Illustrative composite: the quarterly business review is going well until the CRO asks a simple question. Marketing reports 1,400 MQLs for the quarter, up 20%. Sales says it worked fewer than 500 of them, and that most of those were "not real." The pipeline report shows 180 SQLs, but a third of the opportunities behind them were created after the first call had already happened, and a dozen accounts listed as MQLs are current customers. Everyone in the room is quoting the same CRM. Nobody agrees on what any of the stages mean, and the definitions slide from last year's offsite says something different again.
This is not a definitions problem, though it is always argued as one. It is an automation problem. The stages were set by rules that fire once, on one contact, and never look again. The buyer kept moving. The field did not.
The research has been pointing at this for a while. In its 2023 "Saying Goodbye to MQLs" series, Forrester noted that fewer than 1% of leads make it from the top of the funnel to the bottom, and its Buyers' Journey Survey that year found 93% of B2B buyers purchasing as part of a group of two or more people, and 71% in a group of four or more. Forrester's State of Business Buying 2024 put the average buying decision at 13 people, with 89% of purchases involving two or more departments and 86% stalling at some point in the process. A lifecycle model built on one contact crossing one score threshold is measuring a fraction of that reality.
Buyers are also arriving later and better informed. 6sense's 2024 Buyer Experience Report found buyers about 70% of the way through their process before engaging sellers, more than 80% initiating first contact themselves, and 81% already holding a preferred vendor at that point. Gartner's 2025 survey of 632 B2B buyers found 61% preferring an overall rep-free buying experience and 73% actively avoiding suppliers that send irrelevant outreach. Stages that wait for a form, then fire a sequence, are both late and noisy.
And the data underneath is shaky. In Validity's State of CRM Data Management in 2025 report, a survey of 602 CRM users and administrators, 76% said less than half of their CRM data was accurate and complete, and 37% reported that their staff fabricate data to appease decision-makers. Lifecycle stages, which are easy to set by hand and visible in every leadership report, are exactly where that pressure lands.
Diagnosis: why lifecycle stages drift out of sync with reality
At $3M to $30M ARR, the lifecycle model was usually designed once, by the first marketing operations hire, and then extended by everyone who came after. Five patterns explain almost all of the drift.
Point-in-time scores that never expire
The classic MQL rule says: when the lead score crosses a threshold, set the stage to MQL. The score rises when someone downloads a guide or attends a webinar. It may decay later, but the stage does not move with it. A contact who crossed the line eighteen months ago is still an MQL today, counted in the funnel, sitting in a rep's queue and dragging down every conversion rate it touches. The rule stored a moment as a fact.
Forward-only fields and manual overrides
Many platforms are built to move stages in one direction. HubSpot's own documentation, for example, notes that its default lifecycle stage property can only be moved forward by HubSpot tools such as imports, form submissions and the API, and must be cleared manually or by a workflow before an earlier value can be set. That design protects against accidental regressions, but it also means nothing moves a lead back when it goes cold unless someone builds that path deliberately. Meanwhile reps learn that setting a stage by hand clears a queue or makes a report look right. Once manual edits and automated rules both write the same field, nobody can say which one is telling the truth.
Contact-level stages in a buying-group world
Most lifecycle models live on the contact record. But the buyer is a group of people spread across departments, often with one person researching anonymously and another filling in the demo form. The result is an account with six contacts at six different stages, none of which describes where the account actually is. Sales sees an MQL from an intern while the economic buyer has been on the pricing page all week as an unknown visitor.
Several workflows writing one field
Over time, the stage field acquires many writers: the scoring workflow, a form workflow, a list import, the sales engagement tool, the CRM integration and a cleanup job someone built after the last board meeting. Together they produce contradictions and a stage history that reads like an argument. When the definition changes, at least one writer is never updated.
