The PULL framework — Project, Unavoidable, Looking, Lacking.

TL;DR

• PULL is a four-word discovery rubric: Project (a real to-do on their list), Unavoidable/Urgent (it can’t wait), Looking (they’re evaluating options), Lacking (the options are insufficient for a critical reason).
• Demand math: P+U produces real demand. L+L produces a supply gap they can’t close on their own. High on all four = weird NOT to buy. High on two = a deal that will close once a budget cycle resolves. High on zero = a meeting that should never have happened.
• Snyder’s four-question discovery sequence: (1) Why did you take the call? (2) What near-term result are you trying to produce? (3) How does what you’re looking at compare to what others are doing? (4) Demo last, only if PULL is high.
• The retrospective drill: take 40 recent first-touch calls. Separate the 2 who had real PULL from the 38 who didn’t. The lookback is where the framework earns its keep — not the live coaching.
• PULL is methodology, not infrastructure. It tells you what to listen for. It does not score the calls for you, refresh your ICP from drift, generate the next campaign, or fire the next outbound touch. The doc is the input. The loop is the moat.
• Where the framework stops short for an operator: no scoring schema, no integration with CRM/dialer/meeting notes, no closed feedback loop from call score → ICP → campaign → reply → re-score. PULL gives you the rubric. The campaign-running layer is yours to build.
• How we apply it: every Granola/Gong/Zoom transcript flows through a scoring agent that produces a PULL grade with evidence quotes. Scores roll up into a live ICP drift view. Campaigns are written from the PULL-evidenced language of buyers who scored high. The loop closes in days, not quarters.

Why read PULL

B2B operators acquire discovery frameworks the way they acquire dietary plans — fast, with enthusiasm, and with very little follow-through. The ones that survive in production share a property: the rubric is small enough to apply in real time, and the failure modes are concrete enough that operators can self-correct without an external coach. PULL is one of the small set that survives the transit from blog post to live call.

The mechanism: Snyder reduces the messy question “is this a real opportunity?” into four binary-ish judgments that the seller can score during a call without distraction. Most early-stage operators run discovery on instinct, which produces uneven calibration across reps and across weeks. PULL produces calibration that is auditable in retrospect — you can read a transcript six weeks later and reproduce the score, which is the property that makes a framework operationally load-bearing rather than rhetorical.

One upstream point that matters for the rest of this chapter. PULL is a methodology, not a system. Snyder ships a Substack and a book; the productization is yours. The operators who get the most out of PULL are the ones who build the scoring infrastructure around it — the ones who treat the framework as the input to a closed loop, not as the loop itself. The teams who recite PULL without building the loop reply at roughly the same rate as the teams who haven’t read it.

The four words

1. Project. There is a real project on the buyer’s to-do list. Not a casual interest, not a generic line item — a project with a name, an owner, and a deadline. The presence of a project is what separates a real buying conversation from a polite one.
2. Unavoidable / Urgent. The project is unavoidable right now. Something has changed — a hire, a launch, a regulatory date, a customer escalation, a board review — that prevents the buyer from punting it another quarter. Urgency is the variable that determines whether the deal closes this month or never.
3. Looking. The buyer is actively evaluating options. They’ve looked at competitors, read a comparison post, talked to a peer, reviewed an internal build. The absence of looking is a signal the project hasn’t materialized; the presence of looking is a signal the timeline is real.
4. Lacking. The options the buyer has evaluated are lacking for a specific, articulable reason. “Too expensive” is a weak Lacking. “Doesn’t handle the multi-tenant case our European team needs” is a strong Lacking. The specificity of Lacking is the variable that determines whether your product is structurally the answer or only situationally the answer.

The math is multiplicative, not additive. A buyer with Project + Unavoidable but no Looking has demand and no buying motion — they will solve it internally or wait. A buyer with Looking + Lacking but no Project has window-shopping behavior and will spend six months in evaluation without ever sending a contract. Only the buyers who score on all four words are buyers in the operational sense — the rest are conversations.

The asymmetric phrase

Snyder’s framework spreads because of a single sentence: when PULL is high, the buyer would be weird NOT to buy. The framing inverts the seller’s default question. The seller asks “will this person buy?” The PULL frame asks “is this person already in a position where buying is the obvious move and not buying requires explanation?”

The operational consequence is that the seller’s job changes from persuasion to disqualification. The high-PULL buyer does not need to be convinced — they need to be confirmed and removed from the funnel of buyers who are not ready. The seller who internalizes this stops pitching low-PULL buyers, which is the single biggest unlock for a founder running their own discovery. Pitch effort migrates from low-yield calls to high-yield calls within a quarter, and the same calendar produces 2 to 3x the pipeline.

