The CSR floor is the front door of the shop and the highest-leverage hour of every workday lives between 8 a.m. and noon when the booking funnel decides the truck count, the average ticket, and Saturday's revenue. The Friday afternoon CSR huddle is the second-highest-leverage hour. Most service managers walk into it cold — pulling tickets one at a time, asking a CSR who remembers nothing about a call from Tuesday why the customer "called back tomorrow," and watching the booking-percent gap stay stuck quarter after quarter. This lesson is the named workflow that fixes that hour. The Weekly CSR Audit: 10 booked calls and 10 lost calls per CSR pulled every Friday at 2 p.m., AI scores each against the shop's rebuttal library, builds a one-page CSR coaching card per person, and lands on the manager's desk by 2:45 p.m. The 4:30 p.m. CSR huddle stops being a guessing game and starts being a coaching session anchored to specific moments from specific calls — 30 minutes from pull to huddle, three coaching points per CSR, one behavior change per week. The shops running this workflow in 2026 lift booking percent from a 65% baseline to 80-85% in 90 days and recover four to seven calls per CSR per day they would have otherwise lost. The service manager owns the workflow. The CSR floor owns the change. The owner reads the booking-percent trend on the Friday recap and stops asking what happened on Tuesday.

Why the Weekly CSR Audit Is the Named Workflow

The booking-percent gap between a 65% CSR floor and an 85% CSR floor is the single largest hidden revenue lever in most $1M-$15M trades shops. Twenty percentage points on a shop running 350 inbound calls per week at a $480 average booked-call value is roughly $33,600 of weekly captured revenue, or $1.7M annually. Most owners do not see the leak because the calls that did not book do not show up on the same dashboard as the calls that did. The CSR who took the call remembers nothing two days later. The recording sits in CallRail or ServiceTitan unaudited. The pattern that killed the call repeats next Tuesday and the Tuesday after that.

The CSR coaching problem in 2026 is not a script problem. Most shops have a script. Power Selling Pros, CertainPath, Nexstar, and the in-house coaches each ship a perfectly serviceable rebuttal library and a perfectly serviceable booking workflow. The problem is the feedback loop. Without AI, the service manager pulling 200+ recordings per week to surface which CSRs are losing calls on which objections is a 14-hour task. With AI, the same task is 30 minutes and the output is structured, dated, evidence-anchored, and ready to coach. The Weekly CSR Audit is the workflow that closes the feedback loop. It is named so the service manager does not have to invent it every Friday. It is calendared so it happens whether the floor is short-staffed or the dispatch board is on fire. It is owned by the service manager so accountability does not drift.

The output is one page per CSR, the same shape every week, with three coaching moments highlighted and one behavior change agreed at the huddle. The cumulative effect across a six-CSR floor over 12 weeks is the booking-percent number the owner has been chasing for two years. Without the workflow, the CSR floor stays at 65% and the dispatcher stays grumpy about the calls that did not book. With it, the floor moves and the rest of the operation follows.

The Friday 2 p.m. Pull

The pull is the first 12 minutes of the workflow. The service manager opens CallRail Conversation Intelligence, ServiceTitan call recordings, or Avoca post-call summaries — whichever the shop runs — and pulls a structured sample per CSR. The sample is not random. The sample is engineered to surface coaching signal and avoid wasting the manager's time on calls that ran the script cleanly.

Per CSR, the manager pulls 10 booked calls and 10 lost calls from the trailing seven days. The booked sample skews toward higher-value bookings — calls over the shop's $750 expected booked-call value, calls that resulted in dispatch-same-day, and any membership upgrades closed on the phone. The lost sample is engineered: 4 price-shoppers (calls where the customer asked for a quote and the CSR could not book), 3 schedule-window losses (calls where the customer wanted a window the CSR could not offer), 2 after-hours-confusion calls (calls where the customer wanted clarification on dispatch-fee or arrival window and disengaged), and 1 wildcard (the recording the CSR herself flagged as "tough" during the week, or the call the manager noticed on Wednesday).

The Rebuttal Library as the Grading Rubric

AI scores each pulled call against the shop's rebuttal library — the same 8-12 scripted responses every CSR is expected to use on the top five objections. The library is the input the shop already maintains: "just calling for a price," "I'll think about it and call back," "what's your dispatch fee," "I have a guy already," "I want a window on Saturday." Each library entry has the scripted opener, the redirect, and the booking ask. AI's job is not to invent coaching language. AI's job is to match what the CSR actually said against what the library says she was supposed to say, flag the gap, and quote the 8-second exchange that proves it.

