Self-Reported Attribution Surveys for B2B Pipeline Source Data

Most B2B buyers make up their minds before a salesperson, an ad, or a form ever finds out. They read a Slack thread, watch a LinkedIn post from someone they trust, ask a colleague what they use, and only then fill out a demo request. By the time that happens, the trackable trail is short and misleading. Self-reported attribution (SRA) surveys exist to catch what click tracking never could. This piece walks through what SRA is, what a large 2024 dataset from HockeyStack actually shows, and how to use the tool without asking it to do a job it was never built for.

Start with the structural problem. B2B buyers spend most of their journey anonymously, bouncing across channels that leave no cookie, no UTM parameter, no pixel fire. The touchpoints that build vendor preference (peer recommendations, Slack groups, LinkedIn posts, podcasts, community forums) tend to show up later as direct traffic or a branded search. That's not a tooling gap someone forgot to fix. It's a structural limit: influence that happens in a conversation or off-platform simply cannot emit a tracking signal. There's no code for "my friend mentioned it at dinner."

The consequence is predictable. Budget follows what's measurable, not what's actually working. Channels that capture already-formed demand (paid search, retargeting) get credited constantly. Channels that create the demand in the first place get credited rarely, if at all. And the gap is widening, not shrinking: Google's Consent Mode v2 and the EU's Digital Markets Act mean a real chunk of European traffic now arrives without consent for measurement. Attribution data isn't incomplete by accident anymore. It's incomplete by design.

That gap isn't closing. The real question is what fills it directionally, and that's where SRA comes in.

What self-reported attribution actually is and what it claims to do

SRA is simple to describe: ask the prospect how they heard about the company. Usually that's a field on a demo request form, or a question an SDR asks at the top of a discovery call. That's the whole mechanism.

It's worth being precise about what this data is not. SRA is not correlative, and it's definitely not causal. It doesn't measure marketing influence in any statistical sense. What it captures is memory and salience — what a buyer remembers and considers worth naming, at one specific moment.

That "one specific moment" detail matters more than it sounds like it should. SRA only surveys people who already decided to raise their hand. So the data reflects whatever pushed someone over the line, not necessarily what touched them first, or most, or best. A buyer who saw eleven LinkedIn posts over three months, then Googled the brand name and clicked a paid search ad, will very often answer "search." Recency beats sequence. Salience beats accuracy.

That's a real bias. But it's not a reason to throw the tool out. It's a reason to read SRA next to platform data instead of in place of it — a point worth holding onto, because the rest of this argument depends on it.

Implementation, by design, stays light: one open-text or short-select field on a form, or a consistent question SDRs ask early in a call. No new tracking stack required.

What the HockeyStack SRA dataset reveals about where B2B buyers say they come from

HockeyStack's 2024 SRA report pulled 8,528 responses from high-intent prospects (people who'd already filled out a "book a demo" or "contact sales" form) across dozens of B2B companies globally. It's the largest cross-company SRA dataset published so far, which makes the shape of the results worth sitting with.

Search engines took the largest share of self-reported sources, by a wide margin. No surprise there. But the second and third-place finishers are where it gets interesting.

Social media came in second, with 1,670 mentions — 20% of all respondents. Break that down further and LinkedIn dominates the category: 508 mentions, well ahead of YouTube (58), TikTok (31), Facebook (22), and Twitter (9). For anyone deciding where to put B2B paid social dollars, that gap between LinkedIn and everything else isn't subtle. It's the whole story.

Word of mouth came in third: 1,544 responses, or 18% of the total. That bucket covers colleague referrals, partner mentions, Slack recommendations, community posts, and specific people named by name. Here's the catch: this category is nearly invisible to any tool that relies on digital tracking. A referral almost always shows up downstream as direct traffic or branded search. Nothing about the referral itself leaves a trace.

Then there's the podcast number, which is the sharpest gap in the whole dataset. Only 12 of 8,528 respondents — roughly 0.001% — mentioned a podcast as their source. Meanwhile, 46% of the companies in the sample were actively producing one. That's not a rounding error. Either podcasts are influencing buyers quietly, without ever becoming the thing someone names when asked "how'd you hear about us," or a lot of companies are overestimating how much reach their podcast actually has. Both readings point to the same next step: use SRA to test the assumption, rather than assume the podcast is either a hit or a waste based on gut feel.

