How Did You Hear About Us? Options, Survey Design, and Self-Reported Attribution

Your "how did you hear about us?" field is probably lying to you, and the fix is design, not more tracking. Designing a self-reported attribution field that works means treating that one line on your form as a product: exact wording, deliberate placement, a plan for non-answers, and a repeatable way to turn free text into reportable data. A well-designed self-reported field will tell you more about where pipeline comes from than another month of tuning UTM naming conventions. (This guide covers the field on B2B demo, contact, and signup forms, written for demand gen teams, not the version on job applications.)

Here is what a broken field costs. A prospect hears about you from a peer in a Slack community, searches your name a week later, and lands in your CRM as "Google" or "Direct." Multiply that across a quarter and your demand generation strategy gets built on the channels that happened to be last in line, not the ones that created the demand.

What a Self-Reported Attribution Field Actually Fixes

Definition: A self-reported attribution field is a question on a form or in a sales conversation that asks buyers, in their own words, how they first heard about you. It captures the source the buyer remembers, rather than the source your tracking software recorded.

Tracking sees clicks: the ad, the email link, the converting page. It misses everything between people: a podcast mention, a screenshot in a private channel, a booth conversation, a former colleague's recommendation. Many marketers call this layer the dark funnel, and it matters because B2B buyers now do much of their research before they ever talk to a seller.

As a result, every stage of your marketing funnel looks cleaner than it is. A buyer persuaded in conversation can type your URL and register as a plain visit, while Safari downgrades third-party referrers to their origin, so you see the domain but not the page (WebKit, Tracking Prevention in WebKit). The table shows the difference.

Notice the last row. Tracked data is good at the things it can see. The field does not replace it: it fills the rows where tracking reports a channel that did not actually do the work.

Why the Answers You're Getting Are Garbage (and How to Fix That)

If you already have a "how did you hear about us?" field, you have probably seen the answers: "Google," "internet," "not sure," and the classic "I've always known about you." I would not conclude self-reported data is worthless. Bad answers are almost always a design failure, and they come from three predictable causes.

1. The question arrives with no examples. A blank box invites "Google," which is technically true for most people.
2. The question is asked too late or too early. Ask it on a newsletter signup and nobody cares enough to think. Ask it three weeks into a sales cycle and the first touch is long forgotten.
3. There is only one generic prompt. "How did you hear about us?" asks about the first exposure. Many buyers honestly cannot remember that, but they can tell you what made them act today.

The fix is to give people a shape for their answer and a second chance to answer well.

• Seed the placeholder with examples. Use placeholder text like "e.g., a coworker mentioned it, saw a LinkedIn post, heard a podcast."
• Ask a trigger question on the first call. A rep who asks "What made you decide to reach out today?" usually gets a sharper answer than the form ever will.
• Treat "I've always known about you" as data. It usually means brand awareness built over time. Tag it as "Brand, long-term" rather than discarding it.

Where and When to Ask the Question

Placement matters more than wording. The best version of the field on the wrong form still produces junk, because answer quality tracks buyer intent. Someone requesting a demo has a reason to be thoughtful.

Demand gen teams I talk to are increasingly skeptical of friction anywhere before the buyer sees value. One put the principle plainly:

"We're very demand generation focused and we try to gate as little as possible in order to get a lot of interaction with our content." - [VP Marketing, HR tech]

That is the right instinct, and it tells you where the field belongs: on the high-intent moments that happen after trust exists, not at the front door. Use this framework:

• Demo request form: Always ask. Intent is highest, and the buyer expects a short qualifying step.
• Contact sales form: Always ask, optional, placed after the required fields.
• Live discovery call: Always ask, in the first five minutes, as a trigger question. This is your quality check on the form.
• Newsletter or content download: Skip it. Low intent means low-effort answers, and the extra field costs you signups.
• Free trial signup: Ask only if the form is already short. Otherwise, move it into the first in-product welcome step.

How Did You Hear About Us Options: A List for B2B Forms

If you use a dropdown or quick-select chips, the option list decides what you can learn. The list below is a starting point for B2B demo and contact forms. Pick six to eight that match the channels you actually run, keep "Other (please specify)" at the end, and add a free-text follow-up so the specific podcast, person, or community is not lost.

Two notes on the list. First, the AI assistant option is new on most forms, and it is worth adding: tracked referral data from AI tools is often incomplete, so the buyer's answer is a useful cross-check on your AI referral traffic in GA4. Second, review your "Other" answers every quarter. When the same answer keeps showing up there, promote it to its own option.

