Testing copy on 37,035 people for $0

I’ve been furiously saving my thoughts in text files on Dropbox but I never seem to get around to publishing them. So I spent a weekend hacking on a project to make publishing content as easy as writing it.

Separate to that I’d already begun working on a new business. In doing customer research with many technical founders and marketing people we realized they had similar publishing problems, and wanted to use my weekend project.

The road to what people wanted

What became clear as I worked with people and ported my own sites across was that publishing content from any device wasn’t enough. It needed to look great. It needed to promote engagement. It needed to help that content spread. And most of all needed to help drive meaningful value for everyone.

They’re all things I care about too. But it has now become core to this new CMS and content delivery platform. How do I talk about what I’m doing in ways that resonate with my target customers?

Brainstorming headline ideas

The new headline — the one I found in Amazon reviews — brought in >400% more clicks Joanna Wiebe, Copy Hackers

I opened up a spreadsheet and got to work brainstorming headlines and other copy. I took a lead from @copyhackers and read Amazon reviews for inspiration. I found books about content management, books about content marketing, and looked at the language the reviews used to praise these services. I looked for common problems they had that remained unresolved.

And then I mashed them all together in my spreadsheet into as many coherent variations as possible.

Hacking Twitter Ads for copy validation

I had some ideas, but how to test which ones worked? I’ve used services like Optimizely before to split-test page copy but I didn’t have any traffic to test with. I also didn’t have the patience to build that traffic.

So I created a new “Leads on Twitter” campaign.

One of the main benefits to me at the time was I didn’t have drive people to my website (which wasn’t particularly well optimized). I could test all of my ads as different variations on Twitter itself. Work out which version of the copy produced the most engagement. And if a few people signed up to be on the waitlist then that’s just icing on the cake.

But all I was really interested in at this stage was testing the copy, not lead generation.

Here’s a screenshot of just a few of those ad variations:

Turning on the firehose

I’ve had some mediocre responses with Twitter advertising in the past. Not that it’s particularly expensive or has poor ROI but mostly the emotional pain of low engagement rates. Which really shouldn’t surprise me, it’s pretty rare I click no an ad in Twitter myself. There’s a constant stream of things flashing by and it’s competing with a lot of other things for attention.

In 2 days these 37,035 people saw these tweets. And it cost $0.

So I invested some time in refining my targeting. I was looking for a particular types of customer personas, within a specific industries, with particular types of jobs and work to be done. So I focussed on a subset of interests. Targeted followers of specific accounts. Targeted specific countries.

I tried to filter out as many Twitterers as possible so that I in turn was less likely to be lost in the noise.

Now one of the benefits of the “Leads on Twitter” campaign is that you only pay for a “lead”. That is someone that fills in the embedded Twitter card and provides their email address. But people can still click images, links, or otherwise engage with the tweet.

And they did.

Now the total numbers aren’t that impressive, but that was never the point. I wanted to weigh up the relative engagement of different copy options. And as you can see people were clicking though on the tweets:

In just 2 days these tweets were seen by over 37,035 people, it sent 140 people through to the site. And it cost $0.

Here’s the proof:

And the outcome?

In just 2 days I quickly tested 17 different copy variations with a lot of Twitterers. There was no clear winner, but there were differently a lot definite losers that I can exclude from future testing. The Twitter ad manager also gives a good insight into the type of people that clicked on the tweets. There’s some obvious patterns around:

  • Followers of certain accounts were more likely to click than others
  • Almost 75% of ad engagement came from mobile devices

All great insight for when I start running a lead gen campaign, to ensure the full journey through the on-boarding funnel is optimized from start to finish for their experience and expectations.