Leif Johnson Ford of Manor · Performance audit

Manor: Where We Stand, 10/15/26

For Gary Brophy (GM) · Presented by Will Robbins (GSM) · Meeting Fri 10/16/26

39
units delivered Oct 1–14. Same 12 days: Sept 49, Oct 2025 41
48%
of active leads from the last 30 days: no staff-typed touch in 48h+
70
deals already in motion we can still close (about 15 units at our own close rates)
The ask

Three questions, one line each

1
Failing customers on follow-up? Yes. In the middle of the lead, not the first touch.

48% of active leads received in the last 30 days had no staff-typed touch in 48h+ (n=152). Machine answers in 4.5 min. A person: 14.5 min for leads that land 8:30–5, 16 hours for leads that land 5pm–close. verified

2
Reported activity real? Volume yes. Rep credit is inflated.

38% of texts and 63% of emails are the machine, logged under Will's login and signed as the rep. Activity tracks appointments (rho 0.71), not deliveries (rho 0.08). "143 / 381 / 120" does not reproduce. verified

3
Losing leads before contact? Yes, in fixable places.

13 new units ($1.33M) hidden from the site · 26 loaner pages missing $91K in rebates · 105 sales-line numbers missed Oct 1–13 (95 to one line) · Autotrader Wallet quiet since 8/29 (verify). KBB ICO and Cars.com were cancelled on purpose. verified

What we focus on (slide 6)

  • Every morning: the 48h no-person list per pod. This week: the 70 in-motion deals and the 154-customer recovery list.
  • 318 catch-all leads going to pods today. Showroom-visit follow-ups, video promises, one owner for phone-ups. Show rate back from 70% to 80%.
What we're winning

Faster, more appointments, more gross per car

4.5 min
machine first response to an Internet lead, median (was 7.9 the month before). A customer who writes in gets a person in 9.1 min (Oct) vs 19.3 (Sept).
99.7%
of machine-touched Internet leads reached inside 24h (90 days). Only 1% of Manor Internet leads go 24h with no touch; Austin 24%.
+74%
appointments set Oct 1–14 vs Sept same days (68 vs 39). Shown +57% (44 vs 28). VS Insights; Sept read is low confidence.
17.1
shown per 100 leads, last 30 days. Austin 9.1, Buda 9.0. We book 28.3 per 100; they book 19.7 and 14.9.
$3,635
total gross per unit Oct 1–14 vs $3,116 Sept same days and $2,553 Oct 2025 (before chargebacks). F&I $2,924 a unit vs $2,245.
21.9%
mature close, Q3 2026 vs 17.2% Q3 2025. Shown → sold 68.5% this October vs 54.5% last October.
5.5 → 0
rule-breaking rate or negotiation lines per 1,000 machine sends (0 in 391 since 10/10). Texts to an opted-out customer: 1 slipped on 8/15, none since the 8/29 block.
+3.6%
units Jul–Sep vs last year (372 vs 359) on 22% fewer leads. Leads excluding the two cancelled channels: +5.8% Oct 1–13.

Gross is management only. "Reached" counts machine or person. Cross-store numbers are VS dashboard counts, distinct customers.

Booked and shown

We book and show more per lead than anyone in the group

Per 100 leads, last 30 days (all types, bulk imports out)

Manor booked28.3
Austin booked19.7
Buda booked14.9
Manor shown17.1
Austin shown9.1
Buda shown9.0

October vs September, same days

VS Insights, Oct 1–14OctSeptChange
Appointments set6839+74%
Appointments shown4428+57%

Sept is low confidence (restated snapshots). The next lift is the show: 70% of set appointments show (last 30 days), 80% in August.

By pod, per 100 leads

PodBooked (mature 90d)Shown (mature 90d)Booked Oct 1–13Shown Oct 1–13Show rate 7/18–10/15
Oscar Pulido29.220.531.115.676.0%
Matt Cavaness25.017.723.414.377.9%
Eric French22.813.625.313.368.8%
Store (mature 90d)24.716.422.312.270.3% (30d)

By rep, per 100 leads (mature 90d), shown first

RepLeadsBookedShownShow rateRepLeadsBookedShownShow rate
Ajay13636.829.480.0%Shane34126.714.768.3%
Henry14526.220.789.7%Eddie14323.814.074.4%
LaMarcus26726.619.572.3%Pedro17516.011.479.3%
Emely10829.617.675.7%Cesar9717.510.366.7%
Will G20522.015.178.4%Tariq · Orlando34 · 3511.8 · 22.95.9 · 5.781.8% · 16.7%

Mature 90d = leads 6/16–9/13, appointment on the lead (C §4d–4e). Show rate = came ÷ (came + no-show), appointments 7/18–10/15 (C §5b). Ajay's book is repeat and referral buyers. Tariq started 9/1 and Orlando has 35 leads: too small to read.

Year over year

Oct 2025 vs Oct 2026: the lead drop is two cancelled channels. We show and sell more of the rest.

Same daysOct 2025Oct 2026Change
Leads, Oct 1–13 (ICI imports out)283237−16.3%
…excluding KBB ICO + Cars.com (cancelled)224237+5.8%
Units, Oct 1–14 (Axcessa retail + lease)4139−2
Units per 100 leads, Oct 1–1414.316.2+1.9
Appointments set per 100 leads30.433.3+9%
Shown (shown per 100 leads)46 (16.3)47 (19.8)+21% per lead
Appointments cancelled26.7%15.2%−11.5 pts
Shown → sold54.5%68.5%+14 pts
Show rate (shown ÷ shown + no-show)76.4%73.0%−3.4 pts
Front / F&I gross per unit$573 / $1,980$711 / $2,924+$138 / +$944
VS first-contact stamp, Internet, median rough94 min49 min−48%

First-contact stamp is VinSolutions' own field, missing on a third of leads; read the direction only. Gross is management only. October 2026 deals are still closing.

July–September, both years

372 vs 359
units, +3.6%, on 22% fewer leads (1,725 vs 2,203). Excluding the two cancelled channels, leads −6.6%.
  • Units per 100 leads 16.3 → 21.6.
  • Mature close, Q3: 17.2% → 21.9%.
  • Appointments per lead flat: set 33.5 → 31.8, shown 19.3 → 18.6 per 100.

Not wins yet

  • Units −2 the same days; show rate −3.4 pts; Sept mature close 18.7% vs 23.4% Jul–Aug.
What we focus on

One daily discipline: nobody goes 48 hours without a person

9:30am every day: the 48h no-person list, per pod

Every active lead with no staff-typed touch in 48h+, by rep. Pod manager clears it by noon: a call, a real text, or a reason written on the lead.

