29 Jan 2014

You Want To Crowdsource Your Marketing? Here’s My Presentation To SMX Israel’s Give It Up (1/3)

I had the privilege of talking to SMX Israel’s massive, crazy advanced audience for the SEOs Give It Up session this Sunday, and shared my experience and ideas on how you can crowdsource a lot of your marketing. If you weren’t there, you missed an amazing conference – honestly, I learnt more there than probably any other show I’ve been to perhaps except individual panels at SMX Advanced or Pubcon – but you can make it up in part by reading my slides and notes (see below) on crowdsourcing your marketing.

Before talking about crowdsourcing your marketing, I briefly introduced Amazon Mechanical Turk (MTurk) and its functionality.

I gave some advanced tips to getting people with specific demographic or other attributes to do your tasks, so that you can gain better insight or work quality. For example, MTurk doesn’t allow you to exclude anyone, which is a problem if you’re re-running a task you already had some Turkers do. So in such a case, create a one field form, ask for the Turker’s ID, and compare it against those of Turkers who have previously done the work. If it matches, decline them from doing the work and if it doesn’t match, allow them to proceed.

Let’s get to the meat of the preso.

Marketing follows a cycle: research -> campaign -> analysis -> campaign. Each step in that cycle features work that you can crowdsource. Here’s what I have done and suggest you can do as well.

Crowdsource Your Market Research

Before you spend a second on keyword research, you have to understand your audience’s problem(s). What pain will you alleviate for them? And you have to know who the target audience is, in terms of demographics and influencing factors.

Market research begins with intent. It does not start with keywords. You need to get in your customer’s mind. You use interviews to get to this intent.

A key problem in arranging market research interviews is finding potential customers to interview. MTurk has a huge population of workers, roughly representative of the US population. So you can usually crowdsource interviewees in your target audience.

In your interview, your primary question must be “What is the biggest problem preventing you from {achieving your goal}?” Whatever market you’re tackling, substitute the audience’s goal (or your best guess as to that goal). For example, “What is the biggest problem preventing you from losing weight?” “What is the biggest problem preventing you from ranking higher?” “What’s the biggest problem preventing you from being productive in your home office?” (The idea for these problem interviews is stolen from Ash Maurya’s Running Lean, about lean startups, but the application to internet marketing will be clear in a second. He calls this first series of interviews “problem interviews” as opposed to later “solution interviews.”)

When you start your research by asking people this question, something magical happens. It’s called empathy.

(Different from sympathy, which is showing care for or pity for another, but not necessarily feeling what they feel, which is empathy.)

You’ve probably read about creating personas to represent your audience. The first thing you need to do to create personas is to develop empathy. Therefore, your persona building must start with problem interviews.

I can go on with other tips on how to do these interviews, but Justin Wilcox already did an amazing job explaining problem interviews. Here are some (not all) questions from his script:

  1. What’s the hardest part about [problem context] ?
  2. Can you tell me about the last time that happened?
  3. Why was that hard?

The result of all this is a lot of qualitative data on various problems. You’ve got lots of notes, perhaps some audio or video recorded.

To make sense of it all, read through the notes and look for the common problems that people struggle with. What patterns are there?

Armed with this knowledge, you proceed to the next step of research, surveys. The point of these is to rank your problems from most painful to least painful, which you do by collecting quantitative data.

Here’s the goal for internet marketing:

I describe each common problem I’ve noticed from the problem interviews, and ask survey takers to rate it from 1-5 (0-5 would be more logical, but I’m limited by Wufoo’s Likert scales), where 1 is no problem and 5 is a mega-pain.

Then you just export your survey data and average the scores for your problems.

If problem A has an average rating of 3/5, while problem B averages 4/5, your landing page and ads should obviously address problem B in the headlines.

Guess what? Rather than waste money split testing your guesses about what problems the audience has, you jump start your CRO with headlines that have an extremely high likelihood of being your top converters. And obviously you also save the inhouse time and effort to produce those split tests.

It gets even better.

Segmenting is obviously essential to higher conversion rates. (Hat tip ion interactive and aWeber for the lesson.) Know your audience and personalize your appeal, right? Dads have different interests than moms etc.

In your surveys (perhaps also in interviews), ask demographic questions. There are the standard general questions like age and gender, but each market will also have its own particularly relevant questions.

For instance, if you’re selling B2B, a person’s job title obviously indicates to you whether they’re a likely buyer or buying influencer. Or if you’re promoting a rehab clinic, you need to address potential patients differently than their families.

When you have your spreadsheet of survey response data, you can segment your problems by demographic issues. Taking our earlier example, perhaps problem B’s average 4/5 rating is due to guys rating it 3/5 and girls rating it 5/5. Problem A, meanwhile, is a 4/5 problem for guys but 2/5 problem for girls.

Suddenly your conversion strategy changes. Your traffic targeting will split according to gender, and guys will gets ads and landing pages talking about problem A, while girls will get ads and landers talking about problem B. Obviously, your convert rate benefits enormously.

(A way to cross-reference your own surveys and interviews to check for possible errors is to have the crowd read and summarize Quora (qualitative data) and forums (qualitative), as well as Wikipedia, the US Census site and Statsbrain for quantitative data. )

Note: I’m phrasing this all in the language of ads, but obviously it’s relevant to social media links and SEO, too.

Next time: Crowdsource your campaign creation


Leave a Reply