Winning by Design * Dogfood Plan · WS3

SPICED Pulse: Dogfood on WbD First

Before we sell SPICED Pulse to a single client, we run it on ourselves. We score our own Grain calls against SPICED, join them to our own HubSpot pipeline, and run the weekly coaching loop on a small slice of the WbD sales team. Three reasons. First, it makes us better: our own reps get real coaching on Impact and Critical Event. Second, it gives the MDs a working system on real WbD data to demo, not a mockup. Third, it de-risks the customer offer, because we have already lived every step we are about to sell.

Calls from Grain Pipeline from HubSpot Scored in Claude All on our own side
Our stack, our data

Everything runs on WbD's own tools. The data never leaves.

The dogfood pilot uses the exact stack we already pay for: Grain for the calls, HubSpot for the deals and the org hierarchy, and our own Claude instance for the compute. Nothing new to buy, no third party in the loop, no recordings moving off our side. This is the identical three-layer shape we will later stand up on a client's stack, run first on ourselves.

Layer 1 · Data
Grain plus HubSpot
Grain call transcripts, segmented and speaker-resolved, joined to HubSpot deals and their owners on the deal id. The name-join is what kills the quote misattribution that broke earlier attempts.
Raw stays here
Layer 2 · Compute
Our Claude instance
Skills score SPICED, extract the four-part coaching card, and answer role-scoped queries where the data already lives. Outputs are derived insight, never raw recordings.
Runs where data lives
Layer 3 · Interface
The app
Role-scoped views and the query box. A derived-only gate is the single egress chokepoint. A short attributed quote may cross, bounded to a rolling 120-day window. The recording itself stays in Grain, reached by deep link.
Derived-only crosses
Why us first is the honest test

If SPICED Pulse cannot move behavior on our own reps, on our own calls, we have no business selling it. Running it on WbD is the cheapest, fastest, most credible proof we can build, and every problem we hit is a problem we solve once for ourselves and never again in front of a client.

Pilot scope

A small slice of the team, a short window of calls.

Keep the pilot small enough to run by hand and honest enough to show signal. One or two managers, a handful of reps, two to four weeks of Grain calls. We are proving the loop, not boiling the ocean. Names below are placeholders to be filled once we pick the pilot group.

Who

One to two managers, their reps

Manager A and optionally Manager B (to be filled), with their direct reps: Rep 1, Rep 2, Rep 3 and so on (to be filled). Enough reps to see a team pattern, few enough that Dan can inspect every card by hand in the first weeks.

Window

2 to 4 weeks of Grain calls

A rolling 2 to 4 week window of the pilot reps' recorded Grain calls. Enough call volume to score reliably (target at least 5 calls per rep per week), short enough that the data is current and the loop stays weekly.

Which deals

Active pilot-rep deals in HubSpot

The open deals owned by the pilot reps, pulled from HubSpot with their amount, stage, and close date. These are the deals whose SPICED completion we track week over week and the deals the Deal Pulse view reads.

Which calls get scored

Calls tied to a pilot deal

A Grain call is scored when it is a pilot rep's call attached to a pilot deal inside the window. We score for Impact and Critical Event completion first, the two behaviors with the strongest revenue link, then quality 1 to 5 as the coaching signal.

Deliberately out of scope for the pilot: the full team, historical calls beyond the window, non-English calls, and any use of the scores for evaluation or accountability. This is development framing, held by the rep and their manager.

The weekly loop

Anchored to the rituals we already run: Monday review, Tuesday 1:1.

We do not invent a new meeting. The loop plugs into the two starred moments WbD managers already hold, the Monday pipeline review and the Tuesday 1:1. The system does the heavy lifting. The one required act from the manager is small and concrete: deliver one coaching card per rep.

  • 1Over the weekend, the system scores. Claude ingests the week's Grain calls, joins them to HubSpot deals, and scores Impact and Critical Event completion plus quality for each pilot rep, with a verbatim quote behind every card.
  • 2Monday, pipeline review. The manager brings the pipeline diagnostics into the existing Monday review: which pilot deals have a complete Impact and a validated Critical Event, and which do not. Weak-SPICED big deals surface as both forecast risk and coaching trigger.
  • 3Tuesday, the 1:1. The manager delivers one coaching card per rep in the Tuesday 1:1 they already hold. The card is the four parts: what the customer said, what the rep said, the verdict and why, the one next action. The tool selects it; the manager delivers it.
  • 4The rep acts before the next call. One coaching action per rep: quantify the impact, test the critical event. Development framing, held by the rep and their manager, tracked as a week-over-week improvement, not a grade.
  • 5The next week re-scores and shows the delta. The following weekend's scan re-scores the coached behavior from the tagged segments, so the rep and manager see the week-over-week movement. The loop closes and repeats.
The one required manager act

Deliver one coaching card per rep each week in the Tuesday 1:1, then mark it delivered. That is the whole ask. Everything upstream is automated, and the scoreboard is the reps' week-over-week improvement, not manager activity. A ritual this small is a ritual that actually happens.