48%
today (active leads from the last 30 days, n=152)
<25%
by Thu 10/22
<10%
by Wed 11/18

318 catch-all leads go to the pods today

25.5% of the active book sat on the GSM login with no real owner; 98% untouched in 48h+. Redistributing 10/15. From tomorrow they show up on each pod's 48h list. Goal: catch-all under 5% of the book.

DisciplineWhere we areThe ruleOwner
Showroom-visit follow-up28% of visits got no staff follow-up (13 of 47); 47% followed up inside 24hEvery non-buyer visit gets a person touch same day and again next morningRep; desk checks at close
Video promises12 of 12 open promises past due; 3 with no touch sinceSend it same day, or re-date it with the customerRep; pod manager at 9:30
Phone-ups and missed callsOct phone-ups: 40 logged (32 real) → 12 appts → 9 shown → 5 sold / 4 delivered as of 10/15; 105 sales-line numbers missed Oct 1–13; 95 of them dialed one line (Edward Cavaness SLS) that looks dead; 20 Manor customers on the open callback listOne named owner each day for phone-ups and callbacks, worked by noonPod manager on duty (proposed)
In-motion deals + recovery70 deals (18 due today) · 154 customers (46 P1)Logged call and outcome on every row by Sat 10/17Pod managers

Each line reports on the 8:45 text. A note is not a touch; a dial is not a conversation.

Pace and sales

Where we stand: 39 units through 10/14, month-end central ~97

Retail + lease units by month (Axcessa)

Oct 2025118
Jul 2026129
Aug136
Sep107
Oct 1–1439

Same 12 selling days

Oct 1–14, 202639
Sept 1–1449
Oct 1–14, 202541
  • 3.25 a day vs 4.08 Sept, 3.42 Oct 2025. New/used 18 / 21: used is carrying October (21 vs 15 the same days last year).
  • YTD Jan–Sep 978 vs 1,074 last year (−8.9%).
  • Leads are flat: 238 Oct vs 232 Sept, same 11 operating days.

"75 by the weekend": the math

Plan checkpointTargetActualDifference
Sat 10/31812−6
Sat 10/104729−18
Wed 10/14~6139−22
Sat 10/1775~54 expected~−21
  • 75 by Sat needs 36 in 3 selling days = 12 a day. We did 12+ on 11 of ~310 days in the last 12 months. Best day: 16.
  • Month-end: central ~97 (range 90–105). 100 needs 4.1 a day for the 15 days left.
  • 125 needs 5.7 a day, above August (5.2), our best month this year. 125+ happened in 6 of the last 21 months.
  • What moves it this week: the 70 in-motion deals, confirming every open appointment, all 13 reps on Saturday (Sat averages 7.7 deliveries).
Are we failing customers?

The "40%" spot-check: 48 / 37 / 32

Claim: a manager reviewing ~100 leads saw many go 2–3 days without activity.

Supported for people. Weaker for activity of any kind. verified

48%
no staff-typed touch 48h+
37%
no staff-typed touch 72h+
32%
no activity of any kind 48h+ (staff, machine or customer)

Base: active leads received 9/15–10/14, notes synced in the last 24h, n=152, measured to each lead's last sync (conservative). Measured to now: 55 / 37 / 40. Including stale-synced leads (n=257): 62 / 38 / 50. "40%" sits between people (48) and anything (32).

The 84 with no person in 48h+

Pure neglect49
Vehicle sold17
Machine only13
Customer decided / mgr hold / other5

Pure neglect = nothing happened at all in 48h+, or the customer wrote and is waiting. "Vehicle sold" still needs a person to offer another unit.

Who: by pod

PodActive leadsNo person 48h+Waiting on a person nowOct units / goalIn-motion deals
GSM catch-all (Will's login)31897.9%2n/a2
Oscar Pulido pod18573.5%315.5 / 555
Matt Cavaness pod30268.6%313.5 / 7022
Eric French pod40955.9%69 / 7040

"No person 48h+" here = all active leads of any age synced in the last 24h (store 78.2%); older leads drag every pod up. Waiting now = after the 10/15 9:27 CT live reply check.

By rep: no person 48h+ (fresh leads, n in brackets)

LaMarcus (9)89%
Henry (34)71%
Ajay (17)88%
Tariq (19)68%
Cesar (12)83%
Eddie (24)58%
Shane (53)74%
Emely (19)53%
Will G (30)50%
Orlando (7)43%
Pedro (28)32%

Small n for LaMarcus, Orlando, Cesar: directional only. Ajay logs almost nothing in VinSolutions, so his number may be logging, not neglect suspected.

Will's question

Which leads we miss, and how often

Where the customer gets missedHow oftenBase / windowSrc
Active lead goes 48h+ with no person touching it48% (72h+: 37%)152 active leads received 9/15–10/14A §4a
Lead parked on the GSM catch-all login (no real owner)318 of 1,246 (25.5%); 98% untouched 48h+all active leads, 10/15A §6, §8a
Repeat / previous customer: person reaches out within 24h46% (19 of 66 never had a logged person attempt; 16 unexplained)leads 9/15–10/14A §3c
Showroom visit with no staff follow-up28% (13 of 47)visits 9/15–10/14, buyers excludedA §7a
No-show never recovered (never came back or bought)75.5% (77 of 102)missed appts 7/1–9/30C §5c
Internet lead gets a person inside 1 hour43% (8:30–5 arrivals: 78%)252 Internet leads 9/15–10/14A §3a
Customer waiting on a person right now14 (oldest ~13 days; 9 got a machine reply only)active, live-checked 10/15A §5
Customers waiting on a person at 10am, daily average13.8 (4pm: 14.0, mostly the same people)selling days 9/15–10/14A §5
Customer message never answered by a person after 3+ days5.9% (61 of 1,038)message runs 9/15–10/14A §5a
Manor customers on the open callback list20 (of 296 rows)callback file 10/15A §7d
Sales calls missed (store open, rang 16s+, never connected)105 numbers; 95 dialed the "Edward Cavaness (Manor SLS)" line: 134 missed, 9 connected, none reached a desk — check its routingCallRevu 10/1–10/13; of 246 numbers (line-dialed pulled 10/15): sales 105, service/parts 29, robocall 32, unknown 80H-B, D §12
Promised video / commitment past due12 of 12promise files 10/15A §7d
Closed Lost/Bad, customer wrote after, no person replied9 (+13 "did not respond" who did)1,278 Lost/Bad leadsA §7e

Internet leads almost always get a person eventually: 5 of 218 never did, all explained (duplicate, bought, opt-out). The miss is after the first touch and on people-source leads. Phone/walk-in leads start with a person by definition; their number is the first logged call/text/email after the record is made.

