Where is my B2B deal actually stuck?

A deal is stuck at the first gate where a stakeholder has not said yes in their own words: the frontier gate. Reading every call and mail in the buyer’s sequence and laying it out as stakeholder × gate (Priority, Problem, Thesis, Mechanism, Onboarding and resources, Features, Commercials) shows who sits on the committee, who has never been heard from, where it stopped, and the one move that reopens it. A summary of the thread cannot do this; it retells the conversation without the buyer’s sequence or the boundary of the record.

Why does a summary of the thread not tell you where the deal is stuck?

A summary of the thread is written by the seller, for the seller. It compresses a sequence of calls and mails into one story, and the story has a narrator: the person who wants the deal to close. Three things fall out on the way. The order goes first: a summary tells you what was said, rarely when, and never what was said after what. Attribution goes second: the buyer’s words and the seller’s paraphrase of them end up in the same sentence. The boundary goes last: a summary reads as complete even where the record stops, so a question that was never asked looks like a question that was answered.

The CRM does not close the gap either. It holds stages and notes, which is a log of the seller’s activity. It structurally cannot show three things: where the deal is on the buyer’s side, why it is stuck, and the next step. Founders know this. The line that comes up on discovery calls is some version of: we can only create this clarity by going through all our deals ourselves. The loop never closes, because going through every deal by hand is the work a founder-led company has the least time for, and the deals that need the read are the ones that went quiet.

Not a chat window. Every call and mail, read in your sequence. That is the whole method. The rest of this document is what it produces.

What are the seven gates a buyer passes, in the buyer’s words?

A buyer does not move through the seller’s pipeline. A buyer answers a sequence of questions, and each question is a gate. Seven of them, G0 to G6, in the order the buyer answers them:

• G0 Priority. Is this change my priority right now?

• G1 Problem. Have you captured my problem right?

• G2 Thesis. Do I believe this approach solves it?

• G3 Mechanism. Do I see how it works on my side?

• G4 Onboarding and resources. Can my people and my data carry it?

• G5 Features. Does it do the specific things I need?

• G6 Commercials. Is this price fair for what I get?

Two properties of the sequence matter for a stuck deal. The first: a gate is passed by a person, not by a company. The champion can be through G3 while the signer has never heard the mechanism explained. The second: a gate is passed in the buyer’s words, not in the seller’s presentation. A demo that shows how it works is not G3. The buyer saying “so on our side it would run like this” is G3.

G1 is the gate sellers routinely leave open. Buyers rarely contest that the seller understands their problem, and sellers rarely ask them to confirm it, so the loop stays open and everything downstream is built on an unconfirmed reading of the problem. Closing it costs one question: did we capture your problem the way you would describe it?

One rule underneath the whole sequence: an objection not taken out at its own gate comes back at the end as a concession. The lock-in worry never addressed at G4 returns at G6 as a discount request. The mechanism question never answered at G3 returns as “let us think about it”.

What is a frontier gate, and why does the next move aim there?

The frontier gate is the first gate where a stakeholder has not said yes. That is where the deal is stuck and where the next move aims.

Everything before the frontier is on record. Everything after it is speculation, however confident the seller feels about it. The next move does not answer the last message the buyer sent, because the last message is usually a symptom. It goes to the frontier and closes that gate in the buyer’s words.

01 · YOUR RECORD, IN SEQUENCE

T1 · T2 · T3 · … · Tn

• transcripts · mails, messages, proposals · your six intake lines

• record ends at Tn: nothing after it is guessed

02 · STAKEHOLDER × GATE

• S1, the champion: G0 on record, in the buyer’s words · G1 on record, in the buyer’s words · G2 on record, in the buyer’s words · G3 assumed · G4 never touched · G5 never touched · G6 never touched

• S2, the signer: G0 on record, in the buyer’s words · G1 on record, in the buyer’s words · G2 on record, in the buyer’s words · G3 never touched · G4 never touched · G5 never touched · G6 adverse on record

• S3, the seat never heard from: G0 never touched · G1 never touched · G2 never touched · G3 never touched · G4 never touched · G5 never touched · G6 never touched

• Frontier gate: G3

timeline first, then the committee · each mark on record carries the buyer’s own words, with the date

03 · WHAT COMES BACK

• The frontier gate, named

• The one move that reopens the deal

• One next step, dated, in the buyer’s control

A schematic. Yours is built on your record only.

