A Practical Guide to User Needs

Julia Mitelman
On Products
Published in
12 min readMay 6, 2015

As the old saying about the Model T goes (which may or may not have actually been uttered by Ford), “if I’d asked my customers what they wanted, they would have said a faster horse.” So then, how exactly do you find what customers want — or better, what they need?

As product managers know, all you really need is the scientific method. The following is a comprehensive starter kit for turning observations into potent product plans.

Looking for the digest version? Check out the Pocket Guide to User Needs.

The Hypothesis

The most critical factor to creating a great product, rather than just a mediocre one, is asking the right questions.

  1. What are the user’s goals?
  2. What are the obstacles to this goal?
  3. What are the user’s values and priorities?
  4. What is the user’s context and how much can it vary?

Thorough understanding of these questions usually leads to only a handful of truly appropriate (and often oh-duh-how-has-no-one-done-this-before) solutions.

Finding what you don’t know you’re looking for

All products start with observations. Product managers are always looking for opportunities. Some good starting points include:

  1. People watching and dogfooding: what problems do you encounter regularly? Pay attention to those around you; inquire about the experiences of friends, colleagues, and strangers in coffeeshops. If you’re evolving an already existing product, look for inefficiencies. If you’re starting fresh, look at products already available in the space you’re curious about. Question choices and behaviors.
  2. Hacks and workarounds: what are people using atypically? People often work around problems themselves; these are clear signs of issues that might be solved better directly. Ask what the motivation was for the hack and what inspired the solution. Functional fixedness is a tendency that biases us to only use things for what they are clearly designed; users who overcome this were likely compelled by a need, so identify it. Keep difficulty in mind — the harder the workaround is to duplicate or the more creativity it took to come up with, the greater the opportunity.
  3. Industry trends, competitors, and peripheral offerings: what are people excited about? Users are interested in tools that improve their lives, so listen to the buzz. The mere existence of competitors suggests an opportunity, so figure out what problems they’re trying to solve and whether you can do better. Examine how you might create a complementary product to what’s already out there, filling in gaps to complete a fragmented story.
  4. Science fiction: what have those without the bounds of reality imagined to inspire people? While the solutions offered in such fictional worlds may not be possible yet, the underlying problem they solve is likely very real and present today.
  5. Gut and empathy: follow the scent of possibility and let curiosity run you into corners you’d otherwise have never noticed. Your instincts will often tell you when a product isn’t quite right, when a problem isn’t quite solved, or when a vision isn’t quite pointed. Be fearless — run into dead ends, constantly question whether you’re still headed in the right direction, and change paths frequently. This ensures eventual success.

Synthesizing the hypothesis

Armed with observations, you’re ready to interview users. First, you need a prompt: an area of interest to explore that will act as the starting point for your conversation.

The prompt should be specific enough to create a clear purpose but general enough to avoid tunneling your vision. It should have a hypothesis — users have trouble with X and would utilize Y — but, unlike a typical scientific study that validates or rejects one theory, you should be prepared to zoom in and out of the area as you pinpoint the right details.

Here’s a practical example: I once worked at a startup that brought live streaming cable to the browsers of college students. Basic channels were free, but premium ones required a subscription. Some students were using the service but few were signing up. We brainstormed two new offerings: to strengthen the community aspect, we’d add social media integration, e.g. a live chat stream shared with friends; to accommodate student schedules, we’d add some form of DVR. These prompts grounded my questions with particular user goals to validate and contexts to test.

The Interview

Talking to even just a handful of users can provide rich, qualitative data that steers the product in great directions. But not all data is equal: what and how you ask can make or break the quality of your results.

Form matters

These methods, taken from psychology methodology, help produce accurate and effective data.

Avoid leading questions

Commonly known and yet highly misunderstood, a leading question steers the participant towards a particular answer. Avoiding such questions is not intuitive because we routinely employ this technique to get what we want, as part of daily social cues.

Unless their feelings are extreme, most people will default to agreeing with you. This is due to observer-expectancy bias. So when you ask, “was this task difficult” vs. “was this task easy”, you’ll get the same distribution of “yes”, because people subconsciously tell you what they think you want to hear. Instead, ask “How was the task?”, or if that’s not clear enough, mention both states: “Tell me about the ease or difficulty of this task.” This also eases the framing effect, which causes you to get different answers to the same tests based on how the questions are phrased.

When possible, make the question agnostic of the user. Avoid using phrases like “What did you think?”; opt for “Tell me about this.” instead. This lessens pressure on the user and helps downplay tendencies to overestimate personal performance (overconfidence bias), falsely justify choices (choice-supportive bias), and even take on more extreme views to explain behaviors (attitude polarization).

