At 2:07 p.m. on a Tuesday in July, a high-value call lands on the dispatch board — Bel Air homeowner, no-cool complaint, 16-year-old system, the kind of call that historically converts to a $12K-$18K replacement quote at a 38% MPR. Every premium-band tech is on a stop. Jose is wrapping a service call 35 minutes south; Marco is locked on the Marin Park install crew; the only available tech is the rookie at 14% MPR and $1,800 average ticket. Old-school dispatch logic stops there: rookie or hold. The 2026 dispatch AI does not stop there. Dispatch Pro, Sera, or FieldEdge re-evaluates the entire board every 10 minutes against the new call's predicted revenue, ranks the rebalance options (pull Jose from the service call, push the rookie to the Bel Air with a senior advisor backup, accept a 90-minute window slip on a lower-value stop, hold for the 4:30 p.m. slot when Jose clears), and proposes the highest-expected-revenue route across the next four hours of board state. The proposal lands on the dispatcher's screen in 90 seconds. The dispatcher has 5-15% override authority and a documented override criteria document from L2 Ch3 Lesson 1. This lesson is the ops manager's playbook for the re-dispatch loop — the named workflow that turns a high-value 2 p.m. call into the right routing decision rather than the convenient routing decision, what the dispatcher overrides versus what the loop handles automatically, and the 10-minute re-evaluation cadence that drives the 12-18% dispatch yield lift across the day rather than at the morning huddle.

What the Re-Dispatch Loop Is

The re-dispatch loop is the continuous rebalancing of the day's board against new information as the day progresses. Every 10 minutes, Dispatch Pro, Sera, or FieldEdge re-runs the four-variable optimization from L2 Ch3 Lesson 1 (predicted job revenue × tech historical close × travel × capacity) across every open call and every available tech. New information that may have changed in the last 10 minutes: a call landed, a tech finished, a parts run extended, a customer cancelled, an emergency hit, traffic shifted, weather changed, an install delay propagated. The model re-solves the entire board against the new state and proposes adjustments to the assignments still ahead.

The distinction the ops manager defends: the re-dispatch loop is not the morning board build. The morning board build is one-time at 6:30 a.m. — the dispatcher walks the team through the AI-proposed starting board and the override reasoning. The re-dispatch loop runs continuously from 7 a.m. through end-of-day, proposing rebalances every 10 minutes. The morning build owns 0-7% of the day's dispatch yield lift; the re-dispatch loop owns 93-100% of it because the day's reality changes 50+ times between the huddle and the last on-my-way text.

The re-dispatch loop is also what catches the high-value-call-lands-at-2-p.m. scenario. The morning board build had no knowledge of the 2:07 p.m. Bel Air call; the call did not exist at 6:30 a.m. The re-dispatch loop sees the call land in CSR, evaluates predicted revenue ($14K with a fat tail to $24K), evaluates every available tech's expected revenue on the call (Jose at $5,320 expected, Marco unavailable due to install crew lock, rookie at $1,960 expected with high variance), proposes the rebalance (pull Jose from the lower-EV service call, slip that customer's window 90 minutes with notification, route the rookie to a backup low-stakes call). The dispatcher reviews in 90 seconds, applies override criteria, approves or modifies. The call gets the right tech because the loop is running continuously, not because the dispatcher noticed at 11:30 a.m. that the board needed reshuffling.

The 10-Minute Cadence and the Shop Rhythm

The 10-minute re-evaluation cadence from Dispatch Pro, Sera, and FieldEdge is the operational heart of the loop. Old-school dispatch logic re-runs at major events — call lands, tech finishes, emergency hits. The 2026 AI dispatch logic re-runs every 10 minutes regardless of whether an event occurred, because the shop's reality changes every 10 minutes whether the dispatcher noticed or not.