G0 Priority · G1 Problem · G2 Thesis · G3 Mechanism · G4 Onboarding and resources · G5 Features · G6 Commercials

Read the schematic row by row. S1, the champion, is on record through G2 and marked assumed at G3: the seller has explained the mechanism and heard nothing back that confirms it landed. S2, the signer, is on record through G2, has never been touched at G3 to G5, and is adverse at G6: a remark about price or lock-in sits in the record with a date. S3 has never been heard from at all.

The frontier is G3. The adverse mark at G6 looks like the problem and is not. It is what a skipped G3 to G5 looks like when a signer is asked for a price opinion before seeing how the thing would run on their side. The move is not a discount. The move is the mechanism, shown to the signer in the signer’s terms and confirmed in the signer’s words, and the G6 remark is revisited only after that.

Who is the seat you have never heard from?

A deal can be stuck at a seat rather than a gate. The pattern that ends deals at signature is familiar to anyone who has sold into a committee: everyone in the room said yes, the contract went out, and a person nobody had spoken to appeared with a veto. Procurement. The works council. The engineer who owns the data. The deal did not die at signature. It was never alive at that seat.

Practitioners describe the same thing from different angles. Map the entire buying committee, champions, influencers, blockers and economic buyers included [5]. In technical categories the buyer pool is small, several stakeholders evaluate every purchase, and a committee approves the budget [4]. The executive who signs and the people who have to adopt need different messages, and a message that lands with one can put off the other [3].

In the read, the empty row is the finding. An untouched cell is a finding: a person the deal has never heard from, or a question never asked. A row of untouched cells is a person, and the next move often starts with getting that person into the room. Among the named cases, HASE+CO is the long-cycle analog: an established manufacturer selling to real buying committees, where deal states and gate reads were installed per stakeholder so that every deal carried one state, one owner per read and one next move.

How do you separate what is on record from what you are assuming?

Four marks, and they are the whole discipline. On record, in the buyer’s words: the buyer said it, and the date is attached. Assumed: the seller believes it and the record does not contain it. Never touched: nobody asked. Adverse on record: the buyer said no, or said something that reads as no, and the date is attached.

The legend is where deals turn out to be stuck. A gate the seller passed in a presentation and the buyer never confirmed is an assumed cell, and assumed cells are where optimistic forecasts come from. The rule for the read is stated as a boundary: nothing after the record is guessed. No number is invented. Names not in the record are not guessed. If the transcripts stop on a date, the read stops there too, and says so.

Honesty boundaries are a deliverable, not a disclaimer. A read that says the record is blind on the signer after a given date gives the founder a next action. A summary that fills the gap gives them a feeling. One pattern from live sessions: an error in a read, corrected on the spot in front of the person whose deal it was, built more trust than a clean walkthrough would have. The correction showed that the marks meant something.

What is the one move, and what is the one closed question to ask?

The read ends in one move and one question. The move is aimed at the frontier gate and at the person holding it. The question is closed: one commitment, asked so that the answer is yes or no, with the next step that follows the yes dated and in the buyer’s control. Never an “either way”. Asking whether to talk again after the board meeting or whenever suits is not a closed question; it is a deferral with two exits. Asking whether the CTO can join next week to see the integration path on their own data is one.

The next-touch briefing is written per attendee: what to show them, then what to ask. The showing comes first because a gate is passed in the buyer’s words and the buyer needs something to say yes to. The asking is one thing, not a list.