The four discovery questions

Snyder’s prescribed call structure is short enough to memorize and rigid enough to score. The sequence:

1. “Why did you take this call?” Spend the first three minutes on this question alone. The answer surfaces whether there is a Project at all. The most useful failure mode of this question is the buyer who answers in marketing-speak (“we’re always interested in solutions like yours”) — that is a high-confidence signal of zero PULL and the call should end inside ten minutes.
2. “What near-term result are you trying to produce?” Surfaces Urgency. A buyer with a result tied to a specific deadline scores Urgent. A buyer who can’t name the result, or whose deadline is “eventually”, does not.
3. “How does what you’re looking at compare to what others are doing?” Surfaces Looking and Lacking simultaneously. The buyer who can name three competitors and articulate where each falls short is high on both Ls. The buyer who can’t name a single competitor is in window-shopping mode regardless of how positively they speak about your product.
4. Demo last, only if PULL is high. Demoing into a low-PULL conversation is the single most common discovery failure. The demo converts the conversation from disqualification into pitch, the buyer recognizes the pivot, and the rest of the call is theater. The seller leaves the call optimistic; the buyer leaves the call having politely ended an interview.

The fifth move, often missed, is the comparative reference. After the four questions, Snyder prescribes telling the buyer what other people in similar PULL situations have done. The reference functions as a social proof anchor and as a calibration check — if the buyer recognizes the situation, PULL is confirmed; if they don’t, you’ve scored too high on the call. Operators who skip the comparative reference report calls as “great” that produce no follow-through, because their score was not externally validated.

The retrospective drill

Live PULL scoring is useful. Retrospective PULL scoring is where the framework earns its keep. The drill:

1. Pull the transcripts from your last 40 first-touch calls.
2. Score each call against PULL. Most operators discover the modal score is 0 or 1 — meaning roughly 35 of the 40 calls had no real PULL.
3. Identify the 2 to 5 calls that scored high. Read those transcripts in full.
4. Compare against the 35 that scored low. Look for the structural patterns in how the high-PULL buyers described their Project, their Urgency, the alternatives they’d evaluated, and what those alternatives lacked.
5. The patterns become your refined ICP and your refined messaging. The vocabulary your highest-PULL buyers used to describe their problem is the vocabulary your outbound should use.

The unintuitive output of this drill is that the highest-PULL buyers usually do not match the operator’s pre-drill ICP. The drill repeatedly surfaces a segment the operator had been deprioritizing — a specific industry, a specific firmographic cut, a specific stage — that turns out to be where the real demand lives. Operators who run the drill once and revise their ICP outperform operators who run it once and don’t by roughly 2x on subsequent quarter pipeline.

Where PULL stops short

PULL is rigorous on the rubric and silent on the infrastructure. Five gaps an operator hits the moment they try to run it in production:

• No scoring schema. PULL describes the four words but does not prescribe how to encode them. A 0-3 scale? A binary present/absent? A confidence interval? Operators end up inventing their own scoring conventions, which is fine for one rep and corrosive across a team. Without a fixed schema, two reps will score the same call differently, and the retrospective drill produces noise.
• No integration with the meeting layer.The framework assumes the operator has transcripts to score. In practice, the transcript is the bottleneck — Granola, Gong, Fathom, Zoom captions, raw notes — each ships in a different shape, and the scoring lives in someone’s head until a quarter passes and the calls are forgotten. Without an automated transcript-to-PULL pipeline, the retrospective drill happens twice and then stops.
• No closed loop to ICP.The drill produces refined ICP language. There is no mechanism prescribed for keeping that refinement current. The ICP doc drifts in a week, and by quarter-end the team is operating against a stale view of where demand lives. Snyder’s methodology assumes an analyst will rerun the drill; the analyst doesn’t exist on a six-person team.
• No outbound prescription.A high-PULL retrospective produces a set of patterns. PULL does not tell you what to do with those patterns. The leap from “we found three high-PULL buyers in healthcare RCM” to “here is a 5-touch outbound sequence to ten more” is the entire campaign-running layer. The framework leaves it to the operator.
• No reply-data backflow.Even with a campaign running, PULL provides no mechanism for letting replies feed the score. A reply that says “wrong timing, ask me in Q3” is a high-Project, low-Urgent signal that should re-rank the prospect. Without a pipeline that ingests the reply and updates the score, the rep is re-scoring by hand at the volume that breaks the discipline.

These are not criticisms of the framework. PULL is intentionally scoped at the methodology layer — that is the layer where Snyder’s edge sits. The gaps are operational, and the operator who closes them is the operator who turns PULL from a coaching tool into a pipeline engine.

How we run PULL at production scale

The architecture we use across customer engagements. The structure is borrowed from Snyder’s methodology; the operational layer is what we ship.

