For review — not a proposal

Decisions register

The roadmap and informational documents specify what the system does. They do not specify a number of things the software cannot run without. Each one below was a judgment call made in order to ship Phase 1. Every one is wrong until you say otherwise, and every one is a single edit to change.

Assumed No guidance in the documents — needs your call Extended Documents implied it; I made it concrete Followed Straight from the documents

The scoring weights

Assumed

Documents say: Nine qualification questions to capture before a prospect gets partner time: client type, revenue, services, entities, states, urgency, current provider, engagement level, budget.
Gap: No weighting. Nothing says revenue matters more than urgency, or how much a second entity is worth.
I chose: Revenue up to 30, engagement level 24, urgency 16, client type 14, complexity (entities and states) 15, service mix 12, no incumbent provider 3. Capped at 100.
Why: Revenue and stated engagement level are the two strongest predictors of a recurring relationship, which is the stated strategy. Complexity is weighted because it is a Cirology pricing driver, not because it predicts fit.
To change: One table in scoring.py. Tell me a prospect who should be an A and is not, and I will show you which weight is wrong.

Where A, B, C and D begin

Assumed

Documents say: Four lead grades with four routing actions — A gets an immediate consultation, D gets a polite automated response.
Gap: No thresholds. The grades exist; nothing defines who lands in them.
I chose: A = 70+, B = 50–69, C = 30–49, D = below 30. Only A unlocks the calendar.
Why: Tuned so that on a realistic mix, roughly half the pipeline earns partner time. Too loose and the gate does nothing; too tight and you starve the calendar.
To change: This is the most consequential number in the system and the one most worth arguing about. It is one line.

The interview can re-grade a lead

Extended

Documents say: Qualification happens at intake. Discovery happens later, at stage 3.
Gap: Nothing says what happens when the interview reveals something the form did not — an open tax issue, a lender review, active succession planning.
I chose: Five urgency findings each add +5 (capped at +15), which can lift a B into A territory and unlock the calendar.
Why: A mid-sized prospect with an open tax issue and a bank review coming is worth more than the intake form can see. Judging them only on the form loses real work.
Risk: It lets prospects raise their own grade. If that is wrong, the bonus goes to zero and the form alone decides.

How a cross-sell signal gets caught

Extended

Documents say: Two signal tables — "business owner has growth challenges" maps to Growth & Revenue, "poor books" maps to Accounting services, and so on.
Gap: The tables describe signals. Nothing captures one. As written, a partner has to notice the signal and remember to log it.
I chose: Sixteen specific interview answers are wired directly to those signals. Answering the question is logging the signal — no one has to remember anything.
Why: This is the part of the model that depends on nobody forgetting, which is exactly what software should absorb. It is also the hardest piece to buy off the shelf.
To change: The map is one table. Add a question, add its signal.

Which questions get asked at all

Assumed

Documents say: An intake form, then a discovery conversation.
Gap: No interview script exists in the documents.
I chose: Nine questions, branching on client type — an individual is never asked about bookkeeping, dashboards, capital needs or technology. They get five questions; a business gets nine.
Why: Asking a household filer about their KPI dashboard reads as a form that was not built for them, and it costs completions.
To change: Questions are a list. Reorder, reword, add or gate them by any field on the record.

The pipeline refuses to skip stages

Extended

Documents say: Eight sales stages, each with a purpose.
Gap: Whether the stages are a description or a rule. Most CRMs let anyone drag a deal anywhere, which is why stage data is usually worthless.
I chose: Stages advance one at a time. Jumping stage 2 to stage 5 is rejected. Going backwards is always allowed. Touches release in order. A lost lead is frozen until someone reopens it.
Why: Enforced stages make the pipeline report true. If a deal sits at Proposal, someone actually did Solution Design. Without that, close-rate numbers are decoration.
Risk: It will annoy someone who wants to log a deal that arrived late. That friction is the point, but it is a choice, not a law.

Nothing sends itself

Followed

Documents say: Human checkpoints. Nearly autonomous, not fully autonomous. No unsupervised outbound from the primary domain.
I chose: Enforced in the rules layer, not the screen — a person releases every touch and every stage move. Touches are tracked but nothing is emailed yet.
Why: Putting it in the rules means it holds even if someone later builds a different interface on top. A checkpoint that lives only in a button is not a checkpoint.
Open question: When sending is switched on, does a touch go out on approval, or on a timer with an approval window? That changes the design.

Which firm owns the lead

Assumed

Documents say: Cirology and TAC are separate, complementary firms with their own funnels and their own messaging.
Gap: One form, two firms. Nothing says who owns a prospect who wants both.
I chose: Advisory services only → TAC. Anything involving CPA work → Cirology, flagged cross-sell. When a lead's firm changes, its 6-touch sequence is rebuilt in that firm's language automatically.
Why: Cirology is the trust relationship in the stated model, so it leads when both are in play. That is a commercial decision, not a technical one — it may be wrong.
To change: Flip it, or split the record so both firms hold their own pipeline entry.

Every advisory opportunity names its pillar

Extended

Documents say: Four pillars: Growth & Revenue, Operational Improvement, Financial Strategy, Wealth Accumulation & Preservation.
Gap: The pillars are described but never attached to anything the system produces.
I chose: Every advisory opportunity is tagged to exactly one pillar, so a partner sees which practice the work belongs to rather than that "an opportunity exists".
Why: It is how the work eventually gets assigned, priced and staffed. Tagging at capture costs nothing; tagging later means re-reading the file.

The onboarding checklist contents

Assumed

Documents say: Engagement letter, payment setup, portal activation, onboarding checklist, kickoff.
Gap: No actual checklist.
I chose: Eight items for a Cirology engagement, a different eight for a TAC engagement, generated automatically the moment a deal reaches Engagement Close.
Why: Closing is exactly when things get dropped. The checklist should exist before anyone thinks to make one.
To change: This one should almost certainly be replaced with your real intake process.

What I need from you

Four of these — 01, 02, 08 and 10 — are commercial decisions wearing technical clothes. I can hold an opinion on them, but they are yours. The other six are safe to leave until the system has run against real leads.

The fastest way to correct any of them: name a real prospect you have had, and tell me what the system should have done with them. That is worth more than a spec.

← Back to the overview