This is the same discipline practitioners describe when they talk about a repeatable motion: a written map of the buying group, conversion tracked stage by stage, and a loop that folds every lost deal back into the playbook [2]. A founder who has to show an investor the full journey, how long it takes to educate and convince the customer and which approvals sit in the path, is drawing the same matrix from the other side [1]. On the Cosa side it is the deal motion on the methodology page: the read first, the next touch built on it, in that order. What the read becomes once a founder stops doing it by hand is the subject of When should a technical founder hire a sales leader?.

Why does the single likelihood number founders want hide the answer?

Founders ask for one number: how likely is this to close. A single percentage would hide what they need. Probability is not a property of the deal. It is lost at a specific gate by a specific person, and averaging over the committee produces a number that is true of nobody. The matrix shows which person at which gate holds the probability. It also separates two things a single number merges: technical alignment, which is yours, and the path from conversation to contract, which is where the deal is actually stuck and what the read works on.

The pilot case is a common instance: a buyer who said yes to a proof and never started it. It has its own document, Why do B2B pilots stall after the buyer says yes?

What to send if you want the read done for you

If you want the read done for you, the free session is Unstuck Your Deal: one stuck deal, read end to end, returned in writing. Nothing is connected for it. The read runs on the document you send. The package is one Google Doc:

• Every call on the deal, as a transcript, first to last.

• Every mail, message and proposal.

• Everything in order, with timestamps. Anonymized where you need to.

At the top of the same document, six lines:

1. The deal. Which company, which product or pilot, and who on their side you have spoken to.

2. Where it stalled. The last thing they said or wrote, and when.

3. What they need to see. What the buyer says would get them to commit, in their words if you have them: a data challenge, a parallel run, a live test, a reference, a number.

4. Readiness. By which date your product or hardware can run a paid pilot or design partnership alongside your own tests.

5. The round. How many signed pilots or design partners your next round needs, and by when.

6. The ticket. What a pilot costs the partner today, and what it should cost once the product is stable.

Skip what you cannot share; the read says where it is blind. The address, the two routes into the session and the copyable version of the six lines are at /startups#send.

The read in the session is the operator’s own tooling, the Cosa Sales Brain, applied to your record by the person who runs it. Inside an engagement the same Sales Brain learns your buyers, committees, deal states and next moves from the CRM and call data you already produce. Current connectors: Close and Gmail, nothing more. What it reads there is on the solution page.

Sources

1. hard2beat.vc, “10 slides you need in your deep-tech startup pitch deck” · retrieved 2026-09-09

2. The GTM Newsletter, “Founder-led sales playbook: repeatable motion” · retrieved 2026-09-09

3. Bessemer Venture Partners, “A B2B founder’s guide to generating demand from scratch” · retrieved 2026-09-09

4. sparksense.co, “B2B lead generation for startups” · retrieved 2026-09-09

5. salesmotion.io, “Series A startup sales strategies” · retrieved 2026-09-09

Questions this document answers

Can I do this read myself from my CRM?

Not from the CRM alone: it holds stages and notes, not who said what in which order. You can do it by hand from the transcripts and mails, one deal at a time, laying every touch out in sequence and marking each stakeholder at each gate with their own words. It is the loop most founders describe as the thing they cannot keep doing themselves.

What if we only have notes, no transcripts?

Then the read is built on the notes and mails and says where it is blind. An empty cell is a finding, not a failure: it names the question you never asked or the person you never heard from, which is often the reason the deal went quiet.

How is this different from asking a chat model to summarize the thread?

A summary retells the thread in your order and fills the gaps. The read keeps the buyer’s order, separates what is on record from what is inferred, and stops where the record stops. The output is not a summary; it is the frontier gate, the one move and the next closed question, per person.

What if the frontier gate is one we thought we had passed?

That is the usual finding. A gate the seller passed in a presentation and the buyer never confirmed in their own words is an assumed cell. The next move goes back to it and closes it with the buyer’s words before anything further downstream is asked.

Which deal should I send?

The one that went quiet after a yes. Not the largest, not the oldest: the one where a named person agreed to a next step that never happened. One deal, one document, in order, anonymized where you need to.