How to Use Surveys to Build Audiences and Reach Your Customers
Jump start an audience and position yourself as an expert
So you want to get people interested in something you’re working on — a software product, a book, an event — but you don’t have a big audience, or you just want to make yours bigger. One of the fastest ways to jump start an audience is to run a survey.
People love to know what their peers are doing. Everyone has questions that they are maybe too afraid to ask in a public setting, but want to know the answers to. A great survey can quickly spread across your target community if people are interested in the answers and trust that you’re going to present their data in a thoughtful and interesting way.
But done poorly, you could spend a lot of time spinning your wheels or get accused of having a biased agenda. (Think of the absurdly one-sided surveys run by both Democrats and Republicans).
Over the last few years, I’ve created and launched surveys that have reached thousands of individuals across three very different sectors: Asian American men, tech recruiters and candidates, and collegiate esports teams. I’ve also advised others on this strategy, including journalist and author Allison Yarrow, who is surveying new moms for her second book, “The Mother Load.”
All told, these surveys have reached thousands of people and have helped jump start audiences for personal and professional projects. Here’s how you can use them too.
Disclaimer: While some of what I’m going to share are strategies lifted from user research best practices, this kind of survey work is not meant to replace more rigorous research methods. It is first and foremost a marketing strategy which generates data that is informative, if not perfectly representative.
1. Choose an Audience Segment
Who are you trying to reach? Choosing the right audience segment is critical to making the survey hyper relevant, and thus shareable. If your segment is too general, people won’t share it, or they won’t know if it’s meant for them.
Examples of segments that are probably too general:
Examples of segments that are more targeted:
- Meal-kit subscribers
- Moms with kids 0–18 months old
- Recreational runners
These are groups of people who can easily identify themselves as being in or out of the segment, know other people like them, and are likely are involved with communities based on this identity or interest.
Seriously though, don’t worry about being too specific.
Sometimes, you might even pick an audience segment that’s meant to be a subset of a larger audience. For instance, a survey of Airpod users could be used as a way to reach the broader, but more general, Apple consumers or tech early adopters.
If you have a big audience you’re trying to reach, you can always split it into sections and run a follow-up survey for each part of the audience.
When starting my company Midgame, an esports and gaming analytics company, I realized there was this emerging field of collegiate esports that no one was really talking about. So, I knew that starting a survey for this audience segment would get people to share it with others, since it was probably the first one they had ever seen targeted towards them.
2. Define the Survey Focus
Once you have your audience, you still need to go further and focus on a particular aspect of this audience. There’s usually some polling by Gallup or the Pew Research Center on this segment, so general questions aren’t going to be that interesting (unless your category is very new — like “Airpod users” or “Gen Z college grads”).
For instance if we’re going after those meal-kit subscribers, we might focus on a theme like preparation and consumption. So the questions might be directed in this way:
- Are the meals easy to make?
- How long do they really take to prepare?
- Do you actually have the right kitchen tools?
- How good do the meals taste?
- Do they give enough food / too much food?
I’m not a meal-kit subscriber, so I’m spitballing here, but you can see where this is going. These questions are of course different from a focus on packaging and delivery, or pricing or alternatives to the meal kit (cooking yourself / takeout).
The point is that you can’t make a comprehensive survey that covers every aspect of this group’s life, so picking a focus is important. For the purposes of building an audience, you want to pick a focus area around issues that are controversial, taboo, niche, or otherwise less known.
For the Asian American Man Study, I asked a lot of questions around dating—preferences, dating history, and very specific questions like, “How many times has someone told you ‘I don’t date Asian men’?” These were issues that were top of mind for many guys, and they were also questions they couldn’t find the answers to anywhere else.
3. Write Your Questions
Start your survey with three to five demographic questions. This would usually include things like gender, race, age range (e.g. 18–24, 25–34, etc), income bracket, industry/field (if it’s a business survey) plus a few context-dependent ones.
For the meal-kit subscribers survey, you might ask how long they’ve been subscribers and what size plan they’re on. For the mom’s with kids under 18 months, it might include if they have family support for childcare, or whether they are a stay-at-home mom versus working outside the home.