Stages that track marketing activity, not buying behavior
Points for opening an email or downloading content measure how well marketing distributes content, not whether anyone is buying. The signals that actually precede a purchase (several people from one account researching, pricing and integration pages, product usage, a reply to a rep, a meeting held) are often not wired into the stage logic at all, because they live in other systems.
The framework: the Evidence-Gated Lifecycle
The Evidence-Gated Lifecycle replaces "set the stage when a rule fires" with a simple contract: a record is in a stage only while the evidence for that stage is present and recent. Four rules make it work.
Every stage has an evidence contract. For each stage, write down the verifiable events that put a record there, in business language a sales leader would accept. An illustrative contract, to adapt rather than copy: an account becomes Engaged when two or more people from it interact within 30 days; it becomes Marketing Qualified when that engagement includes a high-intent action (a demo or pricing request, a trial start, a reply to outreach) and the account matches your ideal customer profile; it becomes Sales Accepted when a rep accepts it within the agreed window; it becomes Sales Qualified only when a discovery meeting has actually been held and an opportunity exists with its required fields completed. Scores can rank records within a stage. They never move a record between stages on their own.
Stages can expire and regress. Every stage has a time limit and an exit path. If the evidence goes stale (no qualifying activity in 60 or 90 days, as a suggested starting point, not a benchmark), the record recycles to an earlier stage with a reason code. If sales rejects a lead, the rejection and its reason are recorded, and the record goes back to nurture rather than sitting in a queue forever.
One writer, one log. Exactly one automation is allowed to change the stage. Every other system emits events that it reads. Each change writes a stage-change record: the old stage, the new stage, the time, the evidence that triggered it and the rule version. The lifecycle field becomes a view of that history rather than a free-text truth anyone can edit.
The account, not only the contact, carries the stage. Contacts keep their individual status, but the stage that sales acts on and leadership reports on is the account's or buying group's, computed from everyone involved.
Implementation: six steps to lifecycle stages you can trust
You do not need a new marketing platform. You need the stage history you already have, one set of contracts and the discipline to test first.
Audit stage drift from your own history
Export a year of stage changes and compare each record's current stage with the evidence behind it. How many MQLs have had no activity in 90 days? How many SQLs never had a meeting? How many current customers sit in a lead stage? The diagnose-before-you-build playbook covers how to run this read-only. Check: you can state, with numbers, how far each stage has drifted from reality.
Write evidence contracts with sales and marketing together
Agree the entry evidence, the expiry window, the regression path and the owner for each stage, in one page both leaders sign. Keep the list of stages short; every extra stage is another place for drift to hide. Check: a rep and a marketer, reading the contract separately, would place the same ten sample records in the same stages.
Consolidate to one writer
Find every workflow, integration and import that writes the stage field and turn all but one off. Restrict manual edits to a small group, and require a reason when they happen. Route the signals those old writers used (forms, product usage, meetings, replies) into the one stage automation as events. Check: over two weeks, every stage change in the system traces back to the single writer.
Add the stage-change log, expiry and regression
Write a record for every stage change with its evidence and rule version, then build the expiry and recycle paths. Start the expiry windows from your own sales-cycle data and treat them as a suggested starting point, not a benchmark. Check: for any record, you can see why it is in its current stage and when that will next be re-evaluated.
Backtest on deals you already won and lost
Run the new rules against last year's records and compare where they would have placed each account with what actually happened. We hold every system to the same bar: tested on around 20 of the client's own past cases, and 85 percent correct or it does not ship. Check: the new stages separate accounts that became pipeline from those that did not more clearly than the old ones did, and every miss has a written reason.
Run in shadow, then cut over
Compute the new stages alongside the old field for a few weeks without acting on them, and review the differences with sales and marketing leaders. When the shadow stages hold up, switch routing and reporting to them and restate the funnel by cohort so history stays comparable. Check: leadership signs off on the restated funnel before the old field is retired.
Workflow: the evidence-gated stage loop
A suggested operating loop. Adapt the cadence; keep the owners.