Field Design: Free Text vs. Dropdown vs. Hybrid

Most advice on this topic says "use free text, not a dropdown" and stops. A dropdown lists the channels you already believe in, so buyers pick the nearest option and the data confirms your assumptions. Free text lets them name the podcast, the person, or the community you did not know mattered.

But pure free text is messy to report on, so I would run a hybrid on most B2B demo forms.

For a free-text field, use this copy:

Label: How did you hear about us?
Placeholder: e.g., a coworker mentioned it, saw a LinkedIn post, heard a podcast
Helper text: Optional, but it helps us know what is working.

For the hybrid, show five chips: Referral, Search, Social, Event, Other. When any chip is selected, reveal a single line that asks "Tell us a bit more (who, where, or what)."

On required versus optional, I lean optional. A required field gets more answers, but some are typed to clear the form, and checkout research ties each extra field to friction (Baymard Institute, 2024). Optional fields with good placeholders get fewer, better answers, and discovery calls catch the rest.

Designing a How Did You Hear About Us Survey

A form field is not the only place to collect self-reported attribution. A short "how did you hear about us" survey, shown right after signup, after a demo is booked, or in the first in-product welcome step, can collect the answer without adding a field to a form you are trying to keep short. The design rules are the same, with a few survey-specific ones.

• Ask one question, maybe two. "How did you first hear about us?" and, optionally, "What made you sign up today?" First exposure and final trigger are different questions, and a buyer can often answer only one of them well.
• Decide single-select or multi-select up front. Single-select forces a choice and gives cleaner reporting. Multi-select reflects reality better, since most B2B buyers touch several channels, but it makes channel totals add up to more than 100%. Pick one and keep it stable, so month-over-month comparisons mean something.
• Keep the wording neutral. "How did you hear about us?" is neutral. "Which of our webinars did you attend?" is leading, because it assumes the answer.
• Always include "Other (please specify)" and "I don't remember." An honest "I don't remember" is better data than a random pick made to move on.
• Show it at a moment of momentum. Right after a signup or booking, the buyer has just decided to act and the reason is fresh. Weeks later, it is not.
• Store the answer on the contact record. A survey answer that lives only in a survey tool never gets compared against tracked source. Map it to a dedicated CRM property, such as "Self-reported source."

Whichever format you use, form field or survey, the goal is the same: a consistent, stored, reportable answer in the buyer's own words.

What to Do With the Answers Once You Have Them

This is where most teams give up: six months of free text, four hundred unique answers, and nobody looks again. The workflow below fixes that.

1. Export monthly. Pull every answer with the lead ID, tracked source, and opportunity stage.
2. Build a two-level taxonomy. Level one is channel (Referral, Content, Paid, Event, Community, Brand). Level two is detail, such as "Referral > coworker" or "Content > LinkedIn post." Keep naming simple and consistent.
3. Classify with an LLM. Use a prompt like this one: "You will receive a list of answers to the question 'How did you hear about us?'. Classify each answer into exactly one channel and one detail from the taxonomy below. If the answer does not fit, return 'Unclassified' and a short reason. Return a table with the original answer, channel, detail, and confidence (high, medium, low)."
4. Route mismatches to review. When tracked source says Paid Search and the buyer says "a coworker told me," do not overwrite anything. Flag the record for a monthly reconciliation review.
5. Hand context to sales. A buyer who says "my old boss used you" is a warm referral, and the rep should know before the call. Feed the answer into your sales qualification process as context, not as a score.

There is a second variant worth stealing: a self-report question for your own team at closed-won. One solution consulting leader told me how he measured the impact of a specific asset when CRM tracking fell short:

"Basically, whenever a deal closed, I would ask, did you send a storylane? And then the second question was, was the story lane the primary Demo tool, and that's how I was able to do attribution." - [former solution consulting director, time-tracking software]

That is the same idea pointed inward. Two required questions at closed-won gave him attribution his tools could not.

Asking the Question in Conversation (Full Disclosure: This Is Us)

Storylane RepX is an AI agent that sits on your website and talks to buyers while they are engaged: it answers questions, pulls up relevant interactive demos, and qualifies buyers through conversation instead of a form. I am including it here because it changes where the "how did you hear about us?" question can live.

A form asks the question at a single moment, cold. A conversation can ask it the way a good rep would: after the buyer has gotten value, in natural language, with a follow-up when the first answer is vague. If someone says "Google," the agent can ask what they searched for or who mentioned you. And because RepX sends qualified leads and conversation summaries to HubSpot, Salesforce, and Slack, the answer reaches your CRM without adding another field to your demo form.