The implication for the dispatcher's day rhythm is structural. Pre-AI cadence: react to events as they hit, intermittent reshuffling, the dispatcher's attention oscillating between the wall board, the phone, and the radio. Post-AI cadence: every 10 minutes a proposed reshuffle lands in a notification tray; the dispatcher reviews in 15-30 seconds; approves or overrides with one-line documented reason. The dispatcher stops reacting and starts reviewing. Pace calmer, output sharper, the "I didn't notice the rookie was free at 10:30" mistakes disappear because the loop noticed for the dispatcher.

The math the ops manager defends to the owner: three to four 10-minute reshuffles per truck per day, each worth $50-$100 of recovered labor and incremental billable revenue. At a 7-truck shop, 7 × 3.5 × $75 = ~$1,840/day of recovered value. Across 250 working days = ~$460K/year of value the loop captures that the morning board build alone cannot. That number sits underneath the 12-18% dispatch yield lift documented across Dispatch Pro, Sera, and FieldEdge 2026 deployments — the lift is real because the cadence runs continuously, not because the model makes smarter individual decisions than a great dispatcher.

Cadence Beats Peak Intelligence

The principle the ops manager carries into every conversation with skeptical dispatchers: cadence beats peak intelligence. A great dispatcher with 20 years of shop knowledge makes better individual routing decisions than Dispatch Pro on the hardest 5% of decisions. Dispatch Pro makes the routine 95% of decisions in milliseconds and never forgets to make them. Across 50+ reshuffle opportunities per day at a 7-truck shop, the model captures every opportunity; the dispatcher captures the 5-7 that require shop knowledge. The combined output is better than either alone — and structurally, the cadence is what creates the lift.

The 2 p.m. Bel Air Scenario, Walked End-to-End

The canonical re-dispatch scenario for ops manager training: 2:07 p.m. Tuesday in July, Bel Air no-cool call lands in CSR booking. The walkthrough from CSR booking to closed call illustrates every decision point in the re-dispatch loop and where the dispatcher's override authority kicks in.

Step 1: CSR Booking and Board Update

2:07 p.m. The CSR books the Bel Air call into ServiceTitan / Sera / Housecall Pro. The booking captures: customer name, address, complaint (no cool, 16-year-old system, second floor not cooling), time-window confirmed 3-5 p.m. The call enters the dispatch AI's pending queue. Within 10 minutes (next re-evaluation tick at 2:10 or 2:20 depending on cadence start time), the AI evaluates the call against current board state.

Step 2: Predicted Revenue and Tech Ranking

2:10 p.m. The AI computes predicted job revenue: 16-year-old system + no-cool complaint + Bel Air zip + July season + premium-call-type indicator. Historical shop data: this profile has closed at $14K average with a fat tail to $24K. Predicted revenue = $14K with $4K standard deviation. The AI then computes expected revenue per assignment for every available tech: Jose (38% MPR × $14K predicted = $5,320 expected, but Jose is currently 35 min south on a $620 service ticket), Marco (28% MPR × $14K = $3,920 expected, but Marco is locked on Marin Park install crew with capacity block from L2 Ch3 Lesson 1 configuration), rookie (14% MPR × $14K = $1,960 expected with high variance). The ranked recommendation: pull Jose from the south-territory service call, route to Bel Air, accept 90-min window slip on the south-territory customer with notification, route rookie to a backup low-stakes call.

Step 3: Rebalance Evaluation

The AI evaluates the rebalance against five constraints simultaneously: (1) install-crew capacity block — Marco unavailable, accepted. (2) Time-window commitments on Jose's south-territory customer — 2-4 p.m. window already running; flexing to 3:30-5:30 with notification. (3) Customer named-request flag on south-territory customer — clean, no named request. (4) Recall flag on south-territory customer — clean, not a recall. (5) Originating-tech routing — clean, fresh service call. All five constraints clear the rebalance. The AI proposal lands on the dispatcher's screen at 2:11 p.m.