What the MDs demo

Once we have run it on ourselves, the MDs open a working system.

After a few weekly cycles on real WbD data, the MDs have three surfaces to show live in a client conversation. Not slides, not a mockup. The same system we are running on our own team, reading our own Grain calls and our own HubSpot pipeline.

Demo 1

The overview

Team average against the SPICED standard and exactly where the gap sits, on our own reps. The manager and executive read the same picture at a glance: where the team is strong, where it is behind, and on which deals.

Demo 2

The four-part coaching card

The demo that lands. A real WbD rep's call: what the customer actually said, what the rep actually said, the verdict and why, the one next move. Every score deep-links back to the moment in the Grain recording.

Demo 3

Our own Deal Pulse

Every pilot deal with its SPICED completion next to its amount, weighted value, confidence, and risk flag. Our real pipeline, so the leading indicator and the lagging one meet on one screen the MD can point at.

Why demoing on ourselves is the whole point

An MD who opens SPICED Pulse on real WbD data can say, truthfully, "this is us, and here is what it found." That credibility is what turns a pitch into an engagement. We prove it on ourselves, then help the client stand up the identical shape on their own stack.

The proof metric, on us

Impact plus Critical Event completion lift, measured honestly.

The single thing we are proving on ourselves: that weekly coaching of two SPICED behaviors, Impact and Critical Event, driven by each rep's own Grain calls and delivered in the Monday review and Tuesday 1:1, changes what our reps do on calls. Success is behavior changed plus a revenue signal, not a tool shipped. We hold ourselves to leading indicators first and read the lagging ones with appropriate patience.

Leading indicators, what we expect to move first

Weekly completion rate

The share of active pilot deals showing a quantified, customer-validated Impact and an identified, tested Critical Event, scored from Grain. This is the headline. We watch it rise from our own baseline and hold for several consecutive weeks.

Coaching cards delivered

The share of pilot managers delivering at least one card per rep per week, plus each rep's week-over-week quality delta on the coached behavior. Ritual adherence and rep improvement are the earliest honest signals that the loop is real.

Lagging indicators, read with patience

Win rate and stage conversion

Win rate on the coached deals and stage-to-stage conversion on the deals the behavior touched. A guardrail matters here: conversion on coached deals should hold or rise, never fall, while the leading indicators move.

Sales cycle

Cycle length on coached deals over time. On a small pilot in a short window this is a directional read, not proof. We pre-agree that the leading metric is the success signal and the lagging metric is confirmation that arrives later.

The honest framing

A small pilot over a few weeks will show behavior change before it shows revenue. We say that up front. If completion lifts and holds and conversion does not fall, the pilot succeeded, and the revenue read is the confirmation that follows across the next deal cycles. Claiming a revenue win in four weeks would be the dishonest version, and we do not make that claim.

Timeline

Setup, first run, first demo, review point.

A simple four-step arc. Durations are working estimates to pressure-test with Shari, not fixed commitments. Each step ends with something we can point at.

Step 1 · Setup
Wire the stack
Connect Grain and HubSpot, pick the pilot managers and reps, pull the deal list, and stand up the scoring skill and the derived-only gate. Ends when the first calls score cleanly with correct speaker attribution.
Step 2 · First run
First weekly loop
The first full weekend scan feeds the first Monday review and Tuesday 1:1. Managers deliver the first cards, reps take the first actions. Ends when every pilot rep has received one delivered card.
Step 3 · First MD demo
Show it working
After several cycles, an MD walks the overview, a coaching card, and our own Deal Pulse on real WbD data in an internal dry run, then in a live client conversation. Ends with the first MD-led demo delivered.
Step 4 · Review
Read the signal
Review the leading indicators against our baseline, decide keep, widen, or adjust, and write back what we learned into the client engagement design. Ends with a go decision on scope.
Owners

Dan runs it. Shari operationalizes the cadence.

Two clear owners so nothing falls between them. Dan owns the build and the coaching standard. Shari owns the weekly rhythm and the manager ritual actually happening. The pilot managers own delivering their cards.

Dan Smith
Runs it
  • Owns the build: the scoring skill, the coaching card, and the derived-only gate on our own stack.
  • Sets and holds the SPICED standard of good, and inspects the early cards by hand for quality.
  • Owns the MD demo readiness: what the MDs show and how they frame it.
  • Owns the proof read: leading indicators against baseline, and the honest revenue framing.
Shari
Operationalizes the cadence
  • Owns the weekly rhythm: the scan lands before Monday, the cards reach managers in time for Tuesday.
  • Drives manager ritual adherence: one card per rep per week, delivered and marked.
  • Coordinates the pilot managers and keeps the loop on schedule week to week.
  • Tracks delivery and completion so the leading indicators are always current.

Pilot managers (Manager A, Manager B, to be filled) own delivering their coaching cards in the Tuesday 1:1. Pilot reps own acting on the one coaching action before the next call.