Try to keep your core survey to 15–25 questions that are mandatory (except for the open-ended ones). Make sure it can be done in five to seven minutes, or else you’re going to lose too many people along the way. Make most of these questions multiple choice. This means knowing the common answers people might have, which might mean running your questions by someone else.
Consider asking a few questions that deal with a controversial or divisive issue within your community. For the collegiate esports survey, I asked about feelings on NCAA’s involvement in the field — a topic that many have strong feelings on. Topics that have strong emotional content get shared more.
The inverse of the divisive question is the question that might be known to insiders, but not to outsiders. For the Asian Man study, I asked, “How many times have you heard someone say ‘I don’t date Asian men’?” This is a phenomenon that Asian men experience frequently, but comes often as a surprise to others. People tend to share surprising findings.
I’ve used Google Forms for most of my surveys, but Typeform can be a good choice too. I tend to not like the look of SurveyMonkey, but lots of folks seem to use them, so there must be something there.
So much of good survey work is about how you word your questions. This is really important, so don’t skip over this. Here are some basics:
- If you want to get frequency on something, ask how many times it has happened in the last day, week, or month. When you ask how often something usually happens, people give different, more idealistic answers, than what actually happened.
- If you need people to give you a number, try to make it a multiple choice question with ranges. Otherwise, later you’re going to have a hell of a time trying to crunch everything by hand. Things like age, income, meals eaten per week, can all be expressed as ranges. You may need to do some pre-research to figure out what an appropriate range might be.
- If you want to get a sense for someone’s opinion on an issue, try saying “Please indicate how much you agree or disagree with the following statement.” Then include a relatively neutral statement. For the recreational runners, maybe a statement would be, “I enjoy running in competitions,” or “I feel safe running at night.” Then use a five, six, or seven point scale. I usually go with five because it’s easiest: Strongly Disagree, Disagree, Neutral, Agree, Strongly Agree
- If you want to get a feel for motivation, you can ask people to select from a checkbox list of reasons why they do/like/use something. “Please indicate which of the following reasons for running apply to you personally.”
- Try to keep the open-ended questions to two or three max. These are hard to analyze and are more to illustrate a particular perspective rather than hard numbers.
There’s a great book for this called “The Complete Guide to Writing Questionnaires,” by David F. Harris, that I highly recommend if this stuff is interesting to you.
These take longer, but can occasionally reveal really interesting things. Try to not do more than two of these, because they really slow down responses. Also, move them closer to the end of the survey and keep them optional.
Some open-ended questions I’ve used to good effect:
- What do you want people to know about [survey focus or audience segment]?
- What’s the best thing about being [audience segment]?
- What is the one thing you’d like to change about [survey focus]?
4. Set Up Your Systems
Finished your questions? Great! But you’re not out of the woods yet. You’ve got to set up your promotional systems. This is where the marketing part of this tactic comes into play.
Think about it like a Kickstarter project: the survey is only interesting if a lot of people take it. I would shoot for a minimum of 100 responses for an audience segment that is generally reachable (e.g. recreational runners vs. NBA players).
Set a deadline
People are motivated by deadlines and missing out on things. I recommend setting a deadline of two to three weeks — not super long, but enough time for you to promote it pretty hard.
Make sure you get people’s email addresses — if you’ve written the survey well, they’ll want a copy of the results. Make sure it’s clear they’re going to get the survey results and the occasional update on your project when asking for their email.
If your survey is on an issue that many consider private, like sexual habits or relationship woes, you might choose to avoid asking for their email on the main survey. Instead, provide a link at the end to another form that’s just collecting emails for “People who want to see results.” This allows them to disconnect their answers while remaining involved.
Add them to your list
If you use Google Forms or Typeform for your survey, you can use Zapier free at low volumes. Set up an autoresponder that either adds these people to a mailing service, like Mailchimp (also free at low volumes), or directly reply via Gmail.
Ask people where they found you
Make this an open-ended text field that’s not required. Occasionally, you’ll get a gem: some influential person shared it, or it was posted in a private newsletter or mentioned in a podcast.