1. Transcript ingest.Every meeting recorded across Granola, Gong, Fathom, Zoom is auto-ingested into the customer’s context layer within minutes of the call ending. Format-agnostic — we normalize at ingest so the scoring agent sees one shape.
2. PULL scoring agent. Each transcript runs through a scorer that produces a 0-3 score on each of P, U, L, L, with two to four evidence quotes per dimension and a one-sentence rationale. The scorer is calibrated against a customer-specific gold set of 20 manually-scored calls — calibration drift is checked weekly.
3. ICP drift view. Scores roll up into a live view that shows the firmographic, vertical, and stage patterns of high-PULL buyers across the last 30/60/90 days. When the patterns shift — a new vertical starts scoring high, an old one collapses — the operator gets a notification, not a quarterly report.
4. Campaign generation. The vocabulary of the highest-PULL buyers in the last 60 days feeds the outbound copy. Subject lines, opening lines, and value-prop framing are drafted against the language those buyers used to describe their Project, their Urgency, and the Lacking of their alternatives. The copy reads like a paraphrase of a high-PULL call because, structurally, that is what it is.
5. Reply backflow.Every reply is classified against the PULL dimensions. A “wrong timing” reply lifts the Project score and drops the Urgency score on that prospect; a “not the right person” reply triggers a re-routing to the right person and a re-score; a positive reply that books a call increments the conviction on the source PULL pattern.

The aggregate property of this architecture: the PULL rubric runs as a live system, not as a quarterly exercise. The ICP refreshes from real signal weekly. The campaigns refresh from the same signal. The conviction-pattern map updates with every call and every reply, and the team operates against a current view of where demand lives instead of a six-month-old ICP doc.

Operator failures observed when adopting PULL

• Scoring the call in real time without a retrospective.Live scoring is the most common adoption pattern and the least useful. The retrospective drill against 40 transcripts is where the framework changes the operator’s ICP. Operators who skip the retrospective treat PULL as a coaching note and never get the upstream benefit.
• Treating PULL as a sequential checklist. Some operators read the four words as sequential gates — Project first, then Unavoidable, then Looking, then Lacking — and disqualify a call that misses any one in order. The math is multiplicative across the four, not a sequence. A call with strong Looking + Lacking and weak-but-present Project + Unavoidable is still worth a follow-up, often a strong one.
• Inventing scores rather than scoring against the transcript. Operators who score from memory rather than from a transcript drift toward optimism. The score should be derivable from quoted lines in the call. If the operator cannot point to the evidence sentence, the score is fabricated, and the retrospective drill produces no signal because it is calibrated against opinion rather than data.
• Confusing Looking with Looking at us. A buyer who is enthusiastic about your product but has not evaluated alternatives scores low on Looking. The confusion produces inflated scores on calls where the buyer is socially engaged but operationally inert. The cost is roughly 2 to 4 weeks of pipeline time spent on a buyer who was always going to disappear in procurement.
• Demoing into a 0/4 PULL. The framework is unambiguous on this and operators violate it routinely. A demo into a call that scored zero across P, U, L, L is theater. The seller leaves optimistic, the buyer leaves polite, no follow-up materializes, and the slot is spent. The discipline of refusing to demo into a low-PULL call is the highest-leverage habit a founder can build, and it is the one operators resist most.
• Adopting PULL without changing the campaign. The framework reshapes the upstream campaign by reshaping the ICP and the language. Operators who adopt PULL at the discovery layer and leave their outbound copy unchanged see calibration improvement on the calls that come in and no change in the volume of high-PULL calls. The framework is most valuable when applied across the funnel, not only at the call layer.

PULL-adoption checklist

• The four words are memorized and the operator can describe each in a sentence without reference to the source material
• A scoring schema is fixed — a 0-3 scale across each dimension, two to four evidence quotes per score, a rationale sentence — and applied consistently across all reps
• The last 40 first-touch transcripts have been scored and the modal score is known (most teams discover it’s 0 or 1)
• The 2 to 5 highest-PULL transcripts have been read in full and the linguistic patterns extracted
• The ICP doc has been revised based on the retrospective — at least one segment has been added, dropped, or reweighted
• The outbound copy has been revised to use the language of the highest-PULL buyers, not the operator’s pre-drill phrasing
• The four discovery questions are run in order, with demo deferred until after the comparative reference
• No demos into 0/4 PULL — the discipline holds across at least two consecutive weeks before the operator stops backsliding
• Reply data flows back into the score — “wrong timing” replies update the Urgency score, not just the campaign list
• The retrospective drill runs at a fixed cadence — monthly at minimum, weekly at scale — and the ICP doc is updated each cycle

Where this fits

PULL is one chapter in our framework analysis series, and it is the methodological foundation that the others build on. Richard Makara’s GTM Context OS is the closest operationalization of PULL — an open-source repo that scaffolds the directory structure and slash commands for a PULL-based context system. Henry Hund’s Protocol sits one layer up, productizing the messaging-source-of-truth that a PULL-refined ICP should anchor against.

On the discovery layer specifically, our 30-minute discovery call frame is the live implementation of PULL’s four-question sequence, and our closed-won deconstruction guideis the retrospective drill applied to deals that have already closed. The two together — PULL on inbound first-touches, closed-won deconstruction on existing customers — are the input pair that produces an ICP that doesn’t drift.

Related chapters

• Richard Makara’s GTM Context OS — the operational scaffold built explicitly on PULL.
• The 30-minute discovery call frame — PULL’s four questions in a single call structure.
• Closed-won deconstruction — the retrospective drill applied to existing customers.
• ICP hypothesis testing — falsifiable ICP refinement, the campaign-layer compliment to the call-layer drill.