Step 4: Dispatcher's 90-Second Review

The dispatcher reads the proposal: pull Jose from south, route to Bel Air, slip south-territory window 90 min, notify customer. The dispatcher's 90-second mental check against the L2 override criteria document: (1) Comp plan dynamics — Jose is one close from his bonus tier; the Bel Air call is exactly the kind of call that triggers his bonus. Approve. (2) Install crew margin — Marco is locked; not in play. Approve. (3) Recall risk by originating tech — not a recall. Approve. (4) Customer request specificity — no named request on either call. Approve. (5) Unseen variables — weather is clear, traffic is normal, no supplier issues, Jose's morning was solid. Approve.

The dispatcher approves the rebalance at 2:12 p.m. Jose gets the on-my-way text to Bel Air with ETA 2:45 p.m. The south-territory customer gets the window-slip notification text at 2:13 p.m. The rookie gets routed to a backup low-stakes call. The board has rebalanced in 6 minutes from call landing to truck reassigned, and the dispatcher's attention spent on the decision was 90 seconds.

Step 5: The Call Closes

Jose arrives at Bel Air at 2:42 p.m. The system is a 16-year-old R-22 unit with a failed compressor and obvious corrosion across the indoor coil. Jose presents the L2-Ch5 three-option proposal: repair (R-22 retrofit, $3,800, doesn't solve the coil corrosion), partial (replace condenser only, $7,400, leaves the coil exposed), replace (full system replacement, three tiers $11K / $14K / $18K). The homeowner picks the mid-tier $14K replacement with Wisetack financing at $189/month. The call closes at $14K. The dispatcher's 90-second decision at 2:12 p.m. has produced $14,000 of revenue, ~$5,320 of gross margin, and triggered Jose's bonus tier for the month.

The math the ops manager defends after the call closes: had the call been routed to the rookie (no rebalance), expected revenue $1,960; the call likely closes at $3,800 repair only at the rookie's typical close pattern. The rebalance produced $14,000 - $3,800 = ~$10,200 of incremental revenue and ~$3,900 of incremental gross margin on a single decision the loop proposed and the dispatcher approved in 90 seconds. Multiply across 3-5 high-value re-dispatch decisions per week at a 7-truck shop and the loop's value is $40K-$80K of monthly margin contribution.

What the Dispatcher Overrides

The re-dispatch loop captures the routine 95% of rebalances; the dispatcher's override authority handles the 5-15% that require shop knowledge. The five permanent override categories from L2 Ch3 Lesson 1 apply at every re-dispatch tick, not just at the morning build. The ops manager's training discipline: every dispatcher reviewing a re-dispatch proposal runs the same 90-second mental check against the five categories.

Comp Plan Dynamics Mid-Day

The PIP and bonus-tier dynamics from L2 do not pause at noon. The 2:07 p.m. Bel Air re-dispatch could legitimately route to Marco if Marco was not on the install crew — but the comp-tier dynamics that favor Jose at this moment of the month (one close from bonus) are mid-day variables. The dispatcher's override authority encodes: "Push the Bel Air to Jose even if AI proposed Marco" or vice versa, based on the comp tier curves the AI doesn't see. The override gets logged with one-line reason ("Jose comp tier — one close from bonus, route Bel Air"); the model captures the override pattern; Friday review walks the comp interactions.

Install Crew Protection During the Day

Install crew protection from L2 Ch3 Lesson 2 (smart routing) applies just as forcefully during re-dispatch. If the AI proposes pulling Marco from the Marin Park install at 2:07 p.m. to handle a service call routing more efficiently, the configured install-crew capacity block (Lesson 1 of this chapter) flags the proposal as a rule violation. The dispatcher reviews against the L2 install-rule-bend criteria (install completion probability today, redeployable crew identification, explicit logged reason). Most days, the answer is "protect the install"; the rare day with a supplier-delayed install, the answer is "redeploy with logged reason."

Recall Risk Mid-Day

Recalls returning to the originating tech is a configured capacity rule from Lesson 1 of this chapter, so the AI's re-dispatch proposal already routes recalls to the originating tech without dispatcher action. The override case: originating tech is unavailable (sick, on a stacked install, on PTO). The dispatcher's override authority handles the unavailable-originating-tech case by routing to the next most-context-rich tech (typically the partner or apprentice from the original install) and notifying the customer of the substitute. Logged with one-line reason ("originating tech [name] unavailable, routing to [substitute] with install context").

