Discussion starter: a B2B and SaaS SEO question, posted with an answer from Niraj Raut to open the thread. If you have dealt with this on a site, reply with what you saw, especially where it differs.
- SaaS SEO
With a nine-month enterprise sales cycle, how do you prove SEO’s contribution to pipeline when first-touch and last-touch both mislead?
Disclosure: Niraj Raut, who posted this answer, runs the SEO consultancy linked at the end of it.
1 reply
-
adminAdmin
You will not prove it with an attribution model, and searching for the “right” model is usually where these conversations go wrong. On a nine-month cycle, software attribution cannot see most of the journey. The credible approach is a triangulated case: account-level touch data in the CRM, what buyers say drove them, and a small number of tests or time-series analyses that speak to incrementality. You report SEO’s contribution as a defined range, not a single number.
The distinction that matters most is attribution versus incrementality. Attribution describes where organic search appeared in a buyer’s journey. Incrementality asks how much pipeline would have been lost without it. First-touch and last-touch both answer the first question badly, and neither attempts the second, which is the one a CFO actually cares about.
Why both single-touch models break on long enterprise cycles
The problem is not that first-touch and last-touch are biased in known directions you can correct for. They are biased in directions you cannot measure, because the data they rely on is missing in predictable places.
- The tracking window is shorter than the sales cycle. In GA4, the lookback window for most key events defaults to 90 days, with 30 and 60 days as the only other options. Nine months is roughly 270 days. Anything organic did in the first half of the journey falls outside the window before the opportunity even exists. Browser privacy controls make it worse: Safari caps cookies set by script at seven days, so a returning researcher often looks like a new visitor.
- The researcher is rarely the buyer. Enterprise deals involve a buying group. The analyst who found your comparison page in month two may never fill in a form. The executive who books the demo in month seven arrives through a branded search or a sales email. Contact-level attribution credits the executive’s path and ignores the analyst’s.
- Last touch rewards capture, not creation. Branded search, direct, SDR outreach and events dominate the bottom of the funnel. Organic shows up there too, but mostly as branded queries, which are often a symptom of earlier exposure rather than a cause.
- First touch is the touch most damaged by data loss. “First touch” in a CRM usually means the first touch you managed to record, which on a long cycle is often months after the real first interaction. It can overcredit or undercredit organic, and you cannot tell which.
GA4 also retired its first click, linear, time decay and position-based models in November 2023, leaving data-driven and last click options. Data-driven attribution is still bounded by the same lookback window and the same identity gaps, so it does not solve the long-cycle problem either.
A useful picture: single-touch attribution on a long enterprise deal is like crediting a goal to either the player who made the first pass in the move or the one who scored. Neither tells you whether the midfielder’s pass in the middle was necessary, and on a nine-month move the camera missed half the passes. The literal issue is that attribution models distribute credit across recorded events, while the question leadership is asking is counterfactual.
Separate the two questions:
“Where did organic search show up?” can be answered with better tracking. “How much pipeline would we lose without SEO?” needs experiments, time-series analysis or buyer testimony. Report them as different metrics so nobody reads an influence number as a causal one.
A three-layer evidence model that survives a finance review
Each layer fails in a different way, which is exactly why they work together. Finance can poke holes in any one of them; it is much harder to dismiss three that point the same direction.
Layer Question it answers Main source Main weakness 1. Touch evidence Did organic appear in the journeys of accounts that became pipeline? CRM and marketing automation, rolled up to account level Misses anonymous research and untracked devices; not causal 2. Buyer testimony Do buyers say search mattered? Open-text “how did you hear about us” plus discovery-call notes Recall bias; “Google” is ambiguous 3. Causal evidence What changes when SEO investment changes? Staggered rollouts, time series with a control, template tests Slow; pipeline is too sparse to test directly Layer 1: build touch data at account level, not contact level
- Capture landing page, referrer and channel into hidden form fields on every conversion, and keep the full session history on the contact record, not just “original source” and “latest source”. Most marketing automation platforms can store every known touch; the default fields alone are not enough.
- Roll contact touches up to the account and the opportunity through opportunity contact roles, so the analyst’s research counts towards the deal the executive signs.
- Split organic touches by query type and page type. A non-brand visit to a comparison, integration or pricing page is far stronger evidence than a branded homepage visit. Search Console’s branded queries filter lets you track branded and non-brand demand separately at the query level, and branded demand growth is worth reporting as an outcome in its own right.
- Define windows in advance. For example, “SEO-sourced” means the first recorded touch on the account was non-brand organic within 12 months before opportunity creation. “SEO-influenced” means any contact on the account had a non-brand organic session on a commercial page in the 180 days before opportunity creation.
- If you use tools that match anonymous visits to companies by IP address, treat that data as directional. It fills part of the pre-form gap, but match rates vary, particularly for people working from home connections.
Never add up influenced pipeline across channels. Every channel will “influence” most large deals, and the total will exceed 100% of pipeline. Nothing loses credibility with finance faster.