Speed

In the day a person answers in 14.5 minutes. Leads after 5pm wait 16 hours.

First person touch, Internet leads 9/15–10/14 (n=213), by when the lead landed

ArrivedShareMedian to a person≤1 hourSame day
8:30am–5pm48%14.5 min78%94%
5pm to close, store open16%16 hours38%41%
Store closed (nights, Sundays)36%27 min after we open66% in first open hour32%
All (the "104 minutes")100%104 min44%63%

Closed-hours leads wait a median 10 hours for us to open; that's the clock, not a rep. Machine first send: 4.5 min median, any hour.

By hour the lead landed: median minutes to a person

8:30–11am20
11am–2pm18
2–5pm14
5–8pm986
8pm–8:30am645

13% of all rep outreach happens after 5pm; nothing after 8. In the day reps run 3–47 min (Pedro 7, Emely 3, LaMarcus 47 on 14 leads): not the driver.

Why the evening gap

  • Before 10/13 our system never texted a rep about a normal new Internet lead. Reps worked off VinSolutions' own alert.
  • The watcher that should nag at 30 min counts the machine's text as a touch, so it never fired on a machine-texted lead.
  • Tariq and Richard have no phone in the roster: any rep text to them fails (27 of 252 leads).

Fix list

  • Text the rep at 15 minutes on any in-hours Internet lead with no person touch (desk manager if the rep is off).
  • Watcher counts only staff-typed touches.
  • An evening owner, 5–9pm. The 8:30am text lists the overnight leads once.
  • Add Tariq's and Richard's phones.
Customer writes in, person repliesOct MTDSept 1–14
Median minutes9.119.3
Answered within 1 hour70.5%57.6%

Oct vs late Sept first person touch (57 vs 239 min) is mostly mix: fewer evening arrivals in Oct. Daytime alone went 19.3 → 13.7 min.

Follow-through

What we say we will do, and don't

28%
of showroom visits had no staff follow-up (13 of 47 judged, 9/15–10/14). 46.8% followed up within 24h.
12 / 12
open video promises and commitments are past due; 3 with no person touch since. Shane 4, team 3, Eddie 2, Matt 1, Cesar 1, Pedro 1.
20
Manor customers on the open missed-call callback list with no touch since (296 rows; 261 callers aren't in the CRM). Sales-line calls missed Oct 1–13: 105 numbers, 95 to one line that may be dead.
9
leads closed Lost/Bad where the customer wrote after the close and no person replied. Plus 13 "Lost: did not respond" who did respond (90 days).
15
Lost/Bad leads where the customer bought from us. Status hygiene, not a chase, but it breaks every lost report.
29
active leads (10.5% of 277) where the customer wrote last and nothing is scheduled. VS tasks are not in our data unavailable, so this is notes + appointments only.

Also: 9 of 160 active leads (5.6%) got exactly one person touch, then nothing for 7+ days. Keyword-tagged promise and financing rows are medium confidence; unanswered and closed-but-wrote rows are high.

Is the activity real?

Real volume. A third of it is the machine wearing the rep's name.

Oct 1–14 texts: 3,383

Human 2,105 · machine 1,27838% AI

Emails: 1,024

Human 375 · machine 64963% AI
  • Machine = 1,019 fully automatic + 259 AI-drafted, person-approved texts.
  • All of it goes out under Will Robbins' VS login: 987 texts, 466 emails, 0 appointments, 0 sales. VS logs only ~76% of what the send log shows. verified
  • The machine speaks as the rep. 1,330 "[Auto] Text by <rep>" notes Oct 1–15; about 1 in 6 texts names the rep. A manager reading the history sees rep effort that was automation.
  • 74% of hand-written call notes (980 of 1,326) are "left message" or no answer. A note is not a conversation.

"143 calls / 381 texts / ~120 bot"

Does not reproduce. verified

  • Searched 60 VS snapshots × 7 login scopes × every 1–11 day window. No match.
  • "~120 bot" fits one day of machine messages (10/14: 124).
  • Closest days: 10/8 had 142 calls but 192 texts; 10/5 had 387 texts but 256 calls (160 of them the machine login).
Yesterday, Wed 10/14Count
Texts: human / machine130 / 97 (43% AI)
VS calls (human logins / CallRevu logger / machine)109 / 16 / 4
Phone system outbound / connected213 / 91

Calls vs the phone system

  • CallRevu Oct 1–14: 2,581 outbound, 997 connected (38.6%), 676 voicemails.
  • 70% filed "Unidentifiable" → per-rep check possible for 2 of 13 sellers (Will G, Henry). 6 reps have no CallRevu record.
  • 10/13: at least 22 logged call entries have no phone-system counterpart 2 days tested.
Activity vs results

Touches predict appointments, not deliveries

12 reps, Oct 1–14, per working dayAppts setShownDMS units
Logged touches (rank correlation, rho)0.710.640.08
Same, without the 3 near-zero loggers (n=9)0.370.22−0.23
Two-way conversations0.490.510.11

rho runs −1 to 1; 0 = no link. Appointments and activity come from the same CRM logging, so that link is partly mechanical. Units come from the DMS, the independent outcome. n=12; correlation is not cause.

Same leads, very different yield

RepLogged touchesDMS unitsTouches per unit
Ajay180.1
Tariq182536
Henry213453
Shane6924173
Pedro (268 calls, 364 texts)7271727

Two-way conversations per working day

Shane3.5
Pedro2.2
Henry2.0
Emely1.8
Cesar1.7
LaMarcus1.5
Eddie1.4
Tariq1.3
Will G1.0
Ajay0.17
Orlando0.1
Javier0

Two-way = customer sent a genuine reply and a person texted within 48h. Store: ~14 a day (lower bound). Ranks 2–9 are one band: Emely's and Will G's texts are the most under-captured. Ajay sells without logging; Orlando and Javier barely log or sell.