Customer Named-Request Mid-Day

Customer named-request rules from Lesson 1 of this chapter handle the routine case automatically. The override case mid-day: the named tech is locked (install crew, recall return on another job), and the customer was promised the named tech at booking. The dispatcher's override authority chooses: offer the customer the next available window where the named tech is free (push the call), or call the customer and explain the substitute (the named tech's apprentice or partner). Logged with one-line reason.

Unseen Variables Throughout the Day

The unseen-variable category from L2 — weather, traffic, mood, supplier dynamics, street closures — produces the most mid-day overrides. At 1:30 p.m. a thunderstorm forecast for 4 p.m. shifts the optimal routing pattern for outdoor diagnostic calls (do them now, push indoor-equipment calls to late afternoon). At 2:30 p.m. an I-40 closure bottlenecks south-territory routes; the AI's travel time estimates degrade. At 11 a.m. a parts supplier delivery delay reshuffles the install crew availability for the afternoon. The dispatcher's override authority captures these moment-to-moment variables; the Friday review surfaces recurring patterns (annual weather windows, recurring traffic patterns, seasonal supplier dynamics) that get added to the override-criteria document.

The Re-Dispatch Review Cadence

The re-dispatch loop's value compounds only with the review cadence layered on top. The ops manager owns the cadence design across daily, weekly, and quarterly intervals.

Daily End-of-Day Review

4:30 p.m. The dispatcher pulls the day's re-dispatch override log — every override the dispatcher made during the day with the one-line documented reason. The log shows 5-12 overrides at a 7-truck shop on a typical day, categorized by the five override categories. The dispatcher and ops manager walk the log in 10 minutes: identify patterns, flag any system misfires that should have been caught by configuration rather than override, note any new categories for Friday review. The end-of-day cadence is short but compounding — it catches configuration regressions within 24 hours rather than 7 days.

Weekly Friday Re-Dispatch Review

Friday 4 p.m. The ops manager runs the 25-minute structured re-dispatch review from L2 Ch3 Lesson 1, extended to include the L3 Ch3 Lesson 1 configuration adjustments. The review walks the week's override patterns by category, surfaces new patterns for the override-criteria document, identifies configuration adjustments (skill tag updates, capacity rule refinements, override category additions), and produces the cross-functional handoffs from L2 Ch3 Lesson 2 (marketing on named-request data, service manager on tech development, installation coordinator on crew composition risk). The Friday review compounds dispatch yield, override rate decline, and cross-functional intelligence across quarters.

Quarterly Configuration Refresh

Quarterly the ops manager runs the configuration refresh from L3 Ch3 Lesson 1: skill-tag re-tagging against trailing-90-day data, revenue history extension, override criteria document review, capacity rule adjustment. The re-dispatch loop's accuracy compounds across quarters as the configuration matches operational reality more tightly. By quarter four of operation, the override rate has dropped from 15% at week 4 to 7-9% as the model learns shop patterns; the re-dispatch loop is making higher-accuracy proposals; the dispatcher's 90-second review increasingly results in approval rather than modification.

When the Loop Breaks and How to Detect It

The re-dispatch loop fails in three named ways the ops manager monitors. First, configuration drift — the skill tags become stale, the revenue history extends without quarterly review, the capacity rules misfire on new install patterns. Detection: agreement rate between AI proposal and dispatcher decision drops over weeks; override rate stays flat or rises. Fix: trigger an off-cycle configuration refresh.

Second, override discipline drift — the dispatcher stops logging override reasons, the Friday review skips weeks, the model doesn't learn. Detection: override rate stays at 15% for months without decline; new dispatcher patterns don't surface; cross-functional handoffs stop delivering data to marketing and service manager. Fix: re-anchor the override-logging discipline; restart Friday review cadence; re-train the dispatcher on the override-criteria document.