What happens: form fills, meetings booked and held, product usage, replies, pricing and integration page visits, and rep dispositions arrive as events, each tied to a contact and, through matching, to an account.
System role: make every buying behavior visible to the stage logic, including behavior that lives outside the marketing platform.
Owner: RevOps owns event capture and account matching.
What happens: on each new event, and on a daily schedule, the stage automation checks every open record against its stage's evidence contract and expiry window.
System role: decide advance, hold, expire or regress, using the same written rules every time, so two similar accounts never end up in different stages by accident.
Owner: RevOps maintains the rules; marketing and sales leaders approve any change to a contract.
What happens: the single writer updates the stage and logs the change with its evidence, time and rule version.
System role: keep a history anyone can audit, so funnel reports, conversion rates and attribution all read from the same record of what actually happened.
Owner: RevOps, with read access for marketing, sales and finance.
What happens: a stage change triggers its action: routing and a response clock for a new MQL, acceptance or rejection with a reason for a SAL, an opportunity for an SQL, a nurture track for a recycled record.
System role: make every stage mean the same thing to the person who has to act on it, and measure whether they did.
Owner: SDR and sales managers for acceptance and follow-up; marketing for recycled records.
The board narrative
Three statements make lifecycle integrity legible to a board.
Every funnel stage is now defined by verifiable buyer behavior, set by one automation, and re-checked continuously. Stages expire when the evidence goes stale, and every change is logged with the reason it happened.
We were reporting volume that sales did not believe and conversion rates built on records that had gone cold. Marketing and sales now argue about strategy, not about whose numbers are real, and pipeline forecasts start from stages we can prove.
We report stage-to-stage conversion by monthly cohort, time in stage, the share of MQLs accepted by sales, rejection reasons and the number of manual stage edits. Acceptance should rise, manual edits should fall toward zero, and the funnel should restate cleanly when definitions change.
Illustrative example, with made-up round numbers: a $15M ARR company reporting 1,000 MQLs a quarter, of which sales accepts 300, might find after rebuilding that only 450 records meet the new evidence contract, and that sales accepts 350 of them. The headline number shrinks; the number the business can act on grows. Your figures will differ.
Cross-domain: what honest stages unlock
Lifecycle stages are the switchboard of the revenue engine: almost every other system reads them. A stage you can trust is what lets Speed-to-Lead start the response clock the moment real buying evidence appears, instead of chasing cold contacts that crossed a score threshold last year. Sales acceptance and rejection with reasons is a proposed handoff design that the Handoff Orchestrator can track between marketing, SDR and AE. It escalates stalls without changing account owners on its own.
Further down the funnel, the Pipeline Hygiene Sentinel can flag stale or incomplete opportunities without editing deals. The stage-specific checks above are proposed configuration, not universal shipped behavior, and the Forecast Assistant works from a pipeline that entered on evidence rather than on hope. See all systems, or the Sales Operations domain.
For who should own this kind of rebuild as the team grows, see the GTM engineer vs. RevOps manager vs. growth engineer decision tree. Our approach is forward-deployed engineering: build inside your existing CRM, test against your own past deals, and switch each change on only when it proves itself.
Sources: Forrester, "Saying Goodbye to MQLs" blog series (August to October 2023), citing Forrester's Buyers' Journey Survey, 2023. Forrester, The State of Business Buying, 2024 (press release, December 2024). 6sense, 2024 Buyer Experience Report (October 2024). Gartner, Sales survey of 632 B2B buyers (fielded August to September 2024; published June 2025). Validity, The State of CRM Data Management in 2025 (July 2025; 602 CRM users and administrators). HubSpot Knowledge Base, "Use lifecycle stages" (updated July 2026). The Evidence-Gated Lifecycle, the sample evidence contracts, expiry windows and the funnel example are suggested starting points and illustrative figures, not benchmarks.