This matters most for teams that keep product access ungated. Buyers are clear that the hand-raise is the signal worth measuring:

"Our CMO is very much going to be looking at once the person finished the demo or the slew of demos, did it lead to them putting their hand up and going, I want to talk to an account executive." - [senior product marketing manager, compliance SaaS]

Where RepX does not fit: if your traffic is low, your forms convert well, and your reps already ask the trigger question on every call, a well-designed form field plus discovery discipline is enough. RepX earns its place when you have enough inbound volume that buyers are leaving without ever talking to anyone.

Reconciling Self-Reported Data With Your Tracked Attribution

Here is the rule I would tattoo on every marketing ops team: never cut or scale a channel based on self-reported data alone. The field is a second opinion, not a verdict.

One VP of Marketing described checking a modeled attribution platform this way:

"So I like sometimes to just take raw data from the different platforms and compare it against each other." - [VP Marketing, HR tech]

Apply the same discipline to your two sources. Agreement means you can act with confidence. Divergence tells you where to look, and it is often the most useful output of the whole exercise, especially when you layer in intent-based marketing signals to see what buyers were researching.

One more check: audit where your tracked source values fall back to defaults. If a workflow fails silently, leads land in a catch-all like "Direct," and that bucket will look like your best channel for all the wrong reasons.

Common Objections and How to Answer Them

You will hear pushback when you propose this. Here are the three objections I hear most.

"People misremember, so why bother?"
Recall bias is real: people misremember when and where they first saw something, and in my view buyers credit the most memorable touch, not the first. The alternative is tracked data that files word of mouth under Direct. A rough human signal beats a precise-looking wrong one, if you reconcile it.

"Won't this hurt form conversion?"
It can, if you make it required, put it first, or add it to low-intent forms. Make it optional, place it after the primary fields, keep it to one line, and only use it on demo and contact forms.

"Our CRM already records original source. Isn't that enough?"
Original source records the first tracked touch, which is exactly the data that files word of mouth under Direct and podcasts under Organic search. Keep it, and store self-reported attribution in a separate property next to it. The value comes from comparing the two, not from picking one.

A Field-by-Field Template You Can Copy

Three templates, one for each moment: a form answer, a conversational check, and a record for reporting.

Template 1: Demo request form

• Field label: How did you hear about us?
• Quick-select chips: Referral / Search / Social / Event / Other
• Follow-up line (appears after a chip is picked): Tell us a bit more (who, where, or what)
• Placeholder: e.g., a coworker mentioned it, saw a LinkedIn post, heard a podcast
• Required: No
• Position: After name, work email, and company

Template 2: Contact sales form

• Field label: What brought you here today?
• Placeholder: e.g., a friend recommended you, read a comparison, saw you at an event
• Helper text: Optional, but it helps us know what is working
• Required: No
• Position: Last field before submit

Template 3: Live discovery call script line

• Opening question: "Before we dive in, what made you decide to reach out today?"
• Follow-up if vague: "Was there a person, a post, or a moment that put us on your radar?"
• CRM field: Self-reported source (verbatim), logged by the rep during the call
• Timing: First five minutes, before any demo

Build the third template into your standard agenda for sales discovery calls so it happens every time, not only when a rep remembers.

Frequently Asked Questions

What options should a "how did you hear about us" field include?
Six to eight options that match the channels you run, such as a colleague recommendation, search engine, AI assistant, LinkedIn or social, podcast or webinar, online community, event, and "Other (please specify)," plus a free-text follow-up for detail.

Should I still ask "how did you hear about us?" if I have good tracking?
Yes. Good tracking tells you which clicks happened, not which conversations persuaded someone to click. The self-reported field covers the untracked influence your tools cannot see.

What if most people say "I don't remember"?
That is a design signal, not a data verdict. Add example answers to the placeholder and ask a trigger question on the first sales call. Answers like "I've always known about you" should be tagged as long-term brand awareness.

Should the field be required or optional?
Optional, in most cases. Required fields produce more answers but also more junk typed just to submit. Optional fields with strong placeholder text produce cleaner data.

How often should I review the taxonomy?
Review classifications monthly and the taxonomy itself every quarter. Add a new category only when the same answer type shows up repeatedly in the Unclassified bucket.

Can I use an LLM to classify free-text answers?
Yes, and it is the fastest way to make free text reportable. Give the model a fixed taxonomy, ask for a confidence rating, and hand-check the low-confidence results before they reach a dashboard.

Sources

• Anonymized Storylane customer and prospect conversations, 2026 (source of every quote above).
• WebKit, Tracking Prevention in WebKit (webkit.org/tracking-prevention)
• Baymard Institute, Checkout Optimization: 5 Ways to Minimize Form Fields in Checkout, June 2024

Want to see how RepX asks the source question in conversation? Book a demo of Storylane RepX.