A pipeline review is not a status meeting. It is a decision meeting with one subject: which lines move and which lines die. Everything else that happens in the hour — the account updates, the anecdotes, the encouraging news about a buyer who has gone quiet — is either evidence for one of those two decisions or it is conversation.
The distinction has consequences for how the meeting is prepared. A status meeting needs updates, so it is fed by whoever speaks most readily. A decision meeting needs claims and their evidence, so it is fed by a template that asks, for every line, what the stage is and what document puts it there. Where the template is weak the review drifts into a report on activity, and the pipeline comes out of the hour exactly as it went in: same lines, same values, same dates, a month closer to the end of the quarter.
This piece sets out the template as a working instrument: the eleven fields that make a line reviewable, the sixteen questions it asks of each line, the four outcome categories that close every line, a worked review of nine lines that removes $258,000 from a board carrying $528,000, the limits of a forecast built this way, and the one decision the meeting exists to take.
One premise runs through all of it. A pipeline is a set of claims about the future, each made by somebody with an interest in the claim being believed. The review’s job is not to disbelieve them. It is to sort them by what supports them, and then to make the sort visible — because the value of a pipeline is not that it is large. It is that the number in it can be defended in front of the person who has to staff against it.
What the meeting decides
Four outcomes, and every line leaves the meeting in exactly one of them. The value of fixing the categories is that the meeting stops producing a fifth outcome by default, which is the one that has quietly been in use: the line that is neither advanced nor closed, carried at its original value, and reviewed again next month without anyone asking why it survived.
The four are advance, hold, die and requalify. Advance means the line has the evidence its next stage requires, a named next step, and a date. Hold means the line is real but something specific is missing, and the missing thing is written down with a date by which it will be either supplied or the line dies. Die means the line has no evidence behind its stage and no owner willing to date the gap — it is removed from the board, and if the account is worth another attempt it returns as a new line with a new stage. Requalify means the account is real but the opportunity was mis-staged: it goes back to discovery with the open questions named, at whatever value the record supports, which is often nothing yet.
The categories are decisions rather than descriptions, and each one carries a commitment. Table 3 below pairs each outcome with the test that produces it and what it obliges the meeting to do afterwards.
The fields
Eleven fields per line. Fewer than eleven and the meeting argues about facts; more and the template becomes a form that people fill in badly and read less. Each field is here because a question in the next section depends on it.
Table 1. The eleven fields of a pipeline line, what belongs in each, and where it comes from.
| Field | What goes in it | Where it comes from | Why it matters |
|---|---|---|---|
| Account and contact | The legal entity, the person, and their role in the decision | The record, not the memory of the account owner | The person who answers is often not the person who signs |
| Stage | One of the fixed stage definitions, named rather than numbered | The stage definitions the team agreed | A stage nobody can test is a feeling with a label |
| Stage evidence | The document or event that justifies the stage | The record: a signed acceptance, a brief, a letter | Without this field every other field is a claim |
| Stage entered | The date the line entered its current stage | The record | A line that has not moved in a quarter is a fact the review needs |
| Next step | The action, the owner by name, and the date | The last contact | “Awaiting feedback” is not a next step; it is a wait with no owner |
| Value | A range, with the document or the conversation it came from | The quotation, the budget line, or the buyer’s own words | A single figure with no source is a number invented once and repeated |
| Decision date | The date the buyer decides, and what fixes it | A renewal notice, a board calendar, an audit date | A date with a document behind it behaves differently from a date with a hope behind it |
| Authority | Who signs, and the threshold above which it moves | The buyer, or the person they named | Value without authority is a number nobody has agreed to spend |
| Blocking unknown | The one thing that would change the shape of the offer if it were true | The open questions from the call | Naming it gives the hold a condition instead of a hope |
| Last contact | Date, channel, and what was actually said | The record | The gap between last contact and the review is a measurement |
| Last review’s outcome | Advance, hold, die or requalify, with the date | The previous review | A hold that has held three times is a die that nobody has said out loud |
The sixth field does the most work for the least effort. A value entered as a single number cannot be examined; a value entered as a range with its source can be, and the source is usually the thing that turns out to be missing. The two fields that follow it — decision date and authority — then determine whether the range is even relevant, because a value the buyer cannot authorise, decided on a date nobody can source, is not a forecast; it is a figure with a date attached.
The questions
Sixteen questions, asked of every line in the same order. The order matters: evidence first, because a line whose stage cannot be evidenced does not need its value discussed, and discussing it wastes the meeting’s scarcest resource. The questions are grouped by what they test.