Be mindful of cognitive bias

Biases will sway and box users in if you don’t manage them. The interview should be relaxed and free-form, so you won’t be able to prep the entire sequence ahead of time, and that’s okay. Take thorough notes so that even if you don’t notice a bias during the interview, you can reflect later and improve on the next round.

Some common ones to keep in check:

  1. Confirmation bias (as well as selective perception and Semmelweis reflex) sways users to look for confirmation of what they already know or believe, blinding them to other possibilities. During interviews, this can compound with a tendency to treat things that come first to mind as most important (availability heuristic). These can box a user into a particular way of thinking, such as recollecting only a subset of situations in which they encounter a problem. Counter these issues by reminding users about the possibility of alternatives, asking them to take the perspective of others they know, and giving them ample opportunity to provide an exhaustive answer.
  2. Order of information and choices heavily influences the user. First, anchoring creates over-reliance on the first piece of information given. The mere exposure effect biases the user to like ideas simply because they’ve been repeatedly exposed to them (that’s why you like that terrible song on the radio). Users tend to forget the middle of a sequence (serial position effect), and thus may discredit its value. When the user has a choice between two options and then a third is introduced, the decoy effect causes an asymmetrical change in opinion (they might change their mind or might even more strongly prefer their original choice). Users are also fighting belief revision bias and sunk-cost fallacy, which makes them hesitant to change their minds once they’ve already made a choice or investment. Pay attention to changes in opinion when introducing new options; vary order when possible.
  3. Memories are highly biased; a user’s recollection of experiences is still very valuable, so cater to their needs, but take it with a grain of salt. People recall behaviors better when physically in the relevant context (cue-dependent forgetting), tend to remember unusual things (Von Restorff effect), forget negative emotions more quickly than positive ones (fading affect bias), and judge an experience on the average of its peak and ending rather than the whole (peak-end rule).
  4. Self-evaluation of needs, values, and behaviors is greatly biased, so do not over-rely on this. This is why users can’t just tell you what they need or how they will react to a product. People tend to favor outcomes they can predict over unknowns (ambiguity effect), discount the value of future rewards (hyperbolic discounting), overestimate predictability of past events (hindsight bias) and control of current events (illusion of control), overestimate how long a choice will affect them (impact bias), create false associations (illusory correlations), discredit extremes or edge cases (normalcy bias), and underestimate the length or difficulty of future tasks (planning fallacy).

Picking users

Alright, so you’re ready with your prompt and knowledge of cognitive biases. Time for the bravest part — walking up to a stranger and starting a chat.

Most of us don’t have access to a research lab (and for those that do, the self-selection bias may limit the usefulness of this anyway), so where do we find people?

Start by figuring out who you’re looking for. What is the demographic of your ideal customer? What does a power user or a casual user look like? Who are your competitors’ users? What type of location is the most appropriate context for your prompt (e.g. the gym, coffeeshop, library, etc.)?

Be smart and respectful. Stick to public places unless you have the explicit permission from establishment owners. Before you get into the prompt, explain that you’re trying to develop or evolve a product and ask the person whether they’d spare a few minutes (yes, keep it short) to educate you on their opinions. Explain that they will be in no way associated with this product; you’re hoping to create great experiences, so their perspectives would greatly help.

Be ethical. Always make it clear to the person you’re talking to that you’re working on a product. The tone can (and ideally should) remain casual, but once you start gathering data and analyzing, you need to make your intentions clear.

Be brave. If tapping a stranger on the shoulder is intimidating, that’s okay; start small. Interview your friends and then ask to talk to their friends. Talk to your coworkers. Talk to familiar strangers — your regular barista, receptionists, and shop assistants (though be mindful not to distract them from their work). If you’re evolving a product and already have customers, go find a few at a store (if applicable) or throw an event.

Allons-y!

Preface

To ease the burden on users and ensure high quality data, be sure to go over the fundamentals:

  • There are no right or wrong answers. You’re looking to understand current behaviors and choices to find areas ripe for improvement.
  • It’s always the designer’s fault (in industry terms, that’s not true, but users don’t need to know the nuances of the various job titles). Remind them explicitly that any issues you come upon are not the user’s fault.
  • This is an exploration. You’re not looking for any particular answers or support for ideas. You’re looking for perspective from another’s shoes.
  • Even the mundane is valuable and of interest to you. Ask the user to help by overcommunicating their thoughts and experiences.

Question techniques

Let’s start with an example — those prompts from the TV startup. I approached students sitting in common dorm areas to determine whether the social media integration and DVR features would work. The first student I approached, Sara, was already a casual user of the product.