Funnel

Lead to delivered: October cohort vs the mature 90 days

Stage (real leads, all sources)Oct 1–13 (2–14 days old)Sept 1–13Mature 90d (6/16–9/13)
Leads2382321,726
Reached within 1h (machine or person)77.7%66.8%78.0%
Customer replied (floor)46.0%39.9%36.0%
Asked for a day/time (proxy for "appointment requested")15.2%12.1%10.2%
Appointment set22.3%25.9%24.7%
Shown12.2%16.4%16.4%
Shown ÷ set54.7%63.3%66.3%
Shown → sold69.0%71.1%70.7%
Sold (VS flag, Axcessa tie or sold feed)13.4%20.3%22.2%
Delivered (Axcessa, tied by name)8.8%15.1%16.4%
Internet leads only: sold11.7%7.7%12.1%
  • October's cohort is young: 87% of sales land within 7 days of the lead, so read Oct as direction, not final.
  • Same-age comparison (slide 15): Oct delivered 8.8% vs Sept 13.4% vs Aug 16.2%.
  • The leak that is already visible: shown ÷ set down to 55%.
  • unavailable Write-up, approval, manager T.O. and "confirmed" are not recorded in anything exported. Stages between shown and sold are proxies.
  • 19% of deliveries can't be tied to a lead by name, so lead-based close runs low.
The 8% → 20% claim

Our real close is ~21–22%. Nothing steps from 8 to 20.

Where "~8%" shows up (wrong base or young leads)

Raw VS lead count incl. the July Ford ICI mailing import9.7% (July alone 5.4%)
Internet leads only / marketplaces only12.1% / 5.5%
Sept Internet leads, VS Sold flag7.1%
Oct 1–13 delivered (leads 2–14 days old)8.8%
Oct phone-ups12.5% sold (5 of 40 logged; 32 real)

Where "~20%" comes from

Mature 90 days (leads 6/16–9/13)22.2%
/sources page, 90 days20.8%
Sept units ÷ Sept leads19.6%

Possible origin suspected: the 10/13 /sources page shows MTD close 12.2% and "−8.1 pts" vs the same days last month. If the 8 and 20 came from there, it reads down, not up.

What actually changed

+4.7 pts
Q3 2025 17.2% → Q3 2026 21.9%. +2.9 from fewer low-closing Internet leads in the mix, +1.8 from better close inside channels.
  • Units did not rise with it: Apr–Sep 100, 99, 99, 129, 136, 107.
  • Monthly close: 2025 13.2–19.1%; Jan–Sep 2026 17.0–23.8%.
  • Last 45 days are down: Jul–Aug 23.4% → Sept 18.7%; October same-age delivered 8.8% vs Sept 13.4%.
  • Phone-ups: 32 real of 40 logged, 5 sold / 4 delivered (12.5% / 10%) as of 10/15; Sept same age 19.4%; referral/repeat 13.3% vs 32.1%.
Closing

Close by source and by rep (mature 90 days)

Lead → sold, by source

Referral / repeat (214)52.8%
Walk-in (211)37.4%
Phone-up (198)29.8%
Credit-app tools (222)29.7%
Dealer site / OEM (186)13.4%
AutoTrader (118)12.7%
Facebook / chat (41)9.8%
VinCue trade tools (63)9.5%
CarGurus (151)6.6%
TrueCar (250)2.0%
Carfax / KBB (56)0.0%

People-source leads are often created during the visit, so their % is not comparable to a web form. TrueCar: 250 leads, 5 sold.

By rep: actual vs expected for their lead mix

RepLeadsCloseInternet closeActual ÷ expected
Ajay13661.8%23.5%1.72x
Eddie14323.1%9.9%1.06x
Orlando3520.0%7.7%1.03x
Will G20517.1%12.3%0.96x
LaMarcus26722.1%12.5%0.94x
Henry14523.4%18.7%0.91x
Shane34119.4%11.0%0.90x
Emely10818.5%14.3%0.90x
Pedro17514.3%8.1%0.81x
Cesar9714.4%12.3%0.78x
Tariq (new 9/1)345.9%3.6%0.35x

Pods: Oscar 34.8% (1.33x), Matt 21.7% (0.93x), Eric 17.6% (0.90x). Ajay's number reflects repeat and referral buyers landing on him; not a skill gap for the others. Tariq, Orlando, Cesar: too early to judge.

Appointments

Show rate slid from 81% to 70%. No-shows are mostly let go.

Show rate (came ÷ came + no-show)

H1 202681.5%
Sept 202579.7%
Oct 1–15, 202576.4%
Jul 202677.4%
Aug79.8%
Sept70.1%
Oct 1–1573.0%
Last 30 days70.3%

Once a customer shows, ~69% of those leads buy. Internet appointments show 69.5% vs phone-up 82.6%.

By pod and rep (7/18–10/15)

  • Eric pod 68.8% show, 25.3% cancelled · Matt pod 77.9% · Oscar pod 76.0%
  • Henry 89.7% (best; use as the example)
  • Shane 68.3% with 32.3% cancelled, on the biggest book (99)
  • Orlando 16.7%: 2 came, 10 no-shows, of 18
75.5%
of no-shows never came back or bought: 77 of 102 missed Jul–Sep (38 still open leads, 34 marked Lost). Only 19 (18.6%) recovered. Oct 2025: 14.6%.

Confirmations

  • Confirm texts reached 13 of 64 non-cancelled Oct 1–14 appointments.
  • VS "Confirmed" status used once. Same-day no-show chase first ran 10/14.
  • 35 of 110 Sept appointments never got a status update.
  • Sample too small to prove confirmations raise show rate (6 of 10 with vs 35 of 49 without). not proven
Lost deals

Why we lose them, and why we often can't tell

VS lost code, mature 90 days (833 lost of 1,726 leads)

Bought elsewhere29.9%
Out of market17.9%
No response17.5%
System close, no reason14.8%
Credit7.7%
No agreement (price/payment/trade)6.5%
Do not contact4.8%
Bought at Austin / Buda1.0%

What the notes say (lost leads 8/15–10/13, 331 with notes)

41.4%
have no reason written anywhere

Where there is one: bought elsewhere 28.1% · price 14.5% · trade 14.2% · credit 12.4% · payment 8.8% · availability 6.9% (a lead can show several). keyword parse

  • VS has no coded price, payment or trade reason beyond "No agreement reached". We cannot say why deals die.
  • 10/12 weekly review: 2 leads marked lost inside 24h; 8 of 9 "Bad" leads had replied or visited.
  • 9/24 audit: 15 of 236 Lost/Bad were live customers.
  • Group stores: 61 leads coded "bought from another LJ store" since Jan 2025 (22 in 2026). A floor: Austin/Buda sales aren't in Manor's Axcessa.
Money on the table