Layer 2: ask buyers, then compare their answers with the software
Add an open-text “How did you first hear about us?” field on demo and contact forms, and have sales ask the same question in discovery and log the answer. Open text matters because dropdowns anchor responses. Code the answers into categories such as search, AI assistants, peers, analysts, events and podcasts.
Self-reported attribution has obvious weaknesses: people say “Google” when they clicked an ad or searched your brand after a colleague mentioned you. Its value lies in the gap between what buyers say and what the software recorded. If buyers mention search in far more deals than the CRM credits to organic, that is evidence of undercounting. If they rarely mention it despite high influenced numbers, your influence definition is probably too loose.
This layer also catches exposure that tracking misses completely. Search Console’s generative AI reports show impressions only, with no clicks or queries. GA4’s new AI Assistant channel captures some assistant referrals, but traffic without a referrer still lands in Direct. Practitioners also report that clicks from Google’s AI Mode arrive as google / organic with nothing to distinguish them. When a buyer’s research happens partly inside AI answers built from your pages, testimony is one of the few ways to see it.
Layer 3: get some causal evidence, even if it is indirect
Pipeline itself is usually too sparse and too slow to test. Fifteen SEO-related opportunities a quarter will not show a statistically meaningful change within a financial year. So test on leading indicators you have already shown are connected to pipeline.
- Validate the chain first. Using historical cohorts, measure how non-brand organic sessions on commercial pages convert to demo requests, how demo requests convert to qualified opportunities, and the typical lag at each step. If that chain is reasonably stable, a change at the top becomes a defensible forecast of pipeline two to three quarters out.
- Staggered rollouts. When you build a content cluster for one product line or region, hold back a comparable one for a quarter. Compare treated and held-back segments with a time-series method that builds a synthetic control, such as Google’s open-source CausalImpact package. Segments are never identical, but this answers “what would have happened otherwise” better than any attribution model.
- Template tests. If you have enough templated pages (integrations, use cases, locations), split tests on page groups can show whether a change lifts organic clicks. The test proves the SEO lift; the validated chain carries it to pipeline.
Switching SEO off as a holdout, the way paid teams pause campaigns, is rarely realistic because rankings can take months to rebuild. Holding back new investment in one segment is the practical equivalent.
Hypothetical example:
An enterprise software company creates about 60 new opportunities a quarter. Using definitions agreed with finance beforehand, the quarterly report shows 9 opportunities where the first recorded account touch was non-brand organic (SEO-sourced), 26 where someone on the account visited a commercial page from non-brand organic in the 180 days before creation (SEO-influenced), and 13 where the buyer named search or a specific article when asked. The honest summary is a range: organic search sourced 15% of new opportunities and appeared in 43%, with buyer testimony (22%) landing between the two. Add the result of the one staggered rollout run that year and leadership has a case it can challenge on specifics rather than dismiss outright. The numbers here only illustrate the reporting shape.
The reporting set I would put in front of leadership
Metric Cadence What it supports What it cannot claim Non-brand organic clicks to commercial pages Monthly Leading indicator of captured demand Pipeline on its own Demo and contact requests from non-brand organic sessions Monthly Mid-funnel contribution Opportunity quality Target accounts with non-brand organic engagement Monthly Reach into the accounts sales cares about That the engagement caused a deal SEO-sourced pipeline Quarterly Deals where organic was the first recorded touch That the deal would not have happened otherwise SEO-influenced pipeline Quarterly Breadth of organic involvement A total that can be added to other channels Self-reported search share Quarterly Buyer perception, including untracked touches Precise credit Rollout or test lift Per test Incrementality for that segment Generalisation to every SEO activity Watch for this:
Comparing win rates or deal sizes of SEO-touched opportunities with untouched ones is tempting and misleading. Accounts that research heavily touch more channels, including organic, and are often more serious buyers to begin with. A higher win rate among organic-touched deals is selection bias until shown otherwise, so present it as descriptive, never as SEO’s effect.
What to do in the next 90 days
- Agree the definitions of sourced and influenced, and the time windows, with finance and sales operations before pulling any numbers. Pre-agreed definitions stop the report looking like cherry picking.
- Fix capture: hidden fields, full touch history, contact roles on every opportunity, and a non-brand flag on organic sessions based on landing page.
- Launch self-reported attribution on forms and in discovery calls.
- Backfill from the CRM, not GA4. Pull opportunities created in the last 18 to 24 months and reconstruct whatever touch data exists. Standard GA4 properties keep event-level data for 14 months at most, and imperfect history is still better than waiting nine months for the first clean cohort.
- Choose one segment for a staggered rollout and document the control before launch.
If I had to defend SEO’s pipeline contribution next quarter, I would not lead with an attribution percentage. I would lead with a validated chain from non-brand organic demand to qualified opportunities, show sourced and influenced pipeline as a range with published definitions, support it with what buyers say, and point to one segment where holding investment back produced a measurably different result. Each piece has a known weakness, but they fail in different ways, and that is what makes the combined case hard to dismiss.
Need help with this on your own site?
Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.
Add a reply Cancel reply
Replies are open to approved members. Log in or apply to join.