I warmed up with questions about her favorite TV shows; the question was easy and gave me a sense of the importance of TV in her life. Next, I focused on context — when did she use the product and what did she get out of it? Sara let the stream play in the background while she was studying, to keep her from getting too distracted. She would tune back into it when she was tired or bored, watch briefly to relax, then get back to studying. Sara enjoyed the passive nature of the stream — she wanted to “just be a vegetable” and not think about what to queue up. Social media features were useless to Sara, since this would require breaking that passive nature. With her busy schedule, she always streamed her favorite shows later from Hulu. She like the sound of a DVR feature, but knew that she’d never pay for it when free alternatives were so readily available (plus, she’d have to sift through commercials). Next, I asked Sara how she had discovered the service — it was during an Oscars viewing party with friends. They’d had a number of parties for live events, a popular campus practice.

TV was still very much relevant to Sara, but its main appeal was now the curation of content, rather than access, for which there were plenty of superior alternatives. I began to hear this value over and over again as I interviewed more students. Neither social media integration that required active participation (e.g. live chat) nor DVR would have benefited users. A personalized hashtag overlay during major live events, however, worked spectacularly to create a campus-wide conversation during the Superbowl. The product also found new focus — curation. Next round of interviews included prompts around features that would reveal what shows and channels were trending among peers and automatic channelsurfing that would control the remote for you based on what you’re most likely to enjoy.

The interview stages consist of:

  1. The warmup: start with easy, fun questions. Your user is doing you a favor with this chat (though don’t sweat it too much; this makes them like you more, thanks to the Ben Franklin effect). Make them feel comfortable. Keep it light, short, and simple.
  2. Understanding context: start painting a mental picture of the environment your user would be in when facing the issues you’re exploring. Ask for lots of details; nothing is too mundane to inspire. Notice both the content of the description and the type of information shared — this hints at what users are noticing vs. what’s working well enough or is unimportant enough that they don’t even recall it. Find out which factors are in the user’s control and what determines the rest; find out how much the environment varies for that user and between users. Pay extra attention to behaviors, choices, or elements that break your expectations; how were they different and why?
  3. Understanding values: for every behavior the user mentions, figure out the motivation behind it. Become a 6-year old: ask why until you get to the user’s values. Find out the relative prioritization of these values and how they fit into the user’s life overall (e.g. perhaps they won’t upgrade their washing machine because it has physical knobs instead of a touchscreen and their young daughter will twist them and it’s important to include her in family chores so she learns responsibility and independence). Pay extra attention to actions that don’t map to expressed values (e.g. user says he loves cocktails but always goes to the same bar and orders the same three drinks; he’s actually unfamiliar with many of the ingredients and too embarrassed to ask the bartender or look it up in front of his friends).
  4. Zooming in and out: as you learn more about a user’s needs, spend enough time on each element to sniff out the presence of opportunities, and move on to other aspects if there aren’t many. Practice maintaining a balance between depth and breadth to get the most value from each interview. Your gut will develop here with time. Initially, err on the side of depth — knowing where there aren’t opportunities is more valuable than leaving a chat without having really learned anything.

Wrapping up

Keep it short, no more than 15 minutes total, unless your user continues sharing without your influence.

Once you’ve gained some perspective, be sure to take a breather. Thank your user for their time and help and offer them a way to contact you for followup in case they have questions or come up with more thoughts. Clarify (with the user) any part of your notes where you may have skimmed over a detail you wanted to capture.

Immediately following the interview, look back over your notes and fill in the blanks. You won’t remember what happened or what you meant. Write down all the nuanced details; you’ll be thankful for them later.

The Analysis

Give your brain some time to unwind and process (no, really, it will allow your default mode network to synthesize the data), then review your notes.

First, identify the biases. Were there leading questions or social situations that might have swayed the user? This doesn’t make your data invalid (unless you happen to be publishing an academic paper); simply keep in mind how this might impact your findings.

Now, the culmination:

  1. Organize observations on the user’s environment, values, and behaviors.
  2. Extract their goals and reapply these to their context: what obstacles get in the way, and thus what are the user’s needs? Which needs are unmet, and what is the relative pain of these?
  3. Pinpoint the best opportunities for impact.

Note down these findings and brainstorm potential solutions through features (guide to brainstorming techniques coming soon). Analyze the potential of these solutions based on the opportunities and the relative business impact — a need cannot be fixed without a viable product. Don’t take anything off the table if you don’t yet see profitability, but do start to consider revenue as one of the many factors in determining the solution’s potential; it’s not too early for this and may even help expand your thinking.

Rinse and repeat

Figure out what data you need next. What would you like to learn more about? What do you need to validate further, with multiple data points? How should you evolve your prompt based on your findings, to reflect new ideas?

Find more users and keep testing. Let your product morph and evolve and shift. Once the rate of change slows to a steady crawl, you’re probably holding a solid product opportunity. You’re ready for the next type of user study: the wireframes and experience prototypes. More on that later.

--

--