70 deals in motion · 154 customers to recover

18 P1 in-motion deals (due today) · full list in C2 CSV

#CustomerRepStage signalBlocker on file
1Christopher K.Shaneapproval + credit apptrade / negative equity
2Mario B.Eddiecredit app; asking numbersnot documented
3Yomar A.Will (catch-all)credit appcredit / co-signer
4Adrian C.Tariqshown 10/8; asking numberswants different unit
5Aaron K.LaMarcusnumbers promised, not sentwants different unit
6Tyson A.LaMarcusnumbers sentwants different unit
7Christian R.Eddienumbers sentprice vs competitor
8Thomas L.Shanenumbers sentpayment / trade
9Andy W.LaMarcusnumbers senttrade / timing
10Larry S.Henrynumbers sentpayment too high
11Vrutant P.Shanenumbers sentpayment too high
12Taylor T.Shanenumbers senttrade / negative equity
13Nico M.LaMarcusshown 9/19wants different unit
14Charles C.Pedroshown 10/9wants different unit
15Frederick O.LaMarcusshown 10/12trade / negative equity
16Wend B.Tariqshown 10/10not documented
17Kirk F.Shaneshown 10/14not documented
18Johnathan C.Henryshown 10/14not documented

In-motion deals: 70

Approval / credit app 18 · numbers or live negotiation 34 · shown not sold 17 · quiet credit app 1.
P1 18 (10/15) · P2 20 (10/16) · P3 32 (10/17).
Eric pod 40 · Matt 22 · Oscar 5 · GSM house 2 · mgr 1. Shane 18, Pedro 12, Will G 8.

~15 units

at the store's own close rates (range 5.5–17.3). Gross math on slide 25.

Recovery list: 154 customers

P1 46 · P2 36 · P3 72. 64 more screened out (customer decided, bought, thanks-only).
P1 problems: financing unresolved 22 · promised callback 15 · unanswered message 8 · going cold 6 · closed but wrote 2.
By pod P1/P2/P3: Eric 19/11/17 · Matt 14/16/17 · Oscar 9/7/11 · catch-all 1/2/27.

Plus 48 open no-shows never recovered (C2 secondary). Lists overlap. Rows marked verify-in-VS were synced 24h+ ago: open the customer first. Stage is inferred: write-ups aren't exported.

Losing leads before contact

Total leads recovered in October. One feed to verify.

Source (VS rows, ICI import out)JunJulAugSepOct 1–13Per day Jun–Aug → OctJan–Sep 2025 → 2026
TrueCar631048054272.68 → 2.08703 → 749
Phone-up (logged)66606082352.02 → 2.69–
Walk-in / drive-by58608772262.23 → 2.00–
CarGurus34565338181.55 → 1.38395 → 348
Dealer website forms / chat35384825201.32 → 1.54305 → 282
Autotrader (retail)8272423150.64 → 1.15315 → 188
FordDirect / OEM20192116110.65 → 0.85–
VinCue trade tools20142624120.65 → 0.92–
Carfax trade-in2023102380.58 → 0.62372 → 214
Facebook40111660.16 → 0.46–
Impel chat4145410.25 → 0.08–
Autotrader Wallet (last lead 8/29) verify152420000.64 → 0–
Cars.com (cancelled on purpose; last lead 5/21)000000228 → 49
KBB Instant Cash Offer (cancelled on purpose; last lead 2/11)000000984 → 113

Third-party + site leads per day: Jun–Aug 9.1 · Sep 7.4 (−19%) · Oct 1–13 9.1. Dealer website fell 37% in Sept, back in Oct on a small base.

Channel changes

  • KBB ICO and Cars.com: cancelled on purpose. They explain most of the year-over-year drop in lead count (Jan–Sep 984 → 113 and 228 → 49).
  • Autotrader Wallet: last lead 8/29, ~17 leads/mo and ~2 sold/mo while live. If it wasn't cancelled, it's a broken feed. One call to the rep answers it.

CarGurus

  • Leads Jul 46 → Aug 37 → Sep 25 → Oct 1–14 12 (AP slide 10, exact source name)
  • Closes 6.6% (151 leads, mature 90); person ≤1h 42%
  • 14 used units on the lot rated High Priced / Overpriced (10/10)
  • 3 reviews; old "riata-ford" address and service phone (AP slides 7, 9)
Website and listings

What a shopper sees that's wrong

13
new units hidden from the site: VinCue "On Hold". $1,325,990 advertised; 9 are Platinum Super Duty, 7–147 days old. Hold reason unknown. Owner: Will (VinCue).
26
In-Service FCTP loaner pages show no Ford rebate: ~$91,000. Was 31 / $108,750 on 10/14 (AP slide 5); 5 fixed overnight. Owner: Marri + website vendor; Ford rep + counsel confirm.
36
new units show "Leif's Price" above MSRP: $649 tint + $225 doc are built into all 234 live new prices. Gary decides, counsel first (AP item 3).
287
of 287 live VDPs carry an old Riata phone in the page text (512-281-3673 new, 512-601-6381 used). Homepage code lists the old phone and Riata LinkedIn/X. Visible footer is right.
0 / 6
AI search questions name Manor correctly (AP slide 10). Site sits behind a bot challenge: crawlers got 403. Googlebot untested; needs Search Console.
77%
of new units show >$500 less off sticker than the market. Our median $2,126 vs $4,775; 5th of 6 stores (AP slides 3–4). Used: 33 of 53 above market average, 9 over 105%.
20
live pages with under 10 photos; 14 show one placeholder; 2 units 30 days old with none. Window-sticker link has an unfilled code on 274 pages untested.
3
sold units still live (35 of 38 recent sales came off right). Also: 5 new arrivals' discounts not posted ($14,501); 5 units VinCue $1,000 above site.

Impact on leads or sales is not estimable without GA4 / Search Console. Used prices match VinCue to the dollar on all 53.

Loaners

FCTP: the file says 58, not 56

33 + 25 = 58
33 In-Service FCTP + 25 Courtesy Transportation Vehicle, all Active in VinCue's new list verified
  • The action plan's "33" and "31 affected" are the In-Service group only. Couldn't find where "~56" came from.
  • VinCue's own demo flag marks 34 units but overlaps only 29 of the 58. Don't use it to find loaners.
  • $2,401,270 advertised · $2,742,275 MSRP. The "$2.6M" quoted matches neither.
  • Rebate on the page: CTV 24 of 25 show it. In-Service: 26 show none, 5 fixed since 10/14, 2 are 2027s with no rebate.

Age of the 58 (median 122 days)

Under 30 days0
30–598
60–899
90–1199
120+32
  • 32 of the store's 47 new units at 120+ days are loaners.
  • Without loaners, new inventory is young: median 17 days, 15 at 120+.
  • Carrying cost not in scope: no impact dollars claimed.
  • Asks: Ford rep + counsel confirm rebate display on In-Service units; feature FCTP in New Specials (4 of 247 featured today, AP slide 9).
Our AI

Fast first responder. No proof it sells more. Some errors still live.

