The MQL playbook is dead, and it did not die in a board meeting. There was no postmortem, no funeral, no line item struck from the budget. It died in private, at full output, with every dashboard still updating. The scoring model still scores. The SLA clock still ticks. The Monday report still says the machine made its number. A playbook that fails loudly gets fixed. A playbook that keeps producing numbers while the actual decisions move somewhere else just keeps running, and that is the more dangerous way to go.
We helped build versions of this machine and ran them well for years, so this is not a drive-by. The MQL engine was a real achievement of its era. It rested on one load-bearing assumption, and the assumption broke.
Built for a buyer you could see
The playbook has a birthplace: the marketing automation wave of roughly 2008 to 2015. The platforms of that era, and the discipline that grew up around them, turned the funnel into an assembly line with inspection gates. In fairness, the engineering matched the buyer. Research happened in places you could instrument. Buyers found you through searches that ended on your pages, downloaded your gated PDFs, sat through your webinars, opened your emails. Every step of an evaluation left footprints on property you controlled.
So an entire operating system formed around reading footprints. Lead scoring assigned points to page views and opens because digital body language was a fair proxy for intent. Stage gates defined the moment a lead was warm enough to cross the org chart. SLAs governed the crossing. Nurture kept the not-yet-ready warm until they behaved like buyers. Marketing produced MQLs, sales accepted or disputed them, and a weekly meeting adjudicated the disputes.
The load-bearing assumption was never written down because it never needed to be: the buyer researches where you can see them.
The buyer left the building
Then research moved, in two waves. First into peer rooms: private communities, industry Slacks, group chats where a single "who is actually good at this" thread does more shortlisting than a quarter of nurture. Then into AI assistants, where a buying committee asks a model to name the credible vendors, compare approaches, and flag the risks, and receives a synthesized answer with no click, no cookie, and no footprint on anyone's property.
The consequence is blunt. The form fill still happens, but it happens at the end of the evaluation, not the beginning. The demo request is not a hand raised in curiosity. It is a decision surfacing for paperwork. Your funnel now starts where the buyer's process ends.
The machine cannot see any of this, and worse, it is built to reassure you that nothing changed. Leads still arrive. Scores still fire. Targets get hit, or missed by explainable amounts. Which brings us to the failure modes, all of them quiet.
Scoring theater
The first is scoring theater. The model keeps producing scores, and everyone involved slowly stops believing them without ever saying so. You can tell by how the scores get used. Thresholds drift downward in quarters when the MQL target is at risk. Points get added for engagement that is easy to manufacture. Nobody can name the last deal the model surfaced before sales already knew about it.
There is a subtler problem underneath. When a prospect finally shows up and binge-reads your pricing and security pages, the model reads a buyer entering evaluation. It is usually a buyer finishing one. The score is grading the residue of a decision made in rooms you cannot instrument. High score, late arrival. The model is not wrong about attention. It is wrong about tense.
SLA fights over decided deals
The second is the handoff ritual. Whole meetings still get spent on follow-up windows: minutes to first touch, hours to disposition, lead disputes escalated with the energy of a border conflict. Speed to lead was worth fighting over when the lead was early and the first competent conversation could shape the evaluation.
Now look at what the fight is actually about: a lead whose shortlist formed weeks ago in an AI chat and a peer thread. The rep who calls in four minutes and the rep who calls in four hours get the same outcome, because the outcome was settled before the form. The SLA is precise, enforced, and irrelevant. Sales blames lead quality, marketing blames follow-up speed, and the argument consumes the senior attention that should be spent asking where the evaluation actually happened.
Nurture that arrives after the shortlist
The third is nurture. The day-30 case study email lands on a buyer who asked an assistant to compare the top vendors in week one. Nurture assumed two things: that you controlled the drip of education, and that the buyer learned on your calendar. Both are gone. Education is now on demand, synthesized in seconds, from sources you do not choose and cannot edit. Nurture until sales-ready quietly becomes email until irrelevant. The sequences still send. The opens still count. The shortlist was drawn before message four.
What replaces the machine
Not a new funnel diagram. An operating change, in four moves.
First, committee signals instead of lead scores. The unit of analysis becomes the buying group. Several people from one account appearing inside two weeks is a signal worth a meeting; one person downloading a PDF is not. This is the part the account-based crowd got right, minus the software worship. Route accounts, map committees, and let sales work the group instead of speed-dialing the individual.
Second, answer-layer presence. If evaluation happens inside AI answers, then being present, cited, and accurately described in those answers is the new top of funnel. It is also measurable: which questions your buyers ask, which sources get cited, what share of those citations you hold, and what the answers actually say about you. That is a program with owners and baselines, not a channel you toss budget at.
Third, prospect-reported discovery. Put "how did you hear about us" back on the form as free text, ask it again on the first call, and log the answers where finance can see them. It is unfashionable because it is not automated, which is roughly why it works: the buyer is the only witness to the invisible part of the journey. At one client, a Series D identity verification platform, 51 prospects have self-reported AI discovery in the client's CRM. Prospect-reported, not our causal claim, and the restraint is the point. It is testimony from the only party who was in the room.
Fourth, pipeline accountability instead of stage volume. Marketing signs up for qualified pipeline and revenue contribution, and defends leading indicators it can explain: answer presence, committee engagement, reported discovery. The Monday question changes from "did we make the MQL number" to "which accounts moved, and what did we learn." This is the spine of our demand generation work now, and it is a harder job. Volume metrics were comfortable because they always produced a number. Pipeline metrics produce a verdict.
Retiring it without burning the org
Do not delete the MQL field on a Tuesday and declare a new era. The machine is wired into compensation, dashboards, and identity, and teams will defend the metric that pays them. Run two books side by side for a couple of quarters: keep the old reports alive while the new signals earn trust, then move goals and comp once the pipeline view has survived contact with a board meeting.
Keep the parts that were always just good manners. Fast routing on a demo request is not MQL ideology, it is courtesy, and courtesy still closes. What you retire is the belief system: that scored clicks predict revenue, that the handoff is where deals are won, that a buyer can be nurtured onto your timeline.
Above all, change the Monday question, because it is the only lever that reliably works. Teams optimize for whatever leadership asks about, so ask about accounts, answers, and pipeline, and stop asking about the number the corpse still files.
The playbook died in private. The useful move is to notice in public, before your competitors do. If you want a sober read on where your category's evaluation has actually gone, the free self-audit is a reasonable place to start.