Zoom out and the dataset's overall shape says something bigger: most B2B brands are still winning attention later in the buyer journey than they'd like to think. Demand capture channels dominate the numbers. Demand creation channels are underrepresented, not because they don't work, but because they're structurally harder to notice and name.

Where SRA catches what multi-touch models miss: the dark-funnel signal

The same HockeyStack dataset shows the gap in sharp relief: community and dark-funnel channels show up constantly in self-reported answers and almost never in UTM-tracked source data. That kind of gap — community-driven pipeline that's real and substantial yet completely absent from standard digital attribution reporting — is precisely what SRA is positioned to surface.

Why does this keep happening? Because the mechanism is structural, not a fixable bug. A peer referral shows up later as branded search or direct traffic. No UTM parameter survives a Slack message. No pixel fires during a hallway conversation at a conference. The influence is real; the trail just isn't there.

And the stakes are higher than they look, because peer referrals in B2B tend to close at higher rates, land bigger deals, and move faster through the pipeline than tracked digital leads. Yet in any model that requires a clickable, trackable event to assign credit, referrals get systematically undercounted.

There's a second-order effect worth naming directly: brand search campaigns. When someone hears about a company through word of mouth, then Googles the brand name and clicks a paid search ad, that conversion gets credited entirely to paid search. The ROI on brand campaigns often looks fantastic as a result. Some meaningful share of that ROI, though, belongs to whatever generated the demand in the first place — the conversation, the community post, the Slack recommendation. Paid search just happened to be standing at the finish line. SRA is one of the only tools that can surface that dynamic and put a name on it.

The practical takeaway: teams that lean entirely on multi-touch or last-touch data will keep under-investing in the channels — community, referral programs, peer-driven content — that create the intent other channels later harvest.

How SRA fits into a B2B attribution stack without displacing what's already there

Here's the useful way to think about it: SRA and multi-touch attribution have opposite blind spots. Multi-touch sees every tracked click and misses everything untracked. SRA sees untracked influence and is biased toward whatever's most recent and memorable. Put them together, and each one covers the other's gap.

A workable stack looks like this: platform data (Google Ads, LinkedIn Campaign Manager) drives day-to-day optimization. CRM attribution handles pipeline and revenue reporting. SRA sits on top as a gut-check layer, surfacing what the other two structurally can't see.

Worth remembering: attribution model choice alone produces wildly different numbers. The gap between "marketing-sourced pipeline" and "marketing-influenced pipeline," measured on the exact same revenue, can span dozens of percentage points depending on which model a team picks. SRA doesn't resolve that model-selection problem. What it offers instead is an independent data point that doesn't depend on any model at all.

SRA earns its keep most clearly as a diagnostic — when platform data and CRM data are telling contradictory stories, or when a channel's spend-to-pipeline ratio looks weak on paper but keeps getting mentioned by buyers anyway. That contradiction is worth investigating, not ignoring.

And the privacy trend only strengthens the case for having SRA in the mix. As consent gaps widen and pixel coverage keeps degrading, more of the buyer journey becomes structurally untrackable by design. SRA doesn't care about cookies or consent banners. It just asks a question.

One caution, though, worth stating plainly: SRA can't prove causality, can't rank touchpoints by weight of influence, and shouldn't be the sole input into a budget forecast. Any team using a single SRA finding to justify a major reallocation is over-indexing on one data point. That's a misuse of the tool, not a feature of it.

Implementing an SRA program that produces usable data

Getting SRA right comes down to a handful of decisions, each of which is easy to get wrong.

The form field. Open-text fields produce richer, more honest answers. Dropdowns produce cleaner data that's easier to trend quarter over quarter. The practical move: start open-text, see what people actually say, then build a dropdown taxonomy from real responses instead of guessing at categories in advance. And keep the question itself neutral — "How did you first hear about us?" beats "Which of these channels brought you here?" because the second version anchors the respondent to a list someone else wrote.

The sales call question. SDRs asking "how did you hear about us?" at the start of a discovery call catches buyers who never filled out a form at all, and often gets a more detailed answer, since a conversation isn't boxed in by a text field. The catch: that answer needs to land in CRM every time, consistently. If it stays buried in a call note, it's not data. It's trivia.