Internet-type leads, 90 days

Machine touched (685)Not touched (334)
Reached within 1h83.9%84.7%
Customer replied47.9%37.9%
Appointment set14.3%20.4%
Sold8.8%16.5%
Sold after day 1 (removes same-day deals)4.2%3.9%

The raw gap is selection: the machine is kept off leads a rep is already working. After day 1: no lift, no harm shown.

  • ~10,300 machine messages in 90 days (5,436 texts, 4,897 emails). First send median 4.6 min.
  • Only 7 of 371 appointments were machine-first with no person before (5 showed, 5 sold). Label it "assisted", not "created".
  • Does well: first-touch 23.6% reply with no person involved; auto-responder 36.7% (0.8% opt-out); rate/negotiation slips 5.5 per 1,000 → 0 in 391 sends since 10/10.

Errors, counted

ErrorCountStatus
Told a customer a sold unit is "still here"14 sends / 8 customers12 after the 9/28 guard. Gate at every send added 10/15; earlier customers being corrected.
Trade pitch after rep note said "no trade" (the meeting example)2 of 1,331Real, rare. Notes check added 10/15.
Cadence sent within 24h of a rep touch326 of 1,224 (26.6%)After 10/14 gate: 0 of 18
…inside a live two-way thread48 (3.9%)Gated 10/14 5:36pm
Held drafts expired unanswered, Oct 1–141810 customer-reply holds approved by a person
Impel and ours on the same customer4 confirmedOne texted STOP; Impel kept AI on 5 days

Opt-outs

  • STOP after a text: 8.6 per 1,000 machine texts vs 2.9 per 1,000 human texts (18 STOPs in October; small n).
  • Opt-out within 72h: first-touch 4.4%, cadence 3.4%, auto-responder 0.8%. 3 true complaints since 7/17.
Impel vs ours

Impel ran ~90 days at Manor (5/22–8/21). It never showed a sale.

Impel, JulyImpel report 9/13–10/13 (leads mostly 10/7–10/9)Ours, 9/13–10/13
Leads542 worked (400 new)20 new557 created, 350 touched
Messages4,134 (69 bounced)not shown1,967 texts + 1,510 emails
Engaged113 of 400 (28%)~5first-touch reply 18.0% (72h, no person)
Appointments3 AI-set4 (2 AI-only), 1 shown2 strict machine-first
Salesblank on its report1 (our managers also worked it)34 of 96 delivered units touched (not caused)
Speedn/a5 min4.6 min median
Cost~$3,000 / store / month per Will verify invoice; no invoice on fileAPI spend not recorded unavailable

At Austin and Buda Impel is live on 96–99% of Internet leads (slide 29). Not a controlled test: sold per Internet lead was 12.4% with Impel alone (5/26–7/27) vs 12.3–12.8% without. Full history: impel-deck.html. When Impel went off 8/21, leads reached within an hour fell 81% → 67%, then recovered to 73%. Impel risks in its own records: same copy by text and email together, a price $1,232 over the feed, an answer allowing a deposit hold, an 11:30pm text, STOP not honored for 5 days, 2 of 10 users ever active.

Keep

  • First-touch
  • Auto-responder with the human-thread yield
  • Hard gates: opt-out, recent sale, hold/deposit, rate guards
  • Appointment confirms, sold check-ins

Fix

  • Sold-unit block: done 10/15, verify 3 mornings
  • Verify the live-thread gate 3 more mornings
  • "No trade" notes check: done 10/15
  • Expired holds → 8:45 call list
  • Cap repeat pitches
  • Log cost, setter, machine vs human

Don't expand

  • Any "AI closes deals" claim
  • ICI / owner / area waves (5% reply)
  • Re-enabling Impel or a second AI on the same leads
  • Auto-sending money or negotiation replies
RESTRICTED – management only Store gross · no per-deal, cost or pay data

Gross per car is up since July

MonthUnits (new / used)Total gross / unit (GL, net)Front / unitF&I / unitNegative-front dealsChargebacks
Oct 2025118 (69 / 49)$2,883$613$1,96246%n/a
Jul 2026129 (84 / 45)$2,766$298$1,92848%$34.1K
Aug136 (76 / 60)$3,369$511$2,30048%$27.1K
Sep107 (59 / 48)$3,780*$915$2,40035%$41.4K
Oct 1–1439 (18 / 21)$3,253**$711$2,924n/a$15.0K

Front and F&I = deal log (runs 19–28% under GL). *Sept GL transitional. **38 deals, early month. GL new / used PVR: Jul $2,390 / $3,500 · Aug $3,267 / $3,512 · Sep $3,832 / $3,711 · Oct $4,398 / $2,199.

F&I per manager (per credited deal)

JulAugSepOct 1–14GAP / deal (Oct)
Brian Calderon72 · $2,32463 · $2,44755 · $2,52322 · $2,827$146
Jeff Ledesma43 · $2,00563 · $2,21743 · $2,56916 · $3,216$337
  • YTD vs 2025: front +$465/unit ($292 → $757), F&I −$491/unit ($2,482 → $1,991). Total flat. F&I rebuilding since July.
  • Front given up on negative-front deals: $96K–$127K a month Jul–Sep. Often deliberate (aged units, program money): a leak to check, not a loss.
  • Chargebacks: $24.2K–$54.7K a month Jan–Sep; YTD $326.4K. Cause not in the data.
  • Cash not finished: 18 unbooked deals (1–12 days; 3 over 7) · 7 contracts in transit (5–23 days; 3 over 10). Counts only.

Recoverable gross (pipeline, not lost revenue)

(18 × 24.6% + 51 × 20.8% + 1 × 7.9%) = 15.1 units × $3,274 GL/unit

≈ $49.5K

Range $18.1K (all at CarGurus 7.9%) to $56.6K (all at phone-up 24.7%).

unavailable ad cost per sale · AI cost per sale · preventable-loss dollars · written-not-delivered · F&I penetration and chargebacks by manager.

People and coverage

Fix productivity before adding 3–4 reps

DayReps on (of 13)DeskF&IAvg units delivered (Jan–Sep 2026)
Mon1332
3.0
Tue1021
4.1
Wed92*1
3.3
Thu922
3.5
Fri1332
3.5
Sat1332
7.7

*Wed: no bilingual desk manager (Matt and Eddie off). Weekdays deliver 3–4 a day whether 9 or 13 reps are on: the limit looks like traffic and appointments, not hands correlation. Saturday = 31% of 2026 units.

