Attribution is a peace treaty. Sign it before the quarter starts.

At the QBR, marketing’s slide and sales’ slide each claim most of the same pipeline. Someone adds the two percentages out loud and gets well past 100%. The next half hour goes to arguing over a field in the CRM, and the CFO leaves trusting neither slide.

The argument runs on money. SDR plans pay on the meetings and opportunities reps source, and some AE plans pay a premium on self-generated deals. Marketing defends its budget with the same sourced pipeline. When pay and budget hang on one field, each team has a rational reason to fight for it.

That setup buys predictable behavior. An SDR logs a dial on an inbound lead an hour after the form fill, and marketing invites the whole open pipeline to a webinar in week 12. Both moves are rational under the plan you wrote, which makes the plan the thing to change.

Add budget pressure and the argument gets louder. In Gartner’s 2026 CMO Spend Survey of 401 marketing leaders, most at companies with more than $1 billion in annual revenue, budgets held effectively flat at 7.8% of company revenue, up from 7.7% in 2025. Fifty-six percent of CMOs said their marketing organization lacked the budget to deliver its 2026 strategy. A CMO in that seat will defend each sourced dollar, and a CRO whose reps are paid on the same dollars will defend them right back.

Most of the journey happens where you cannot see it

The first touch your CRM records usually arrives late. In 6sense’s 2025 Buyer Experience Report, built on nearly 4,000 responses, buyers first engaged sellers 61% of the way through their journey and initiated 79% of those engagements. They chose from their Day One shortlist 95% of the time, and the vendor they contacted first won about 80% of the time.

Buyers would also rather do the work alone. At Gartner’s CSO & Sales Leader Conference in May 2026, Director Analyst Alyssa Cruz said two-thirds of B2B buyers prefer a rep-free experience, though they still value sellers at high-impact moments. Forrester’s State of Business Buying 2026, built on a survey of nearly 18,000 buyers, counts 13 internal stakeholders and nine external participants in the average purchase decision, and 94% of those buyers report using AI along the way. Your form captures whichever one of those 22 people clicked.

Social platforms erase more of the trail. SparkToro’s Rand Fishkin wrote in 2024 that more than half of the major social networks hide some or all of their referral data. In a 2023 SparkToro test, Google Analytics marked 100% of visits from TikTok, Slack, Discord, Mastodon and WhatsApp as direct traffic, with no referral information. A buyer who clicks through from one of those apps arrives with no source attached, and the credit lands on whoever touched the lead next.

Sourced, influenced and self-reported

Sourced answers one question: whose action created the opportunity record? Each opportunity gets one source under written rules, and sourced totals add to 100%. Treat it as bookkeeping: use it to decide who gets paid, and expect it to say little about why the buyer picked you.

Influenced covers more ground: which teams touched the buying group before or during the deal? Count any logged touch on a buying-group contact in the 90 days before the opportunity was created or while it is open. Influence overlaps by design, so the totals run past 100%. No commission rides on it.

Self-reported fills the gap your software leaves: where did the buyer hear about you? Put a required open-text field on demo and contact forms, then ask again on the first call. Code the answers weekly into a fixed list that includes podcasts, peers, communities and AI assistants. Buyers compress a long journey into one memorable answer, so read the trend across quarters and keep it out of anyone’s paycheck.

Split brand search from category search. A buyer who typed your company’s name into Google heard about you somewhere first, so “Google” should prompt a follow-up question on the first call. In an October 2025 post, Refine Labs COO Evan Hughes listed growth in branded search and better self-reported attribution signals among the evidence his team shows for brand programs.

The treaty, clause by clause

Draft it in the last month of the quarter and sign it before day one of the next. Six clauses cover the ground.

1. Definitions. Sourced, influenced and self-reported mean what the appendix says, including the 90-day influence window and the qualifying source actions: inbound form, event lead, SDR-booked meeting, partner registration and AE-created opportunity.
2. Data sources. The CRM opportunity record is the system of record for sourced. Marketing automation and the activity-capture tool feed influenced, and the coded form field feeds self-reported. No spreadsheet edits after the quarter closes.
3. Tie-breakers. When inbound and outbound touches both land in the 14 days before creation, the earlier timestamp wins. Partner-registered deals go to the partner. A lost opportunity reopened within 90 days keeps its original source. Expansion belongs to the account owner unless a marketing program reached a new buying group.
4. Comp implications. SDR, AE and partner commissions pay on sourced only. Marketing bonuses use sourced pipeline weighted by its win rate, so volume that does not convert pays less. Influenced and self-reported data stay out of every paycheck.
5. Review cadence. RevOps publishes the attribution report on the fifth business day of each month. Marketing, sales and finance review it together once a quarter. Definition changes take effect only at the next quarter start, with 30 days’ notice, and apply to opportunities created after that date.
6. Dispute process. A team files a dispute within 30 days of opportunity creation, with timestamps as evidence. RevOps rules within five business days, and appeals go to the CFO, whose ruling is final. A settled record stays settled at the QBR.

The CRO, CMO, CFO and head of RevOps sign it. The CFO’s signature carries the most weight, because finance is the party that can stop paying for the argument.

Taking it to the CFO

Put all three numbers on one page with each rule printed beneath it, so the CFO can see why influenced runs past 100% and sourced does not. For demand programs, add the efficiency measures Hughes uses: win rate by source, sales velocity and cost per opportunity.

Show where self-reported answers and software attribution disagree. A channel that buyers name often and your software credits rarely is the first candidate for a lift test.

Propose that test for next quarter as a holdout or geographic split, the lift-based measurement SparkToro’s Rand Fishkin argued for over click attribution in 2024. Count only the lift, priced at contribution margin, and bring that number before finance builds its own. A Refine Labs example shows how far the math moves: 12x ROI on a naive LTV:CAC basis, 9x on contribution margin and 2.7x on incremental contribution margin. A CFO who watches marketing volunteer a test it could lose reads the rest of the deck differently.

The treaty costs one meeting and a concession from each side. Book that meeting before the quarter opens, or book the same argument into the next QBR, with better slides and less trust.