"How did you hear about us?" Open text or a dropdown?

Pew Research Center tested the same question both ways after the 2008 US presidential election: "What one issue mattered most to you in deciding how you voted for president?" When the economy was offered as an option, 58% picked it. When people answered in their own words, 35% named the economy (Pew).

The bigger gap is in what fell outside the list. In the open version, 43% gave an answer that wasn't one of the five options. In the list version, where people could still volunteer something else, fewer than one in ten (8%) did. The list changed what people said.

Order matters too. Pew notes that in self-administered surveys, which is what an online signup form is, people tend to choose items at the top of the list. Shaifali Agrawal at Recast makes the same point about this exact question: people "are more likely to select the first visible option," and a drop-down "introduces biases and limits the range of responses" (Recast).

So if "Google" sits at the top of your dropdown, some of the people who heard about you from a friend and then searched your name may pick Google.

One well-known test of this question comes from Refine Labs, a marketing agency. Megan Bowen, its CEO, describes a required free-text "How did you hear about us?" field on their "Book a Strategy Call" form, with "no leading text or suggestions to prompt responses." They compared the answers with what their tracking software said for 620 form fills (Refine Labs).

The software said 78% of those came from web searches. In their own words, people named web search 12% of the time. Most named places the software can't see, which Refine Labs calls "dark social": social networks, podcasts, communities and word of mouth, all of which "happen out of sight of our attribution software." Those made up 85% of what people named. Some people named more than one place, so these are shares of mentions, not of people.

Two cautions. This is one agency's own form and its own buyers, so the numbers won't carry over to your audience. And typed answers have limits of their own. Recast points out that people struggle to remember where they first heard of a business, and recent touchpoints tend to win (Recast). A typed answer tells you what people remember. That's still worth knowing, as long as you read it that way.

A dropdown is easier to count. Clint Fontanella at HubSpot recommends multiple choice for this question because open text "will be more difficult to analyze later," and says to always include an "Other" option (HubSpot).

That's fair if thousands sign up each week and nobody will read a text box. Two catches:

• Your list has to be right first. The only way to know which channels belong on it is to ask openly for a while.
• The "Other" box only helps if you read it. Pew's 8% suggests people rarely bother to type when a list is in front of them.

Typed answers look messy until you label them. Read each one, give it a short label for the channel, and count the labels. Keep the person's words next to the label so you can check later. Recast suggests reading the first few days of answers to spot patterns before you set any rules, such as treating "LI" as LinkedIn (Recast). For the full method, see how to analyze open-ended survey responses.

The numbers below are invented to show the method. They are not from a real survey.

Say a small software product asks the question as open text at signup and gets 120 answers in a month:

Now say the old dropdown offered Google, LinkedIn, Instagram, A friend, and Other. Podcasts, communities and newsletters add up to 39 of 120 answers, about a third (worked out from the table). None of them had a place on that list. Instagram, which did, got no typed answers at all.

This is the step where Enform does the counting for you. The tour below shows it reading a batch of replies.

Use the counts to write the list, then keep asking openly in a smaller way.

• Order the list by count, not by habit. Or shuffle it, if your form tool can, so no option always sits at the top.
• Keep an open box for "Other". Ask "Where?" next to it, so a write-in names a place instead of "online."
• Add one follow-up for the vague picks. If someone chooses "A podcast" or "A community", ask which one. That's the answer you can act on.
• Run the open version again every few months. New channels show up, and a list can't tell you they did.

Keep the question short and optional. On Pew's online panel, open questions that ask for a word or a short phrase are skipped less than ones that ask for several sentences, with medians of 13% against 18%. The gap showed up for people on phones, not on desktops (Pew). "How did you hear about us?" already asks for a phrase, so don't add "please describe in detail." As for where to ask it, HubSpot suggests at checkout or just after a purchase, or on a signup or registration form.

Three ways to ask and count it:

• Google Forms with a spreadsheet. A short answer box costs nothing. Google Forms lets you add "Other" with a typed answer to multiple choice and checkbox questions, but its dropdown has no "Other" option (Google). Counting the typed answers in Google Sheets is slow past a few hundred. See Enform vs Google Forms.
• A general AI chat. Paste the column into ChatGPT and ask it to group and count. It's fast, but check the counts against the answers before you plan a quarter on them.
• A survey tool that reads the replies. Enform writes the survey from a description, you share one link, and it reads every reply and counts what people said. Every count opens to the replies behind it, so you can see what "some podcast" meant before you buy an ad there. It's built for typed answers. If you only ever want a fixed dropdown, any form tool will do. See pricing.

For why open questions are worth the extra reading in general, see open-ended vs closed-ended questions.