Units per 100 leads (Jul–Sep): Ajay 64 · Henry 24 · Eddie 21 · LaMarcus 18.5 · Will G 18.5 · Shane 16.5 · Emely 13.9 · Pedro 12.9 · Cesar 12. Same leads, 5x spread.

+29.6
units a month if the 6 reps under 12/mo reach 12 (Eddie, Will G, Pedro, Emely, Cesar, Orlando)
~3
units in a new hire's first full month (our 7 hires/returns in 2026). 3–4 new reps = 12–20 units in 2 months, on a thinner lead pool (~42 → ~34 leads each)
  • Add now: 1 BDC / appointment seat. No BDC since May; 318 catch-all leads; 105 sales-line numbers missed in 13 days (95 to one possibly dead line).
  • Plan: a second truck closer. Ajay did 29 of 52 new Super Duty deals Apr–Sep and logs no calls (key-person risk).
  • Gate Wed 11/18: fair lead share for all; 8 reps at a 10+/mo pace; 65%+ of set appointments show up; a Saturday up-log shows customers waiting. Then add 1–2, not 3–4.
  • 77% of scheduled rep-days show VS activity (Oct 5–14); Javier 0 of 7. 4 floor exits 9/29–10/3; 2 starters never showed.
Certifications

"About 11 incomplete": we can't verify it

Unknown
No file on the Mini, the MacBook or the Ford feeds lists certifications by employee. Do not quote 11 until the report below arrives. unavailable

What we do have

  • Will's own Ford University: 9 of 29 required modules complete as of 10/6 (list grew from 20 to 29 as modules were done).
  • DDT utilization 93% in Sept (55 of 59 new units); 1 Oct deal "not utilized" in the 10/13 pull. Delivery compliance, not certification.
  • Sept new-unit survey response 19% (11 of 59).

What we need

  • Per-employee certification / required-training report from Ford University / STARS: each person, each course, complete / due / overdue, due date. Sales, F&I, service, managers.
  • Who can pull it: the Dealer Training Coordinator or the GM's Ford Dealer Connection login. Will's login shows only his own transcript.
  • Fallback: ask the Ford area training manager to email the certification compliance report.
  • Then: count gaps by person and role, tie each to the Mon–Thu program, re-run weekly.
Training plan (proposed, weekly)

Weekly at 1pm: Monday teaches one topic, Tuesday–Thursday drills it

Week (Monday)Why (our data)BaselineDay 14 (11/2)Day 30 (11/18)First in the room
1 · 10/19 · Phone28 of 89 opportunity calls set an appointment; Matt 38%, Oscar 6%; 105 sales-line numbers missed Oct 1–13 (95 to one line)31% appt set on opportunity calls38%45%Shane, Eddie, Cesar, Tariq, Orlando, Oscar, Eric
2 · 10/26 · Fact FindingPedro 189 touches per sale; Shane 45% of 114 appointments came in59% of set appointments come in62%65%Pedro, Shane, Cesar, Orlando, Eddie
3 · 11/2 · Trial ClosingSept visit-to-sale: LaMarcus 28%, Will G 43%, Pedro 45% vs Eddie 75%41% of visits reach a desk request (51 of 123)track+5 ptsLaMarcus, Will G, Pedro, Cesar
4 · 11/9 · Objections & Deal Mgmt52% negative new-car front on both desks; GAP $146 vs $337 per deal35% negative-front deals (Sept)35%30%Henry, Emely, Will G, LaMarcus, desks, F&I
Every weekShow rate · same-day appointment status70% last 30d · 68% updated (Sept)76% · 90%80% · 95%Everyone
ProgramAttendance (live or make-up) · calls reviewed per rep per weeknot measured85% · 190% · 2Tracker sheet

How

  • Mon 40 min, everyone in (nobody is off Monday): number · one skill · role-play in threes · commitments
  • Tue–Thu 15 min: the line of the day, role-play in pairs, one commitment
  • Off on a drill day: 10-min role-play with your manager within 48h. Recorded.

Leaders

  • Mon Will · Tue Matt · Wed Oscar · Thu Eric (each on a day he's scheduled)
  • Peer examples: Henry (91% show), Tariq (8 of 8 shown), Jeff (GAP)
  • Week 5 (11/16) the rotation starts over with Phone

Dates

  • Approve Fri 10/16 · baseline frozen + Session 1 Mon 10/19, 1pm
  • Day-14 review Mon 11/2 · Day-30 review Wed 11/18 (with the staffing gate)
  • Full plan: training-deck.html. Targets are our own best recent numbers
Manor vs the group

Manor books and shows more per lead. Austin's bot hides slow people.

Internet leads unless notedManorAustinBuda
Leads, last 30d (bulk imports out, all types)5551,096791
Median min to first staff outbound*128 person-only (5.6 if our machine's texts count)164.241.3
…leads created 8a–8p / after hours14.5 (8:30–5) / 16 h (5–9pm), person-only61.6 / 672.615.1 / 617.8
Staff touch within 1h42% person-only (77% blended)34%53%
No staff touch inside 24h (or ever)1%24%6%
Customer replied after first staff outbound55%29%37%
Appointments set per 100 leads (30d, all types)28.319.714.9
Shown per 100 leads (30d)17.19.19.0
VS sold, last 30d (per 100 leads)99 (17.8)216 (19.7)106 (13.4)
Mature Internet close (leads Jun 1–Jul 15, 91d+)12.4%12.4%6.9%
All non-bulk leads, same cohort16.1%20.5%13.5%

*Not like-for-like: Manor's first outbound blends our machine and staff (machine texts are logged under user names); Austin and Buda exclude Impel. Manor person-only: 128 min median / 42% within 1h on web leads Sep 15–Oct 12 (cross-store report), 104 min overall and 14.5 min for leads landing 8:30–5 on the slide-10 cohort. Still ahead of Austin (164 / 34%), behind Buda (45 / 52%). Comm-history census of Internet leads Sep 1–Oct 12 (n = 395 / 743 / 738). Sold = VS Sold flag; no Austin/Buda Axcessa in our data.

Impel at the sister stores census

  • First touch on 99.3% of Austin and 96.1% of Buda Internet leads, median 3.7 / 3.8 min, around the clock (after hours 4.6 / 4.5 min). Manor: Impel on 7.1% until the 10/13 pause; our machine 4.6 min median.
  • Austin's VS tile reads 99.2% "answered in 15 min" because Impel posts "Clock stopped". Staff median is 164 min; 65 leads (8.7%) never got a person.
  • Buda's Impel never posts the clock-stop note, so its tile shows 49% "Missed" in October although Impel reached 96% inside 4 minutes.

