So many businesses fall into the trap of doing research once, then thinking they’ve solved the problem of research. But since design research is the only way to reliably create revenue-generating A/B test ideas, you’re going to need to do it often.

There is no way for me to find a shade of gray in this: researching only once is wrong. Why?

• Your insights aren’t double-checked. Perhaps you found a handful of people who were power users, not representing the majority of customers. Perhaps you interviewed a bunch of unserious tire-kickers. Either way, you’re not clearly vetting who you’re talking to.
• Research goes stale as your market changes, competitors come and go, and your customers’ attitudes to technology develop. And it goes stale faster than you think.
• Research typically answers a narrowly-scoped question. Think: how many questions can research answer for the problems you specifically face? For each question, there’s a separate, unique research project waiting to happen.

Here are some of those questions, generically framed:

• How do customers buy our product?
• How can we follow through after a sale?
• How do customers vet competitors?
• Why do people prefer us over competitors?
• How long is the sales timeline? Is it a matter of minutes (cheap B2C stuff) or months (enterprise)?
• During that timeline, what questions are prospective customers asking?
• What navigation paths do people take through our funnel?
• What proportion of customers takes each path?
• Is there anything – demographic, behavioral, etc – that differentiates each of them?
• How and when do customers move from free plans to paid plans?
• Where do people pause most frequently on the checkout form?
• Where aren’t people focusing their attention on a given page?
• What’s going wrong with the product?

All of these change over time, and each one represents a different focus of research. And that’s a lot! And you don’t know the answer to any of them.

So, let’s talk about how to establish a routine of research – which provides revenue-generating test ideas into perpetuity, while giving your business a complete portrait of your customers.

Establish Questions

First, you need to establish what questions to ask. I wouldn’t worry about covering every single base here, but good questions should always:

• Connect to revenue generation in some capacity. Put another way, answering the question should provide insights into paying customer behavior.
• Address something the business is questioning. Perhaps you think you already know the answer, and want it confirmed. More likely, though, you have only a premonition, and have spent a lot of time speculating.
• Explain either observed or expressed behavior. Observed behavior is what someone actually does; expressed behavior is what somebody says they do. And the two rarely match.

Come up with 7-10 of the most pressing; you don’t need to do a full audit of questions right now. That should keep you plenty busy!

Match Research Methods to Each Question

Each question needs a way of answering it. For each question, figure out a corresponding research method. For example:

• How do customers buy our product? Interview 3-5 customers on the phone and run a survey of all past customers.
• How can we follow through after a sale? 10 days after purchase, send customers a form that asks them about their experience – and if they need any additional help.
• How do customers vet competitors? Why do people prefer us over competitors? Interview 5-7 customers on the phone.
• How long is the sales timeline? During that timeline, what questions are prospective customers asking? Interview 3-5 prospective customers who haven’t purchased yet, and survey everyone who has.
• What navigation paths do people take through our funnel? What proportion of customers takes each path? Use Google Analytics to assess this.
• Is there anything that differentiates each of them? Survey people from each branch, if you can track their email addresses. (Mixpanel is great for this.)
• How and when do customers move from free plans to paid plans? This requires a combination of Google Analytics, surveying, and 3-5 interviews.
• Where do people pause most frequently on the checkout form? VWO has this built into its behavior reporting functionality. You can also use Hotjar to assess similar behavior.
• Where aren’t people focusing their attention on a given page? Get heat & scroll maps of each page of your site, using either Crazy Egg, VWO, or Hotjar.
• What’s going wrong with the product? Interview 3-5 people on the customer support team, and survey active customers 30 days after they’ve signed up.

Throw each of these questions, and their corresponding research plans, into a Trello board. This will help you track when you last answered the question, and it will help you prioritize what to research next.

Plan New Research Every Week

Go through one of these questions (and its corresponding research methodology) and put it into practice. Wondering how to do that? The A/B Testing Manual provides a deep dive into each one.

Synthesis & Testing

Once you get your results back, it’s time to synthesize the research you put together. Here’s our tutorial on synthesis, which you should read if you haven’t seen it yet.

The result of synthesized research is at least the beginning of an approach to a new design decision on your funnel. And that’s something you should test.

Asides

There are a few more things to note here:

Audit Questions Monthly

Finally, every month, you should go back into your Trello board, reprioritize the next few weeks of research, and add any questions (and corresponding research plans) that have popped up lately.

This is when you get a chance to be more comprehensive with your research plans! What are you missing right now? What’s come up as a result of existing research? Do any questions need to be clarified? Are any subsets of your customers being neglected in your research plans?

Resourcing

Ideally, you’ll have a full-time researcher on staff, solely focused on conversion. More realistically, this requires at least a half-time resource to do well – otherwise, you won’t be as intentional or strategic about your research. Fortunately, there are lots of independent consultants (ahem) who do this fairly well.

Also Read

Wren Lanier has a great piece about establishing a weekly research cadence that you should read, especially if you work on a SaaS. It’s quite different from the method I prescribe, focusing more on interviewing & customer support. There are many ways to skin this proverbial cat!

← Back to the Blog Check Out Draft’s Store →