Evidence: How much is enough?

I recently tried to identify why the back brake on my bike was not working well. After taking the brake system apart, I discovered that the lever at the handle was badly bent out of shape. Reason for malfunction found! Could not be bent back, brake system replaced, done.

When you are running an organization, your problems are much larger. Which technology should a company invest in to gain a strategic advantage? Which product benefits should be highlighted to consumers? Which new features will users want in a year from now?

One way of making these decisions is going by your gut, or your industry expertise. But it is probably good to collect feedback from the world outside your building and do some (market / customer / user) research.

How much evidence do you need to address your business problem? Will a few user interviews do the trick? Do you need a survey or conduct some ethnographic research? Or is the information already available somewhere, in your in-house data bases?

Let’s start here: The research you conduct should be proportionate to your problem. If you are pondering a major change in strategic direction, you should probably take your time, allocate a considerable research budget and think deeply about which data you want to collect and how it will help you make your decision. In contrast, if you want to tweak a minor feature, a few quick user interviews or a small scale A/B test may do the trick.

Also, research to support business decisions is not an exact science. There are no standards for how much research a particular problem warrants. There is always more to be learned, always some doubt remaining about whether the evidence was sufficient. How do you deal with that?

It’s simple! Time and money are the key factors in determining where to draw the line in your research efforts. Business decisions need to be made in a timely manner, and if the strategy session with the CEO is in six weeks from now, your research needs to be done by then. Likewise, if you have a research budget of $30k you cannot commission a project that costs $100k. Business research is conducted in support of business goals and business decisions, and these are constricted by time and resources.


Sometimes, you want to recommend that the company push back its strategy session by a couple of weeks to allow for collecting more comprehensive information about the customer. Or you lobby for an expansion of the research budget so that the size of the research better fits the size of the problem.

If you are unsure if the evidence will be enough, work from the concept of understanding. After collecting your evidence, do you feel you understand the customer enough to move ahead with your business decision? Or is there something missing?

If you feel your understanding is insufficient to make your business decision with confidence, go back, collect more evidence. If there is no time / no budget, frame your research results with caution signs, so that their limitations are clear to all.

Often, organizations develop a sense over time of what constitutes an acceptable amount of evidence. Some are happy with 4-6 interviews to explore fairly major problems, others want a large-scale study. Some don’t trust qualitative insights at all and feel that a quantitative survey must be conducted to support important decisions.

Usually, research results are not perfect. They do not meet all your insights needs. They do not include all possible stakeholders. But they provide valuable real-world guidance from users and customers that counter-balance gut-based decision making. Now, gut and industry experience and business sense still play an important role and insights stemming from research are only a part of the bigger picture.

Plus, when research results come back, their relationship to your business problem is usually not one-to-one: brake lever bent = brake not working = install new brake. Rather, they are subject to interpretation and often beg further analysis.

Research is no replacement for business sense, vision and leadership. These must have the last word and take responsibility for the future path of the product or organization.