Third, vendor configuration regression — a Dispatch Pro / Sera / FieldEdge feature release reverts a configuration setting (the auto-fill toggle reactivates, the clustering weight resets to default, the fast-fill threshold drops). Detection: dispatch yield declines over 2-3 weeks without explanation, override patterns shift toward call-count optimization, install-crew breaks reappear. Fix: review the vendor's release notes monthly, re-verify the configuration screens after every feature release, escalate to vendor support if the regression cannot be reversed.

The ops manager monitors all three failure modes through the weekly Friday review and the quarterly configuration refresh. Shops that detect failure modes within two weeks recover; shops that detect within a quarter watch the lift erode for six weeks before recovery. Detection cadence is the structural discipline that protects the lift over years.

The Loop as the Shop's Rhythm

By month 3 the loop becomes the shop's operational rhythm. Every 10 minutes a proposed reshuffle lands; the dispatcher reviews in 30-90 seconds; the day's board adjusts continuously. Techs receive on-my-way text updates with credible ETAs. Install crews stay together. Recalls return to originating techs. The shop runs at top-quartile RPT because the configuration captured profit-per-truck as the target.

The dispatcher's role evolves from board-mover to loop-overseer. The 5-15% of decisions requiring shop knowledge get full attention; the 95% of routine rebalances are handled by the loop. Lesson 3 of this chapter builds the scorecard that measures the loop's output. The loop is the engine; the scorecard is the dashboard; the configuration is the chassis. Together they produce the 12-18% lift in 60 days and compounding RPT improvements across years.

Key Takeaways

• The re-dispatch loop is continuous board rebalancing every 10 minutes across every open call and every available tech, against the four-variable optimization (predicted revenue × close × travel × capacity). The morning board build owns 0-7% of the lift; the loop owns 93-100% of it.
• Cadence beats peak intelligence. The model makes routine decisions in milliseconds and never forgets to make them; the dispatcher handles the 5-7 daily decisions requiring shop knowledge. 3-4 reshuffles per truck per day × $50-$100 = ~$1,840/day recovered at a 7-truck shop = ~$460K/year.
• The 2 p.m. Bel Air scenario is the canonical re-dispatch case. Call lands at 2:07; AI evaluates predicted revenue ($14K), ranks techs by expected revenue, proposes rebalance (pull Jose from south-territory service call, slip that customer's window, route rookie to backup). Dispatcher 90-second review approves; call closes at $14K. Incremental margin vs. no-rebalance: ~$3,900 on one decision.
• The dispatcher applies the five permanent override categories at every re-dispatch tick: comp plan dynamics, install crew protection, recall risk by originating tech, customer named-request, unseen variables. The 90-second mental check against the L2 override-criteria document is the same discipline at 2 p.m. as at 6:30 a.m.
• Three review cadences compound the loop's value: daily end-of-day (10 min, dispatcher + ops manager walk the override log), weekly Friday review (25 min, structured pattern walk with cross-functional handoffs), quarterly configuration refresh (skill tags, revenue history, capacity rules).
• The loop fails in three named ways: configuration drift (skill tags stale, capacity rules misfire), override discipline drift (logging stops, Friday review skipped), vendor configuration regression (feature release reverts settings). Detection through weekly review and monthly vendor release-note check; recovery within two weeks at disciplined shops.
• Override discipline does not pause at noon. The PIP, bonus tier, install crew, recall, named request, and unseen variable categories all apply mid-day. The dispatcher's 90-second review at 2:12 p.m. is the same discipline as the 6:30 a.m. board walk.
• By month 3 the loop becomes the shop's operational rhythm. Every 10 minutes a proposed reshuffle lands; the dispatcher reviews in 30-90 seconds; the day's board adjusts continuously. Override rate drops 15% → 7-9%; dispatcher role evolves to loop-overseer; ops manager's configuration discipline produces compounding RPT improvements.
• The loop is the engine, the scorecard is the dashboard, the configuration is the chassis. Lesson 1 built the chassis; this lesson runs the engine; Lesson 3 builds the dashboard that proves the lift to the owner.