Table 2. The sixteen questions the review asks of each line, in the order it asks them.
| Group | The question | What a bad answer tells you |
|---|---|---|
| Evidence | Which document puts this line at this stage? | The stage was entered on an activity, not on a commitment |
| Evidence | When did that document arrive, and who read it? | The stage is older than the line suggests, or newer than the relationship |
| Evidence | If I asked the buyer for that document today, would they send it? | The evidence is the account owner’s summary of a conversation |
| Evidence | What has changed in this account since the last review? | Nothing has changed, which is itself the finding |
| Owner | Who is doing the next thing, and on what date? | The next step belongs to the buyer, so nobody owns it |
| Owner | What does that person need from us to do it? | The step is blocked on a document nobody has written |
| Owner | Who at the buyer is going to say no, and what would they say? | The objection has not been sought, so the line has only one view |
| Value | What is the range, and where does each end come from? | The figure is a memory, or the seller’s opening ask |
| Value | What would the buyer compare this against? | The comparison set is unknown, so the value is unanchored |
| Value | What has this buyer spent on the nearest equivalent? | There is no precedent, which makes the value an assumption |
| Timing | What date does the buyer decide, and what fixes it? | The date is a quarter, or the seller’s quarter rather than the buyer’s |
| Timing | What has to be true before they can start? | The gating condition is unknown, so the date cannot be tested |
| Timing | What happens if they do nothing — who notices? | No consequence on the buyer’s side, which is the strongest predictor of requalification or death |
| Movement | What did we say we would do after the last review, and did we? | The team’s own commitments are slipping, which invalidates the rest |
| Movement | Is this line further forward than it was at the last review, in evidence rather than in words? | The line has been carried on sentiment for one or more cycles |
| Movement | If this line died today, what would we lose? | The honest answer is often nothing yet, which is why it should die |
Two disciplines make the list usable. The first is that every question is answered from the record, not from recollection — a review where the account owner says “I think the proposal went over in September” has produced a task for the following day rather than an answer. The second is that no question asks for a probability. There is no percentage in the template, and no place to write one, because a probability entered into a pipeline line is a judgement that cannot be audited and will be remembered as a commitment. Where a forecast needs weighting, it needs a method the team can defend, and inventing a per-line confidence on the way into a meeting is not one.
The outcome categories
Table 3 is the part of the template that turns the meeting into a decision. Each row names an outcome, the test that produces it, and what it obliges.
Table 3. The four outcome categories, the test that produces each, and what each obliges the meeting to do.
| Outcome | The test | What it obliges |
|---|---|---|
| Advance | The evidence for the next stage exists or is dated, the next step has a named owner, and the decision date traces to a document | Carry the value into the forecast with the date it will be reviewed again |
| Hold | The line is real, one specific thing is missing, and an owner has dated the step that supplies it | Keep the line on the board at a value the record supports, with the condition and its date written down |
| Die | No evidence behind the stage and no owner willing to date the gap | Remove the line; if the account is worth another attempt it re-enters as a new line at a lower stage |
| Requalify | The account is real but the opportunity was staged on an activity rather than a commitment | Send it back to discovery with the open questions named, at whatever value the record supports |
The category that requires the most discipline is hold, because it is the one that can be used to avoid deciding. A hold with a condition and a date is a decision: the meeting has said what would have to arrive, by when, and what happens if it does not. A hold without those two things is the fifth outcome in disguise — the line continues, the value continues, and the review has produced a note rather than a decision. The test for a legitimate hold is therefore one sentence long: can somebody who was not in the room read the condition, the date and the consequence?
The category that requires the most nerve is die. A dead line costs the account owner something to admit and costs the team nothing to keep. That asymmetry is the reason pipelines inflate, and it is the reason the review has to be a meeting with a chair rather than a round-robin of updates: the person who benefits from the line staying alive is usually the person presenting it.
Worked example: nine lines, $528,000, $258,000 removed
The example is constructed for this piece. A nine-line pipeline is reviewed on 7 October 2026 for a quarter ending 31 December. The board opens at $528,000 across the nine lines, and each line is tested against the sixteen questions in the order set out above.
Five of the nine cannot produce the document their stage requires. Of those five, three have no owner willing to date the gap and die; two describe accounts that are real but were staged on an activity — a site visit, a proposal sent without an accompanying acceptance — and requalify back to discovery. Four survive: three advance, with evidence, a named next step and a decision date traced to a document; one holds, because a single item is missing and its owner has dated the step that supplies it.