These are places you can cultivate later on for promoting survey results. They also are great places in general to market to your audience (since they were able to send people to your survey).
(Optional) Name the survey and make a custom domain
If you want to go really pro here, you can name the survey. If you have a relevant brand in the space, you can name it after yourself (e.g. The KIND Bar Healthy Snacking Survey).
Most of mine have been named something generic (e.g. The State of Tech Hiring Survey), which is meant to imply definitiveness, but is also just because we didn’t have a relevant brand to lead with.
5. Promote the Survey
Alright! You’re finally ready to get the word out about this bad boy. Time to promote in every way you can.
1. Send an email out to all your friend / relevant contacts
Tell them about the goals of the survey and the kinds of people you’re looking to reach. Ask them to share it with the right people and communities.
2. Post on your social media channels
If you can, tie it to some important event, like a book announcement, a funding announcement, a new product launch. Even if you can’t tie it to an event, make sure to announce the survey in a big way.
3. Add it to your email signature and social media profiles
Find a way to reference the survey, and change whatever link you normally have in your bio on Twitter and Instagram to the survey. Add a line in your signature about the survey and when it closes.
4. Work your list
When people complete the survey and sign up for updates, you can send them three emails.
- Instant responses: Thank people for taking the survey and ask them to spread the link, retweet on Twitter, or share on Facebook. Everyone has different sharing preferences, so give them options.
- Updates: As time goes on, you can take a peek at early results and give a sneak peek to respondents — everyone loves knowing about things before everyone else (This is considered bad form from a scientific research perspective, but remember we are marketing first here!). This then serves as a reminder and motivation to share again. Depending on how long you leave the survey open, you can do this up to two times.
- Final push: Like a Kickstarter, you get to make one final push of your respondents, asking them to promote before the deadline. If you don’t have a lot of responses, you can announce you’re extending the deadline. In any case, definitely leave the form active for a few days after the official deadline to capture stragglers.
Shoot for at least 200 responses. Less than 100 means something went really wrong in either the survey link, the targeting, or the promotion. Over 500 is wonderful, and over 1000 (especially if you don’t already have a big list) is amazing.
6. Crunch the Numbers
Now you have to actually figure out what this data means. Don’t be nervous. Even if you aren’t super strong in math, I’ll give you everything you need.
1. Get that data into a spreadsheet
I know Google Forms gives you some basic charts, but you don’t want to present your hard won data with such blandness. We’ll talk about presentation in the next section. Anyway, with Google Forms, there’s a link to see the data in a spreadsheet. I personally use Google Sheets, and my formulas are set for that. But, they are usually based on Excel, so you should be good if you’re using that.
2. Protect the original data
The danger of a spreadsheet is that it’s easy to overwrite the data. So, start by duplicating the sheet and protecting the first one so if you do mess up, you can easily recover the data.
Do all your analysis on the copy of the original sheet. If things look super weird mid-way through your analysis, see if you deleted anything, and, as a last resort, re-copy and paste the original data and start over.
3. Freeze the top row and word wrap the text
This makes sure that wherever you are in the spreadsheet, you’ll be able to see the questions you’re working with. Word wrap makes sure you can see the data inside each cell instead of just letting it run over.
4. Create a filter on the data
Select the first row and drag all the way down to the last row with actual data in it. Then go to Data→ Create A Filter. This will allow you to remove certain results or sort the results from high to low or A to Z. You’ll know the filter is on when you see three horizontal lines on each first row cell, which you can click on to do things.
5. List out all possible responses for each question in your survey
In this step, you are setting up the things you want to chart. For each multiple choice question, you need to list each possible answer for the questions at the very bottom of the survey under the responses.
If there’s an order you think makes sense (No kids, one kid, two kids, three or more kids) then list them in that order. Put the items one column to the left of the data for that chart. Then do the next column to the right lower, so you leave space for the data, and alternate. See the image for reference. You’ll see why we do this in the next step.
