Pipeline & CRM Discipline

Two other topics already cover the two halves of a single deal: recognizing which deals are worth pursuing, and assembling the package that gets one funded. Neither answers a different question that determines whether a broker actually has a business at the end of the year — out of every ten conversations that start, how many turn into a closed loan, and does the broker know that number well enough to run the practice on it? A broker who works one deal at a time — sourcing it, qualifying it, packaging it, closing it, then starting over — is not running a pipeline. They are running a series of unrelated projects, and revenue arrives erratically because nothing is being tracked, scored, or followed up on a schedule.

Pipeline and CRM discipline is the operating system that sits underneath deal sourcing and deal packaging. It defines the stages every deal passes through, the criteria a deal must meet to advance from one stage to the next, how often and how a broker follows up at each stage, what data has to be captured and kept current, and how often the broker steps back and reviews the whole book of business rather than just the deal directly in front of them. None of this requires judgment about whether any single deal is good — that is what qualifying a deal and packaging it are for. This is about running many deals in parallel, at different stages, without any of them going dark.

Every broker eventually faces the same constraint: more suspects and prospects than there is time to chase, and not enough hours to give every deal the same attention. Deal scoring turns the decision of where to spend time into a repeatable filter rather than a mood-based judgment call. A workable rubric weights a small number of factors and produces a single 0-100 score at intake and again at every stage gate: sponsor track record and liquidity (25%), property and asset fundamentals such as occupancy, condition, and market (20%), deal size relative to the broker's minimum fee threshold (15%), likelihood the sponsor commits to an exclusive mandate (15%), documentation completeness at intake (15%), and timeline realism — whether the sponsor's expected closing date is actually achievable (10%).

The score is not a prediction of loan approval; it is a prediction of whether this deal deserves the broker's limited hours relative to everything else in the pipeline. A practical threshold: deals scoring 65 or above get full packaging attention and move toward a mandate; deals scoring 40-64 stay in nurture cadence at the prospect or qualified stage without full resource commitment; deals below 40 get archived rather than worked, because the honest cost of a low-scoring deal is not that it fails — it's the mandated or marketed deal sitting right next to it in the pipeline that didn't get the attention it needed because time went to a deal that was never going to close.

A pipeline is only as useful as the data behind it, and a CRM record that is incomplete or stale produces a forecast that is wrong in a specific, dangerous direction: it looks healthier than it actually is. A minimum data standard, enforced at every stage gate, should capture: the borrower's and sponsor's legal entity names and roles, property address and type, loan amount requested and use of proceeds, target closing date, source and referral attribution, the current stage and the date it was entered, the deal score, and — the single most important field — a next action with a due date. No deal record should ever sit in the CRM without a scheduled next action; a deal with no next action is a deal nobody is actually working, whatever stage the record claims.

Three additional rules keep the data trustworthy rather than merely complete. First, deduplicate by entity and property before creating a new suspect record — the same sponsor circling back on the same deal through a different channel should update one record, not create a second one that inflates the pipeline count. Second, a stage may not be skipped or advanced without a note documenting how the gate criteria were actually met; a stage change with no justification is usually a broker being optimistic, not a deal actually progressing. Third, every suspect and prospect with zero logged activity for a fixed window — 90 days is a reasonable default — is automatically archived, so the top of the funnel reflects live opportunity rather than accumulated dead weight. And when a deal dies, it is marked closed-lost with a required loss reason (financed elsewhere, missed minimum size, documentation never completed, unrealistic rate expectations, and so on) rather than simply abandoned — loss reasons, reviewed in aggregate, are what tell a broker whether the qualifying screen at the top of the funnel needs to change.

Cadence at the individual-deal level, shown in the table above, keeps single deals from going dark. A separate review cadence, at the level of the whole pipeline, keeps the book of business from drifting. Daily, in the first 10-15 minutes of the day, the broker triages every next action due that day or overdue across the entire pipeline — nothing worked today should be a surprise, and nothing overdue should carry silently into tomorrow. Weekly — Friday afternoon is a natural slot — the broker walks every live deal stage by stage: closing anything with an overdue next action and no real update as closed-lost with a loss reason, recomputing the pipeline's probability-weighted value, and flagging which deals need escalated attention in the coming week.

Monthly, the review shifts from individual deals to the funnel itself: actual stage-to-stage conversion rates over the trailing 90 days, compared against the broker's historical average or target. The stage with the largest gap between actual and target conversion is the bottleneck, and it determines where the broker's next month of effort goes — more suspect volume if the leak is early in the funnel, or a process fix, such as faster package turnaround or tighter lender targeting, if the leak is later. Quarterly, a CRM hygiene audit — duplicate records, orphaned next actions, suspects that should have auto-archived but didn't — is paired with resetting activity targets using the quarter's actual measured conversion rates rather than assumptions carried over from the prior quarter.

