Self-Reported Attribution: A Practical Guide
· Cole Stark · 7 min read
Self-reported attribution means asking customers how they found you, usually with a "How did you hear about us?" field on your signup or demo form. It catches the channels tracking software misses: podcasts, creators, communities, word of mouth and AI assistants. People misremember, so it doesn't replace tracked attribution. Used together, the two show far more than either alone.
Below is how to set it up, how to analyze the answers, where it goes wrong, and when a tool is worth paying for.
What is self-reported attribution?
Self-reported attribution is a way of crediting marketing channels using what customers say, rather than what tracking records. You ask one question at a meaningful moment, like signup or a demo request, and store the answer with the contact.
Software attribution is the opposite approach. It reads clicks, UTM tags and referrers, then credits channels by rule. My guide to marketing attribution models covers how those rules work and when to use each.
The two answer different questions. Software attribution says which link someone clicked. A self-report says what they remember as the reason they came.
Why software attribution misses so much
Many of the moments that make a B2B buyer try a product leave no click. A colleague's recommendation in Slack, a podcast on the commute, a creator's video watched on a phone, a ChatGPT answer that named you. When that person later types your URL or searches your name, analytics records direct or branded search.
Consent choices, ad blockers and device switching remove more of the trail. My guide to the challenges of marketing attribution covers each gap. Self-reported attribution is the cheapest way to see into them.
How to ask buyers how they found you
Where to ask
Ask once, at the moment that matters most to your business:
- On demo request forms, for sales-led products.
- On signup or in the first onboarding screen, for self-serve products.
- In the first sales call, as a backup, logged in the CRM by the rep.
One place is enough to start. Asking on every form annoys people and splits your data.
Free text or dropdown
Free text gives truer answers. People write "a video from a Rust YouTuber" or "my CTO", which tells you far more than a category. The cost is cleanup: someone has to sort the answers. That's the trade-off of any open-ended question.
Dropdowns are easy to report on and easy to bias. People pick the first plausible option, a pattern related to the serial-position effect, and a list can't include a channel you haven't thought of.
I'd use free text for any company under a few hundred signups a month. Above that, a short dropdown with an "Other" text box works, as long as you randomize the order and include "Don't remember".
Required or optional
Make it required on demo requests, where people expect a few questions. On a free signup, test it. A required field adds friction, and an optional one gets skipped by people in a hurry.
When Dreamdata, a B2B attribution software company, tested a self-reported field on 100 of its own demo requests in 2022, about 70% filled it in and 49 gave a useful answer. Many simply wrote "Google". Expect a similar share of vague answers and plan your categories around them.
Store it with the contact
Save the answer as its own property on the contact in your CRM, next to the first-touch UTM source and landing page. Copy it to the deal or opportunity, so you can report on pipeline and revenue by self-reported source later.
Self-reported attribution analysis
Raw answers aren't useful until they're grouped. Self-reported attribution analysis starts with a short list of categories, like search, AI assistant, podcast, creator, community, referral, event, social and other.
Then work through the answers:
- Map each answer to a category, keeping the detail in a second field, like the creator's name. AI does a good first pass, but check it weekly.
- Report by category each month: share of signups, share of activated users, share of pipeline and share of revenue.
- Put the self-reported source next to the software source for the same contacts. Where they agree, you can trust both. Where they disagree, the self-reported answer usually reveals the channel that started it.
- Watch the trend, not the exact share. Answers are noisy, but a creator category that doubles after a sponsorship is a real signal.
The comparison in step 3 is where most of the value is. If tracking says "direct" and people say "podcast", your podcast spend is working and your dashboard can't see it.
Where the answers go wrong
Self-reports have known biases. Treat the data as directional:
- People remember the last thing, not the first. Someone who read three of your posts and then saw a creator video will often say "YouTube". That's recall bias, and it favors recent, vivid touches.
- "Google" hides the real source. Search is often how people come back after hearing about you somewhere else.
- Small numbers swing. With 40 answers a month, one podcast mention can look like a trend.
- Self-reports credit channels, not spend. They tell you a channel matters, not how much each extra dollar returns.
None of this makes the data worthless. It means you weigh it with the tracked numbers and with branded search trends, rather than on its own.
Combine self-reported and software attribution
The best setup uses both, for different decisions:
| Decision | Use |
|---|---|
| Daily bids, budgets and ad tests | Software attribution |
| Which channels to fund next quarter | Self-reported and software side by side |
| Whether creators, podcasts and communities work | Self-reported, checked against branded search |
| Reporting revenue by channel to a board | Both, with the gap between them explained |
This matters most for channels that rarely produce clicks. My guide to how to measure influencer marketing uses self-reported answers as one of three layers for exactly this reason. For developer tools, it's often the best view of community and creator work, which I cover in developer marketing metrics.
Self-reported attribution solutions and companies
Most B2B startups don't need software for this. A form field, a CRM property and a monthly spreadsheet cover it until answers number in the thousands.
When you outgrow that, self-reported attribution solutions fall into three groups:
- Your CRM or form builder. HubSpot, Salesforce and most form tools can add the field, store it and report on it.
- Post-purchase survey tools for ecommerce. Fairing, for example, runs "How did you hear about us?" surveys after Shopify purchases and ties answers to order value.
- Attribution platforms that blend answers into channel reporting. SegmentStream, for example, classifies free-text answers with AI and re-credits conversions that tracking recorded as direct, brand or organic.
Before you compare self-reported attribution companies, check that the tool can export raw answers. You'll want them for your own analysis whatever the vendor's dashboard says.
Self-reported attribution shows what tracking can't
Self-reported attribution is the cheapest measurement fix most B2B startups aren't using. Add one question, store the answer, sort it monthly and read it next to your tracked data. It won't give you precise credit, but it will show you which channels people remember, and those are often the ones your dashboard undervalues.
Setting this up alongside tracked attribution is part of my growth analytics work. For where it fits in the wider measurement setup, see my marketing measurement plan guide.
Self-Reported Attribution FAQs
What is self-reported attribution?
Self-reported attribution is crediting marketing channels based on what customers say about how they found you, usually through a "How did you hear about us?" question on a signup or demo form. It captures channels that tracking misses, like podcasts, creators, communities and word of mouth, and works best alongside software attribution.
Is self-reported attribution accurate?
Self-reported answers are directional rather than precise. People tend to remember the most recent or vivid touch, and many answer "Google" when search was only how they came back. It's still the best view of channels that produce no clicks, so compare it with software attribution and branded search rather than relying on it alone.
Should "How did you hear about us?" be free text or a dropdown?
"How did you hear about us?" works best as free text for most startups, because people name the specific creator, podcast or colleague. Dropdowns are easier to report on but bias answers toward the first options and miss channels you didn't list. A short dropdown with an "Other" text box is a good middle ground at higher volumes.
Should "How did you hear about us?" be required?
A "How did you hear about us?" field should usually be required on demo request forms, where people expect several questions. On free signup forms, test it, because a required field can lower completion. Either way, include a "Don't remember" option or accept short answers, so people don't type nonsense to get past it.
How do you do self-reported attribution analysis?
You do self-reported attribution analysis by grouping free-text answers into a short list of channel categories, then reporting each category's share of signups, pipeline and revenue every month. Compare the answers with software attribution for the same contacts. The disagreements show which channels start journeys that tracking credits elsewhere.
Does self-reported attribution replace multi-touch attribution?
Self-reported attribution doesn't replace multi-touch attribution. Software attribution is better for daily optimization of tracked channels like paid search, while self-reported answers show the untracked channels that start many journeys. Use both, and treat the gap between them as information about your marketing rather than an error to eliminate.