The AI prompt is structured. Input: the call transcript, the rebuttal library, the call outcome (booked or lost), the CSR's name, the ticket value if booked. Output, per call: objection encountered, scripted rebuttal expected, CSR rebuttal actual, gap-or-match flag, quoted 8-second exchange, recommended coaching focus. The prompt template is locked. The service manager does not rewrite it weekly. It is one of the named prompts in the shop's L2 prompt library and lives in a shared Notion page or Google Doc with a version number on it.

The 2:15 p.m. AI Score Build

Once the 20 calls per CSR are pulled, the AI score build runs for roughly 18 minutes across the full floor — 6 CSRs × 20 calls × roughly 9 seconds of AI processing per call equals about 18 minutes wall-clock with parallel processing or sequential batches. The manager pours coffee. The AI does the work that used to take a Friday and a Saturday morning combined.

The AI returns five artifacts per CSR. First, an objection heatmap — which of the five top objections this CSR encountered most and which she handled cleanest. Second, a rebuttal-fidelity score: percentage of calls where the CSR used the scripted rebuttal versus drifted. Third, a quote bank — three 8-second exchanges showing the cleanest book and three showing the cleanest loss. Fourth, a booking-percent delta — this CSR's week versus the floor's median and versus her own trailing four weeks. Fifth, a recommended coaching focus — the single objection where the gap between scripted and actual is widest and where the volume justifies the attention.

The AI does not write the coaching plan. The service manager writes the coaching plan. The AI surfaces the evidence. This distinction matters. If AI writes "Marisa needs to handle price-shoppers better," the coaching card is generic and the CSR tunes out. If AI surfaces the quoted exchange where Marisa said "the dispatch fee is $89 and that's the same as everyone" on a call that went silent for six seconds before the customer hung up, the manager has a coachable moment and Marisa has a specific behavior to change.

The One-Page Coaching Card Format

The manager spends 90 seconds per CSR editing the AI output into the shop's one-page coaching card. The format is fixed. Top of page: CSR name, week dates, week's booking percent versus floor median versus her own trailing four-week trend. Second block: three quoted exchanges — two cleanest books, one cleanest loss — with timestamps and ticket value where applicable. Third block: this week's single coaching focus, written as a behavior change ("when the caller asks the dispatch fee on a no-cool call in July, name the fee, then immediately pivot to the same-day window — do not pause"). Fourth block: the script reference the CSR opens before her next shift. Fifth block: the manager's prediction for next week ("if Marisa applies this on three price-shopper calls Monday-Tuesday, booking percent on her row lifts 4-6 points by Friday").

The page is printed and lives in the CSR's binder. The page is also dropped into the shop's shared coaching folder so the CSR can re-read it on her phone Sunday night. The cumulative file across 12 weeks becomes the CSR's growth record and the manager's evidence base for the next comp conversation. The artifact survives the week, not just the huddle.

The 4:30 p.m. CSR Huddle

The huddle is 30 minutes. Six CSRs around the row. Manager has the six coaching cards stacked. Format is locked. Round one: each CSR reads her cleanest booked exchange aloud to the floor — 20 seconds per CSR, two minutes total. The floor hears the language that worked. Round two: each CSR hears her cleanest lost exchange — the manager reads it, not the CSR, to remove defensiveness — and the manager states the coaching focus in one sentence. Two minutes per CSR, 12 minutes total. Round three: the rebuttal-library refresher — manager reads the scripted rebuttal for the most-missed objection across the floor that week, and the floor runs a 90-second role-play in pairs. Six minutes. Round four: the commitments. Each CSR states one behavior she will change Monday-Tuesday and signs the coaching card. Five minutes. Closing: manager states the floor's booking-percent target for next week and the specific row-level expectation per CSR. Two minutes.

The huddle is not a vent. It is not a feedback ramble. It is structured so a tired Friday-afternoon CSR floor can absorb three specific coaching moments per person in 30 minutes and walk out with a specific change to make Monday. The named-workflow discipline is what makes the huddle repeatable across 50 Fridays a year regardless of which manager is running it and which CSRs are on the floor. The script for the huddle is laminated and lives in the manager's binder next to the coaching cards.

