
3. Crowdtesting: How many $’s is your idea worth?

If you liked this – subscribe for more Customer Development tips, tricks & experiments (Email or RSS).