Volume and aggregation. One answer is an anecdote. A few hundred answers over a quarter start to look like a pattern. Smaller programs may need to aggregate over six months just to get category distributions that hold still long enough to read.

Taxonomy maintenance. Real answers are messy. "A colleague," "our Slack group," "my boss mentioned it," and a specific person's name are all really the same thing: word of mouth. Normalize them into consistent categories, but keep the raw response stored too, in case the categories need rebuilding later. Revisit the taxonomy every quarter, since new channels show up constantly.

Where the data lives. SRA data belongs on the contact or opportunity record in CRM, not in a standalone spreadsheet nobody opens. Tied to the record, it can be sliced by segment, deal size, industry, or ICP fit. Disconnected from pipeline, it's just a curiosity.

Reading SRA output alongside pipeline data to make channel decisions

The wrong question is "what does SRA say is working?" The better question is: where does SRA disagree with the tracked data, and what does that disagreement actually mean?

Three scenarios come up constantly, and each one calls for a different response:

• SRA and platform data agree. High confidence. The channel is genuinely contributing, and both signals back each other up.
• Platform data looks strong, SRA mentions are thin. That's a flag for over-crediting — probably a last-touch or model artifact. Worth investigating before scaling spend further.
• SRA mentions are common, platform data shows almost nothing. This is the classic dark-funnel signal: the channel is influencing buyers, just not converting in a way anything can track. That's an argument for holding or increasing investment, even without a clean line from spend to conversion.

Three dashboard questions tend to organize the analysis well: Is awareness growing in the channels a company actually invests in? Are buyers finding the brand through those same channels, or different ones entirely? And is word of mouth growing as a share of total sources — a slow, lagging signal of brand and community health?

Segmenting by deal outcome adds another layer. Comparing SRA answers from closed-won opportunities against ones that stalled out reveals which channels bring in buyers who actually fit, not just buyers who show interest and go quiet.

The podcast example from earlier is a good template for how to use this data responsibly: 46% of companies run one, but 0.001% of demo requests mention it. That's a hypothesis worth testing, not a verdict. Maybe the podcast needs better distribution. Maybe it's reaching the wrong audience. Maybe it's working exactly as intended and just isn't the thing buyers name when asked. SRA raises the question. It doesn't answer it alone.

What to avoid: treating one quarter of data as statistically meaningful, using SRA by itself to justify a big reallocation, or dropping SRA percentages into a board deck without flagging the memory bias baked into every response. SRA earns weight over time, by consistently confirming or contradicting other signals — not by producing one surprising number in one reporting period.

Where SRA fits in a full-stack B2B growth marketing system

SRA is a diagnostic instrument. On its own, it doesn't grow anything. What makes it valuable is whatever system exists to act on what it reveals.

If SRA shows community and word-of-mouth driving a large chunk of pipeline, the right response is a content and community strategy, not a tweak to a Google Ads bid. If SRA shows LinkedIn as the second-most-cited channel among well-fit buyers, that finding should shape targeting, creative, and budget allocation on the platform itself. The insight only matters if something downstream changes because of it.

The stack that makes this work: CRM as the system of record, with SRA tied directly to the opportunity; platform data driving day-to-day optimization; SRA sitting alongside both as the channel-level reality check. None of the three should live in a silo. Read together, they tell a fuller story than any one of them alone.

Teams that get the most out of SRA are the ones asking a sharper question first: what's actually constraining growth right now? Channel visibility? Weak creative? A landing page that leaks conversions? An attribution gap that's hiding a working channel? SRA answers that last question better than almost any other tool available. It's wasted, though, on a team using it as a standalone metric to report and move on from.

There's a compounding effect worth naming, too. SRA data collected over several quarters builds something no third-party benchmark can offer: a specific, proprietary picture of how a particular type of buyer, in a particular category, actually finds and evaluates vendors. That picture gets sharper as the dataset grows. It's not something a competitor can buy or copy, because it belongs to the pipeline that generated it.

For sales-led B2B companies running paid media on Google and LinkedIn, SRA functions as the missing feedback layer, the piece that connects what the ad platforms report (clicks, impressions, conversions) to the upstream influence those same platforms structurally cannot see. Without it, optimization stays boxed in by whatever the dashboard happens to show, which was never the whole picture to begin with.

Handled well, SRA turns one form field into a running research project. Not a report someone glances at once and forgets. A signal that gets more useful every quarter it keeps running.