What Not to Coach in the Huddle

The Weekly CSR Audit huddle does not coach attitude, tone, or "energy." Those are real issues but they belong in the one-on-one, not the floor huddle. The huddle coaches specific objection handling against the scripted rebuttal library. If a CSR has an attitude problem the manager addresses it in a Wednesday one-on-one with the coaching card in front of her, references the trend, and either documents progress or escalates to the owner. Mixing personnel coaching with audit coaching contaminates the workflow and the floor stops trusting the huddle. The L2 Cardinal Rule applies through the workflow: nothing the AI surfaces about a CSR goes to the CSR without the service manager's eye on it and without the human moment of the huddle to deliver it.

Measuring the Workflow and the 90-Day Target

The workflow is measured against five numbers and the service manager owns all five. First, booking percent floor-wide — target 80-85% by day 90, baseline 65%. Second, booking percent per CSR — manager watches each CSR's row, calls out the laggard, and intervenes when one CSR drags the floor median. Third, rebuttal-library fidelity — percentage of scored calls where the CSR used the scripted rebuttal — target 75-85%, baseline often 30-45% in unaudited shops. Fourth, recovered calls per CSR per day — calls that would have been lost under the previous floor performance but now book — target four to seven per CSR per day. Fifth, the booking-to-show-rate funnel — booked calls that arrive and convert at the diagnostic — target 92%+ show rate (referenced in Lesson 3 of Chapter 2, the named CSR Show Rate Coaching workflow).

The 90-day arc has three milestones. Day 30: workflow runs every Friday without manager prompting; floor booking percent lifts 5-8 points; rebuttal-library fidelity lifts 15-20 points. Day 60: floor at 75-80% booking; per-CSR variance compresses (the spread between the top CSR and the bottom CSR shrinks from 22 points to under 10); coaching cards become routine and the floor expects them. Day 90: floor at 80-85% booking; per-CSR booking-percent target tied to comp tier (the comp redesign that follows from sustained 90-day discipline); the manager spends less time fighting fires and more time coaching cleanly because the workflow surfaces the work.

The Numbers That Prove the ROI

At a six-CSR floor handling 350 inbound calls per week, lifting booking percent from 65% to 82% adds 60 booked calls per week. At a $480 average booked-call value, that is $28,800 of weekly captured revenue, $1.5M annualized. At a 38% gross margin on incremental service revenue, the workflow contributes roughly $570K of annual gross margin against a service manager time cost of roughly 90 minutes per week (78 hours per year) and an AI tool cost of roughly $200-$400 per month for CallRail Conversation Intelligence or ServiceTitan call summaries layered on the existing CRM. The ROI is not subtle. The workflow pays for itself in week one and compounds for as long as the manager runs it.

The workflow also pays for itself in CSR retention. The Power Selling Pros and Service MVP coaching data from 2026 shows CSR turnover at unaudited shops runs 60-90% annually; at audited shops with structured weekly coaching, turnover drops to 25-35%. The retention gap alone is $8K-$15K per saved CSR (recruiting cost, training cost, ramp-time cost, lost-booking cost). Across a six-CSR floor over 12 months, the retention contribution is $30K-$60K of avoided cost. The math is durable.

AI Vendor Stack and Prompt Discipline

The workflow runs on a layered stack the shop almost certainly already pays for. Layer one: the call recording and transcription source — CallRail Conversation Intelligence ($300-$600/month at typical shop call volume), ServiceTitan call summaries (included with the ServiceTitan seat), or Avoca post-call notes (bundled with the Avoca contract at $1K-$3K/month). The shop picks whichever it already runs; the workflow does not require buying a new tool. Layer two: the AI inference layer — ChatGPT Team or Claude Team at $25-$30 per user per month, or the in-software AI that comes with ServiceTitan / HCP / Sera. Layer three: the prompt library — the shop's locked prompt template, version-controlled in Notion or Google Doc, plus the rebuttal library in the same shared folder. Layer four: the coaching-card output format — a Google Doc template or Notion template the manager fills in 90 seconds per CSR.

