At a glance
Asking buyers how they found you can reveal recommendations and shared resources your tracking misses. Preserve their exact answers, use consistent categories and show how many people didn’t answer. This guide helps marketing operations teams add that evidence to reporting without treating memory as a complete journey record.
Your CRM attributes an enquiry to search. The prospect tells your sales team that a colleague recommended you first. Those accounts don’t have to conflict: one describes a recorded visit, while the other describes what the person remembers leading them there.
Before adding “How did you hear about us?” to a form, decide what you want to learn. Where someone first encountered your business, what prompted today’s enquiry and what helped them evaluate you are different questions. Your report should preserve that distinction.
Use the answers alongside the campaign history you already hold. They can reveal shared resources, recommendations and unanswered buyer questions. Keep the original wording so your team can investigate those details, rather than reducing every response to a channel label.
From raw answers to a defensible content decision
Consider four illustrative answers: “a colleague sent the budgeting guide”, “Google”, “your LinkedIn post, then the website” and “I don’t remember”. Preserve all four. Code the first as a reported colleague referral with a named resource, the second as unspecified search, the third as multiple reported influences, and the fourth as unknown recall.
The first response gives the content team a useful question: what part of the budgeting guide helped the colleague explain the decision? The second may need no further interpretation. The third should not be forced into one winning channel. The fourth belongs in response-quality reporting, not in a guessed acquisition category.
If several independently collected answers name the same resource, inspect the resource and ask appropriate follow-up questions about how it was used. That can support a decision to clarify or expand a particular section. It does not supply a revenue allocation or show that all readers had the same experience.
Our reporting methodology connects observations with recommendations. Apply that discipline here by recording the raw evidence, interpretation and proposed action separately. Try the exercise with your own responses and compare interpretations before agreeing the reporting categories. Use the accompanying counts and non-response rate to show how much evidence the decision rests on.
Decide whether you’re asking about discovery, timing or influence
Discovery and influence are different questions. “How did you first hear about us?” asks about remembered discovery. “What prompted you to contact us today?” asks about the immediate trigger. “Which information helped you evaluate us?” asks about decision support.
Don’t combine all three into one field and later treat the answers as comparable. Choose a question appropriate to the moment and the decision the data will inform.
Pew Research Center’s questionnaire guidance describes how wording and response options affect answers. Keep the question neutral and provide an honest route for people who don’t remember or don’t know.
An open response can capture unexpected sources but requires coding. A fixed list is easier to summarise but can omit useful answers or steer the respondent. A short list with an open explanation can work if the options are carefully designed and the form remains usable.
Keep the buyer’s own words
Store the exact wording, response date, question version and collection point. Add the coded category in a separate field. Never replace the raw response with the analyst’s interpretation.
Keep blank, unknown and unusable responses distinct. A blank may reflect non-response. “I don’t remember” is a valid answer about recall. An unrelated entry is a different data-quality issue.
Include the associated person and account identifiers where appropriate. If several people at one account answer, retain their perspectives separately rather than selecting the most convenient response as the account’s single truth.
A practical coding guide
| Raw response | Suggested code | Interpretation limit |
|---|---|---|
| A colleague sent me your article | Colleague referral; content mentioned | Doesn’t reveal how the colleague found it |
| Search, unspecified | Doesn’t distinguish paid, organic or the query | |
| Your LinkedIn post about account research | LinkedIn; content topic recorded | Doesn’t establish whether it was the first exposure |
| We used you at my previous company | Previous experience | May concern a different use case or organisation |
| A conference conversation, then your website | Multiple reported influences | Don’t force a single source without a stated rule |
| I can’t remember | Unknown recall | Keep in the denominator for response-quality reporting |
These example rules are a starting point. Define how multiple influences are counted and whether categories can overlap. If they overlap, percentages may sum to more than 100%; label the chart accordingly.
Use a second reviewer on a sample of responses. Disagreements reveal where category definitions need improvement. Don’t silently revise historic coding without recording the rule change.
Report answer quality alongside channel shares
Dreamdata’s published test examined 100 demo submissions and classified 49 as containing useful answers. Its small, company-specific design doesn’t establish a universal response rate, but it shows why answer quality and missingness need reporting alongside channel shares.
Check the denominator before sharing a headline such as “40% came from referrals”. Does that mean 40% of all enquiries, everyone who answered or only the answers you could classify? Show the counts. Otherwise, a neat percentage can conceal how little of the audience you’ve heard from.
The study also compared self-reports with tracked account journeys. That is a useful design idea, provided recorded data isn’t assumed to be complete ground truth either.
Why the denominator changes the story
Imagine a fictional cohort of 120 enquiries. Eighty people answer the question, and sixty answers can be coded meaningfully. Eighteen of those sixty mention a colleague recommendation.
Referral mentions represent 30% of usable answers, 22.5% of answered forms and 15% of all enquiries. Each calculation is correct, but they answer different questions. Report the usable-answer share and the total cohort so readers can assess the gap.
If one of those respondents also has a paid search interaction, retain both records. Their colleague may have prompted the search. You don’t need to choose a winner between the form answer and the tracking system to learn something useful from them.
If the exact sequence matters, ask a suitable follow-up question during a conversation. Record the clarification as new evidence, not as permission to reconstruct every missing touch.
Keep CRM reconciliation simple
Create separate fields for reported discovery, reported trigger, raw text and coding version if those questions are collected. Keep tracked source and campaign history in their existing fields.
Link the response to the relevant conversion event. Someone answering again six months later may be describing a new reason for contact. Overwriting the earlier answer would erase useful context.
Give ambiguous cases a review status. Don’t let an automated classifier turn “a friend mentioned something on social” into a confidently named campaign. If classification is automated, retain the raw text and inspect a sample for systematic errors.
Use the answers to investigate what helps buyers
Look for patterns worth investigating: resources that people share, communities they hear from and gaps in tracking.
Treat the answers as clues for decisions you can investigate. Repeated mentions of a useful resource may justify improving it or making it easier to share. They don’t tell you the exact budget that resource deserves or how much revenue it independently caused.
Combine the findings with your demand generation measurement framework and with account-level evidence. Review unanswered and ambiguous cases as part of the result rather than treating them as an inconvenience.
Talk to ABM Logic about campaign reporting if you want buyer explanations, tracked activity and follow-up context to support the same commercial decisions.