A broker starts a quarter with 40 suspects sourced through referrals, marketing, and cold outreach. Applying this broker's trailing conversion rates stage by stage: start with 40 suspects. Half of suspects become prospects: 40 × 50% = 20 prospects. Half of prospects pass the intake screen to become qualified: 20 × 50% = 10 qualified deals. Half of qualified deals convert to a mandate: 10 × 50% = 5 mandated deals. Every mandate this broker takes gets marketed (100%), so marketed also equals 5. Of marketed deals, 80% generate at least one actionable term sheet: 5 × 80% = 4 deals with a term sheet. Of those, 75% execute the term sheet and move into closing: 4 × 75% = 3 deals in closing. Finally, roughly two of every three deals that reach closing actually fund — diligence findings, third-party reports, or a late change in terms kill the rest: 3 × 66.7% ≈ 2 closed loans.

Two closed loans out of 40 starting suspects is a 5% overall conversion rate (2 ÷ 40) — a realistic order of magnitude for a broker running a disciplined but not hyper-selective top of funnel. At an average closed loan size of $9,000,000 and a 1.00% (100 bps) origination fee, each closed deal is worth $90,000, so this cohort of 40 suspects produces $180,000 in fee revenue (2 × $90,000). That works out to $4,500 of revenue per suspect sourced ($180,000 ÷ 40) — a number worth knowing, because it runs the funnel math in reverse: a broker with a $360,000 quarterly revenue target, at this same historical yield, needs 80 suspects that quarter ($360,000 ÷ $4,500), or roughly 27 a month, to hit the number without changing the fee rate, deal size, or any conversion rate along the way.

Funnel math describes a cohort moving through time. A second calculation — probability-weighted pipeline value — answers a different question a broker, or a broker's manager, needs answered at any single point in time: what is the pipeline actually worth right now, given everything currently sitting in it? Raw pipeline value, deal count times average fee, overstates this badly, because it treats a deal that just entered Qualified the same as a deal already sitting in Closing. The correction is to weight each deal by the trailing close probability for its current stage — not the stage-to-stage conversion rate used in the funnel example above, but the probability that a deal in that specific stage eventually closes.

Take a broker's live pipeline on a given day, with a $90,000 average fee per deal and trailing close-probabilities by stage of 10% (Qualified), 20% (Mandated), 35% (Marketed), 60% (Term Sheet), and 85% (Closing): 8 qualified deals contribute 8 × 10% × $90,000 = $72,000. 5 mandated deals contribute 5 × 20% × $90,000 = $90,000. 4 marketed deals contribute 4 × 35% × $90,000 = $126,000. 3 deals at term sheet contribute 3 × 60% × $90,000 = $162,000. 2 deals in closing contribute 2 × 85% × $90,000 = $153,000. Summed across all five stages, the probability-weighted pipeline value is $603,000 ($72,000 + $90,000 + $126,000 + $162,000 + $153,000).

Compare that to the pipeline's raw face value: 22 total deals (8+5+4+3+2) at $90,000 each is $1,980,000. The weighted figure, $603,000, is only about 30% of that raw number — and it is the weighted figure that belongs in a revenue forecast, a compensation projection, or a conversation with a lender or manager about what the quarter actually looks like.

A broker's calendar should mirror the pipeline stages it has to feed and service. A workable structure blocks Monday, Wednesday, and Friday mornings, roughly 8:00-10:00, for prospecting — outbound calls and messages that move suspects to prospects — because this is the activity most easily crowded out by urgent-but-not-important closing fire drills, and the one whose absence shows up as a hole in the pipeline eight to twelve weeks later. Tuesday and Thursday mornings are reserved for deal execution: packaging qualified and mandated deals, and coordinating diligence on deals in closing — work that requires uninterrupted focus.

Afternoons across the week stay reactive by design. Lender calls, sponsor questions, term-sheet negotiation, and diligence requests all arrive on their own schedule, and a broker who fills afternoons with more prospecting has no slack left to hit a same-day response commitment on a term sheet. The first 15 minutes of every day are the CRM triage described above — non-negotiable, before any other work starts. Friday afternoon, 3:00-4:00, is reserved for the weekly pipeline review, and the last business day of the month adds the monthly funnel-conversion review on top of it. What is deliberately absent from this structure is unscheduled time for prospecting: it is the first activity to disappear under pressure precisely because its payoff is invisible for two to three months, which is exactly why it has to be protected on the calendar rather than left to whenever time allows.