The prompt template is the single most important artifact and the one most shops fail to lock down. A drifting prompt produces drifting output. The locked prompt names the rebuttal library file by version, names the call-outcome field (booked / lost / abandoned), names the output fields (objection / scripted rebuttal / actual rebuttal / gap-or-match / quoted exchange / coaching focus), names the constraint (quote exactly 8 seconds; do not infer beyond the transcript; do not invent coaching language; surface the evidence and let the manager write the coaching). The prompt is reviewed quarterly by the service manager and the owner. The constraint discipline is what keeps AI from producing generic coaching that the CSR tunes out.

Integration with the Friday Owner Recap

The Weekly CSR Audit feeds the Friday Owner Recap (the lesson immediately preceding this one in Chapter 6). The owner recap shows the floor's booking percent, the floor's rebuttal-library fidelity, the trailing four-week trend, and the names of the top and bottom CSRs by booking percent. The owner does not read the coaching cards. The owner reads the trend and asks one question: "is the workflow running?" The service manager answers yes-with-evidence (six coaching cards in the folder, six signed commitments, the floor's number moved this week) or no-with-reason (one CSR out, one card not signed, plan to recover). The audit becomes part of the shop's operating cadence rather than a side project the manager runs when she has time. That is the difference between an AI workflow that compounds and an AI workflow that fades.

The Manager's Deliverable for This Chapter

By the end of this lesson the service manager builds three artifacts. First, the locked prompt template for the Weekly CSR Audit — versioned, dated, owner-approved, in the shared folder. Second, the one-page coaching-card template — the shop's specific format with the five blocks named above, also in the shared folder. Third, the calendar entry — every Friday 2:00 p.m. through 5:00 p.m., titled "CSR Audit + Huddle," recurring through the next 12 months, with the service manager as owner and the owner copied on the invite as accountability.

The three artifacts together convert the Weekly CSR Audit from a good idea into a named workflow. Within four weeks the floor expects it. Within eight weeks the booking-percent number moves. Within 12 weeks the CSR floor's row-by-row variance compresses and the bottom-performing CSR either lifts to within 5 points of the floor median or the manager has the evidence base to make the comp or seat decision. The next lesson covers the same discipline applied to technician onboarding: the 30/60/90 ride-along scorecard and the AI-built onboarding tracker that gets a new tech to veteran metrics in 90 days rather than 270.

Key Takeaways

• The Weekly CSR Audit is the named workflow. Friday 2:00 p.m. pull, 2:15 p.m. AI score build, 4:30 p.m. CSR huddle. 30 minutes from pull to huddle. Same shape every week, 50 weeks a year. The named-workflow discipline is what makes the huddle repeatable across managers and CSR rotations.
• The sample is 10 booked and 10 lost calls per CSR. Engineered, not random. Lost sample: 4 price-shoppers, 3 schedule-window losses, 2 after-hours-confusion calls, 1 wildcard. Booked sample skews to high-value bookings and same-day dispatches. The sample is designed to surface coaching signal, not summarize the week.
• AI scores against the rebuttal library; the manager writes the coaching plan. The locked prompt template returns objection / scripted rebuttal / actual rebuttal / gap-or-match / quoted 8-second exchange / recommended focus. The manager edits to the one-page coaching card in 90 seconds per CSR. AI surfaces evidence; the human coaches the moment.
• The one-page coaching card has five blocks. CSR name and trend, three quoted exchanges, this-week coaching focus as a behavior change, script reference for next shift, manager's prediction for next week. Printed for the binder, dropped in the shared folder for Sunday-night re-read. The artifact survives the week.
• The 4:30 p.m. huddle is 30 minutes, four rounds. Cleanest book quoted by each CSR, cleanest loss read by the manager with coaching focus stated, rebuttal-library refresher with 90-second role-play, commitments signed by each CSR. The huddle does not coach attitude or tone — those belong in the one-on-one.
• The 90-day arc lifts booking percent from 65% to 80-85%. Floor-wide. Rebuttal-library fidelity lifts from 30-45% to 75-85%. Per-CSR variance compresses from 22-point spread to under 10. At a six-CSR floor, the workflow contributes roughly $570K of annual gross margin against $200-$400/month tool cost and 78 hours of annual manager time.
• The deliverable is three artifacts. Locked prompt template (versioned, owner-approved), one-page coaching-card template (five blocks), and the calendar entry (recurring Fridays 2-5 p.m., owner copied). Without the three artifacts the workflow fades; with them it becomes operating infrastructure and feeds the Friday Owner Recap as evidence the booking-percent trend is being managed, not hoped for.