Table 4 sets out the arithmetic.
Table 4. One review’s arithmetic: the board as opened, the value removed by the evidence test, and the number the meeting signed. Constructed for this piece.
| Line | Amount | What it measures |
|---|---|---|
| Board as opened, nine lines | $528,000.00 | The pipeline as reported before the review |
| Value on lines that could not evidence their stage | $258,000.00 | 48.9 per cent of the board, described in claims rather than documents |
| — of which died | $188,000.00 | Three lines, no evidence and no dated gap |
| — of which requalified | $70,000.00 | Two lines, real accounts, staged on activity |
| Value on lines that survived the test | $270,000.00 | Four lines, 51.1 per cent of the board |
| — of which forwarded | $218,000.00 | Three lines with a named next step and a dated decision |
| — of which held | $52,000.00 | One line, one condition, one date |
| Number the meeting signed | $218,000.00 | 41.3 per cent of the board as opened |
The arithmetic is plain. $188,000.00 + $70,000.00 = $258,000.00 removed; $218,000.00 + $52,000.00 = $270,000.00 carried; and $258,000.00 + $270,000.00 = $528,000.00, which is the board the meeting opened with. The signed number is $218,000.00 rather than $270,000.00 because a hold is a decision to wait, and a number that includes waiting is a number that has already been promised to somebody outside the room.
Two things in the example are worth separating. The first is that the five failing lines were not dishonest. In each case somebody had done real work — a visit, a proposal, a conversation that went well — and had recorded the work as progress. The template did not catch a lie; it caught a category error, which is the more common failure and the harder one to see, because the account owner’s account of the line is accurate about everything except what it implies.
The second is the disposition of the $258,000.00. Removing a line is not the same as losing the account. The three that died were removed because nobody would date the gap, which is a statement about the opportunity rather than about the relationship; two of the three re-entered the board in the following quarter at a lower stage and a smaller value once a brief arrived. The two that requalified kept their accounts and lost their stages, which is the cheapest correction available: the discovery questions in the account file were already written, and the next call had a purpose it had been missing.
One further figure belongs in the meeting’s minutes and not on the board: the rate at which the team’s own commitments from the previous review were met. Fourteen had been made; nine were done. That number is not a measure of the pipeline. It is a measure of whether the pipeline’s next-step dates mean anything, and a review that ignores it is testing a forecast against a team whose reliability it has not checked.
What a forecast built this way cannot promise
Seven limits. Each is a question a reader will ask of the number in Table 4, and each has the same answer: the forecast cannot tell you that.
It cannot promise revenue. It reports what the record supports on the date of the review. Between the review and the quarter end, buyers change jobs, budgets are frozen, a contract renewal slips, and a competitor is invited in. A forecast that describes its evidence is a forecast that can be audited after the fact; it is not a prediction, and the difference is the whole of this section.
It cannot see the deal that never entered the record. Coverage is unknown. The account where nobody wrote the brief down, the referral that arrived as a phone call, the renewal nobody thought of as a sale — none of them is in the board, and the forecast cannot tell you what share of the eventual quarter came from outside it. A number built this way is precise about the lines it contains and silent about the lines it does not.
It cannot read a buyer’s budget cycle. A decision date traced to a document is a date somebody wrote down, not a commitment to decide. The renewal notice says the contract expires on 31 March; it does not say the buyer will run a process. The field improves the date from a hope to a claim with a source, and a claim with a source is still a claim.
It cannot correct itself. The evidence test is applied by the same people whose lines are being tested, reading the same record they wrote. Where an account owner is generous with the definition of evidence, the test passes and the forecast inflates; where an owner is harsh, a real line dies early. The template makes the standard visible, which is a real improvement on invisibility, and it does not remove the judgement.
It cannot distinguish hold from lost. A hold has a condition and a date, and the date can be met or not. What it cannot do is price the cost of waiting: the attention a held line consumes, the account owner’s time, the buyer’s patience. Those costs are real and they are nowhere in Table 4.
It cannot be compared with last quarter unless the stage definitions held. Change the definition of a stage between reviews and the trend across them means nothing — the board is measuring a different thing in each period. This is the least discussed and most common way a pipeline stops being informative while continuing to be produced.
It cannot produce a percentage. Nothing in the template assigns a probability to a line, and the review does not derive one. If a number arrives weighted, it came from a method the team chose; that method should be stated, and it is not stated here, because this publication holds no measured rate at which lines at any stage convert. A weighted pipeline with an unstated method is a single figure doing the work of an argument.