6. Generate data for each response
OK, here’s where the formula magic comes into play. You’re going to count up the number of people who gave a particular answer to a question. Then, you’re going to divide this number by the total number of responses for that question. Simple enough, right?
For instance, in the image above, we are trying to calculate the percentage of people who said they were 18 to 24 in our survey out of all respondents.
A simplified version of that would be: People aged 18–24 divided by All respondents.
So that first part —the number of people who gave a particular answer (the numerator) — comes out with this formula:
Let’s explain each of those elements shall we?
- =COUNTIF — a formula where you list a range and a value, and the formula counts the number of times that value appears in the range.
- DATACOL — Data column, which is the letter of the column in which the data range you are trying to count is in. In the example above, it is D.
- $ — Formulas have a special property where if you copy it from one cell to another cell above/below or left/right of the original cell, the formula will adjust relative to its new position. Sometimes this is useful, but sometimes you need an absolute position, so you use the $ symbol to indicate the following element stays put.
- DATA1STROW — Data 1st row, which is the number row of the data range you are trying to count begins. In the example above, and in most cases, it will be two.
- DATALASTROW — Data last row is the number of the final row in the data range you are counting. In the case above, 664 people took my survey so the final number is “665” since we started not at one but two.
- ANSWERCOLANSWERROW — the letter of the column and the number of the row containing the answer you’re looking for. The way we’ve set this up, it’s C668
The second part, the total number of responses to a given answer, is simpler.
You could just literally go with the number of people who took the survey and call it a day (in our case 664). But if you made questions optional, then some people might not have filled these out at all and so the percentages won’t add up to 100%. You solve that by using a different formula:
- =COUNTA is a formula that just counts if there’s anything in a given cell range at all. If it’s blank, it counts nothing. If it has anything — a number, a piece of text, etc — it counts one. This makes it great for counting total number of responses.
Inside of the parentheses of COUNTA is the same range we used in the first part. It should look familiar.
So, going back to the example scenario I laid out. Here is the actual formula as it appears inside the cell.
Once you have this formula, you can drag it downwards to copy it into the cells below. Those cells should automatically populate with the correct percentages. Remember to use the $ symbols!
You can then also copy a cell and paste it into another cell for your following question, and it should work correctly. Just click the cell once to select it, hit CTRL-C or -C to copy it, and then click on the cell you want to move it to and hit CTRL-V or -V to paste it. Google Sheets and Excel will automatically update the formula for this new column or row.
Don’t directly copy the text of the formula from the bar at the top, or else the formula won’t update and the data will be incorrect.
7. Use the formula to generate all your data
Making sure the formula works is the hardest part. Then it’s just manually copying and pasting all the answer responses and generating the percentages.
8. Optional — create data segments
There’s one last step you can take if you want to slice and dice the data further. And that’s to create data segments.
This is a more advanced analysis, but it can reveal incredibly interesting results. Maybe you’re doing a parenting survey and you want to see how mothers and fathers differ in the number of diapers changed per week. You then split the data into two parts, and then chart both segments.
Maybe parents tend to change 20 diapers a week on average. But, when you split out moms and dads, it’s more like 30 diapers by moms and ten diapers by dads. That difference becomes an interesting story and factoid, i.e. “Moms change 3x as many diapers as Dad’s”.
So to make this, you use the filter we created in step four. Go into the column you want to separate (for instance, the gender column in the parenting category) and click the green upside down pyramid until a chart comes up. Use Filter by values… Click Clear, then click the response you want to dig into specifically look at.
What will happen is the data you filtered out will disappear. It is just being hidden, so the data at the bottom won’t update. However, you can select all the data that is visible, copy it, create a new sheet, and paste them into a new sheet. You will see that only the filtered results came along. You’ll have to recreate the formula, since the total number of rows is less now.
Luckily, you only need to do it once. Then just make sure to copy and paste the formula for all the answers you want to see.
Make sure to then do the other part of the segment so you can compare (e.g. moms vs dads).
7. Create the Actual Charts
If you made a survey with only ten or so questions, you could probably make a chart for every question. But if you have 20 or more, you will have to be judicious. This means looking at all the questions and sharing the ones that are the most interesting or useful.