What it costs
The review is document work, and it is priced at the measured Base rate of $0.00700 a call, with the counts stated as planning assumptions.
Fourteen calls prepare and close one review of nine lines: one read per line to assemble its evidence and its last outcome, one to build the pack in the template’s order, one to write the four decisions back into the record, and three to draft the follow-up notes for the hold and the two requalifications. Fourteen calls is $0.09800, or $0.01089 a line. At the metered ceiling of $0.00955 the same fourteen calls are $0.13370 — $0.03570, or 36 per cent, more, because a ceiling is what an agent authorises rather than what it spends. Re-reading the nine line summaries on Qwen flash at its measured $0.00875 adds $0.07875, taking the review to $0.17675: 80.4 per cent more than $0.09800 for a second read of the same record, which is worth doing only where the pack is the input to a decision somebody outside the team will rely on.
Twelve monthly reviews at this size are 168 calls, $1.17600 a year. The 0.50 USDC credited to a workspace is seventy-one calls at this rate: five reviews of nine lines, $0.49000, with a single call to spare. Five reviews is five months of the process at the counts above, which is the useful way to read the balance — the review is cheap per month and the constraint is the number of months it covers, not the price of an individual meeting.
Set against what the review removes, the $258,000.00 the example took off the board is 2.63 million times the $0.09800 the review cost. The comparison is not an argument for the template, because a review that removes nothing is also $0.09800, and a badly prepared review can remove the right value for the wrong reason. It is an argument about where the cost of a review actually sits: not in the calls, which are a rounding error, but in the hours of the people in the room and in the accuracy of the record they are reading.
The decision
Fix the stage definitions to documents, apply the four outcomes to every line at every review, and carry only the value whose stage evidence survived the test. Then write the number down with its date and the number of lines behind it — $218,000 over four of nine lines, on 7 October, say — and read against it at the next review. A forecast that records its own basis can be improved. A forecast that records only its total can only be argued with.
Two smaller decisions follow from it and are worth taking at the same meeting. Define what a hold obliges, in one sentence, and refuse any line whose condition cannot be read by somebody who was not in the room. And define what evidence each stage requires, in the stage sheet rather than in the review, so that the argument happens once rather than nine times an hour.
The agent’s part in this is narrow and it is the part that makes the meeting possible. It reads each line’s record, extracts the fields in Table 1, applies the sixteen questions as checks against the record rather than as questions to a person, builds the pack in the template’s order, and writes the meeting’s decisions back into the record with their dates. It cannot decide that a line advances, because that decision rests on whether a document means what the account owner says it means. It cannot choose between two dates. And it cannot sign the number the meeting signs, which is the one figure in the process that has to belong to a person with a name.
Where this goes
- The pillar: x402.review/2026/10/31/commercial-operations/ — one record, two readers, and the approvals between them
- The role that keeps the record: studio.x402.press/roles/sales — the proposal, the follow-up and the mailbox it sends from
- The record the review reads: CRM hygiene for small teams — the fields, the duplicates and what a weekly pass costs
- What happens to a held line: following up a lead without losing it — a reason per touch and a stop rule written first
- The service it buys: x402.press/services/deepseek/ — the canonical price, seven chains
- The review: x402.review/2026/10/31/deepseek-review/ — the measured handshake and what it does not promise
Method note
The price was read from the live 402 challenge on Base on 31 October 2026, a Saturday and therefore outside DeepSeek’s weekday peak windows: 7000 atomic units under exact, which is $0.00700, and 9550 under upto at the max_tokens tested. Qwen flash’s $0.00875 was read from its own challenge in the same way. Call counts are planning assumptions, stated where they are used: one read a line, one to build the pack, one to write the decisions, three for the follow-up notes.
The nine lines, the amounts in Table 4, the dates and the fourteen prior commitments are constructed for this piece; they describe no real company, no real account and no real quarter. The amounts are illustrative and are not offered as typical deal sizes — this publication holds no average deal size, no conversion rate and no win rate, and quotes none. The eleven fields, the sixteen questions and the four outcomes are the discipline as the sales role requires it, not the results of an audit of a live pipeline. No accuracy rate for extracting a line, no time-to-review and no share-of-board removed is quoted as a benchmark anywhere here, because none has been measured.
x402.review is written and edited by N.P. Vincent. The publication is connected to x402.press, which sells the services reviewed here; that interest is disclosed on every piece that touches a service, and it is stated here rather than left to inference.