Some charts are good for creating a benchmark. For instance, the distribution of kids that our parenting survey has. The data might not be exciting, but it’s useful to just understand.
Some charts are useful because they are surprising or make a point. Maybe we discover that 50% of Airpods users are actually using Android smartphones. That might be interesting because you would assume that more of the userbase would have an Apple phone. Or the example of the diaper changing differential between moms and dads.
Here are a few ways to think about your charts:
You have to give people a picture of who responded and how they skew, since, as we identified up top, this is not a representative sample. Are your respondents skewing female or male? Wealthy? Older? On the West Coast?
If you were smart, you made these questions multiple choice. So, you can now easily make charts that show respondents broken down by race or by number of kids.
2. General responses
This is the meat of the survey. Questions that answer usage, or preference, or frequency.
In general, if you have two to four data series (e.g. types of answers — like student/teacher/parent), you can create a column chart where the blocks are upright. If you have five or more data series, or if the names of your data series are long, you’ll need to go with a bar chart, where the blocks go long.
As you can see in the chart above, I asked for years of experience. Since the data labels are short, we can do a column chart.
3. Segmented responses
This is what we talked about in the optional final step of the section Crunching the Numbers. In this example, we see that when we look at competitive esports gamers, the number of hours they train varies based on how serious they consider themselves. Very serious commitment gamers (cyan) are much more likely to train over ten hours a week (52%) compared to not very or somewhat serious gamers (21%).
Tips on making charts
- I use Keynote to make the charts, but you can use Google Sheets as well. Keynote gives you more flexibility on how your charts work.
- Give your chart a good title. It can be simple, like in the case above, “Esports Training Schedule.” Simply describe the data. Or, it can be more interpretive, like “Very Serious Gamers Train More Hours Than Casuals.”
- If the data is not obvious, try to include the original question as a subtitle or somewhere in the chart. Both Google Charts and Keynote allows you to create subtitles.
- Include the data values in the chart. This way, others referencing the data can actually get real numbers; they don’t have to just eyeball it based on where the bar is against the background.
- Sometimes, you will need to rewrite an answer to make it easier to understand/fit the space in the chart. In the Asian Man study, I sometimes shortened answers like “Asian men do not get adequate representation in media,” to “Lack media representation.” You are the expert here; don’t be afraid to summarize where it makes sense. Just be ready to explain if someone asks, and try to make sure your summary doesn’t deviate from the intent of the answer.
- Pick a set of colors for your chart. Use these same three to six colors across all your charts so they have a unified feeling.
- When doing a segment chart, make sure you include a legend so people know which color means what.
- Make sure to include your survey’s name and a link to its website, or your company’s site, or your name at the bottom, so that you get credit for this data. Expect that people are just going to embed the chart on their site or post it on social media, so make sure there’s a way for people to find you.
- Use column charts if you have a few answers and the answers are short in length (few words).
- Use bar charts if you have a lot of answers and/or the answers are long.
- Use pie charts sparingly, since people struggle to understand how to compare the segments. Donut charts (where there’s a hole in the middle) can make it easier, but as you will find, making room for all the data values can be annoying.
- Line and area charts can be good if you are tracking data over time (Revenue or user numbers by year).
8. Promote the Charts
You can do a few things here.
You can post the charts directly on social media. Make sure to space these out so you can get traffic on each one separately and start conversations.
And you can also create a PDF with additional data that you share if people give you their email. With something like Mailchimp, you can attach a link to download the PDF of the survey results as part of the welcome email.
Don’t forget to share the data with your list, and make sure to ask them to share the results as well.
Congratulations, you’ve done it. You now have proprietary data on a topic. This makes you an expert and provides you with a built-in list of people who likely trust you and want to learn more from you. If you have some compelling enough findings, you could even use this as the basis of an op-ed or published piece.
Make sure to reference this research in future work — you’ve earned it. And, consider doing a follow up survey down the road on something you didn’t cover initially, or that digs into a particular area you could only give surface treatment for.