What it does and doesn't prove

  • Closing can't show Impel helps or hurts: Austin 12.4% = Manor 12.4% mature Internet close; Buda 6.9%. Manor ran Impel until 8/21, so this is not a bot vs no-bot test.
  • A bot first-touch makes the VS response tile meaningless as a people measure. Judge staff speed on comm history.
  • unavailable Impel dashboard (separate login); customer replies to Impel aren't logged in VS.
Summary

Five biggest problems · five biggest opportunities

Problems

  1. Leads go quiet after the first touch. 48% of active 30-day leads: no person in 48h+. 318 leads (25.5% of the book) sat on the GSM login, 98% untouched (going to pods 10/15).
  2. Promises and follow-through slip. 28% of showroom visits got no follow-up; 12 of 12 video promises late; 105 sales-line numbers missed in 13 days (95 to one possibly dead line); phone-ups have no single owner (40 logged, 32 real, 5 sold as of 10/15).
  3. Appointments leak at the show. Show rate 81.5% (H1) → 70%. 75.5% of no-shows never recovered. Confirm texts reached 13 of 64.
  4. Activity reports overstate rep effort. 38% of texts and 63% of emails are the machine signed as the rep; 70% of calls can't be tied to a person; activity tracks appointments, not units.
  5. The website leaks shoppers. 13 units ($1.33M) hidden; 26 loaner pages missing $91K rebates; 77% of new units behind market on advertised discount; old Riata phone on all 287 VDPs.

Opportunities

  1. 70 deals already in motion ≈ 15 units at our own close rates. 18 due today; Eric's pod holds 40.
  2. Manor already books and shows ~2x the sister stores per lead (17.1 shown per 100 vs 9.1 / 9.0). One BDC / appointment seat to own the 318 catch-all leads, callbacks, the 154-customer recovery list and the same-day no-show chase.
  3. Show rate back to August's 80%. On Sept's 134 came-or-missed appointments that's +13 shown, ≈ +9 sold at Sept's 67% shown-to-sold (math, not a forecast).
  4. Lift the 6 reps under 12/mo to 12 = +29.6 units a month: more than 3–4 new hires add in their first two months (12–20).
  5. Cheap website fixes: post the $91K of loaner rebates, unhide stale holds, fix phones and page code, confirm the Autotrader Wallet feed (~17 leads/mo when live).
Action plan (full version: gary-G-action-plan.md)

Who does what, by when, and how we'll know

24–48h (by Sat 10/17)

  • Work 18 P1 / 20 P2 / 32 P3 in-motion deals · pod managers · every row has a logged call + outcome by Sat
  • Clear the 14 waiting + 46 P1 recovery customers · reps; managers check · 0 waiting over 1 store hour at 10am
  • Confirm every Sat appointment; same-day no-show call · pod managers · Sat show ≥ 80%
  • Call 20 Manor callers + 12 late promises · named owners · 0 overdue
  • Decide tint + FCTP rebate display · Gary, counsel · decided 10/16
  • Vendor calls: AT Wallet (verify), CarGurus, TrueCar · Will · cause known for each

7 days (by Thu 10/22)

  • 9:30am 48h no-person list per pod · pod managers · 48% → under 25%
  • 318 catch-all leads redistributed to pods (started 10/15) · Will · catch-all under 5%
  • One named owner a day for phone-ups and callbacks · pod manager on duty · callbacks worked by noon
  • Website: FCTP rebates, On Hold review, page phones + code, sold-still-live · Marri + vendor; Will · 0 left on the defect list
  • Training starts Mon 10/19, weekly 1pm · Will, pod managers · attendance ≥ 85%
  • AI: sold-unit block, gate check, no-trade check · Will · 0 wrong-unit sends
  • CallRevu map for 6 reps; VS login for Richard; Ajay logs · Will · 13 of 13 checkable

30 days (by Wed 11/18)

  • Day-14 (11/2) and day-30 (11/18) reviews + staffing gate · Will, Gary
  • Hire the BDC seat · Gary · 0 Manor callbacks over 48h proposed
  • Show rate 70% → 80% · pod managers
  • Coded lost reasons · Will · "no reason" 41% → under 10% proposed
  • Reprice 73 units behind market, aged first · Will; Gary past $3K
  • Saturday up-log; fair lead share; Javier / Orlando plans · desks

60–90 days

  • Add 1–2 reps only if the gate passes; second truck closer with Ajay · Gary
  • Riata listings cleanup (AP item 9) and GA4 / Search Console / GBP read · Marri
  • Negative-front 35% → 30%; Brian GAP $146 → $250 · desks, F&I
  • Close certification gaps from the Ford report · training coordinator
  • Impel final invoice + AI cost per touched unit · Will
  • Re-run this audit mid-December, same definitions · Will
What we need

From Gary and others: data and decisions

Data exports / access

WhatFromUnlocks
Certification report by employee (Ford University / STARS)Training coordinator or GM loginThe "11" question
GA4 viewer, Search Console, Google Business ProfileMarri / website vendorVisits → leads; are crawlers blocked
13 months of ad and vendor invoicesAccounting / marketingCost per lead and per sale by source
Impel invoice + contract; AI usage exportWillAI cost per touched unit
On Hold reason for the 13 hidden unitsWill / inventoryStale hold vs real hold
CallRevu call detail + employee mapCallRevu adminRep-level call check; callback speed
Written-not-delivered log; F&I penetration and chargebacks by managerAccounting / CDKDeals lost after write-up; F&I coaching
Feed push log and price rules (tint, doc, rebates)Website vendor via MarriFix price display at the source

Decisions

  1. $649 tint: in the price or a true option (counsel first).
  2. FCTP rebate display on In-Service units (Ford rep + counsel).
  3. One owner for phone-ups and callbacks each day (proposed: pod manager on duty). The 318 catch-all leads move to pods 10/15.
  4. One BDC / appointment seat: yes or no.
  5. Hold the 3–4 rep hire until the 11/18 gate.
  6. Send Austin's east-town web pages to Manor (AP item 8).
  7. Reprice past the $3,000 cap on 5 aged units (AP item 10).
  8. Approve weekly 1pm training from Mon 10/19 (Monday topic, Tue–Thu drills).
  9. Impel stays off; no AI expansion until verified.
  10. Reset October: ~54 by Sat 10/17, ~97 month-end (not 75 / 125).
Appendix

Definitions and sources

Definitions

  • Person / staff-typed touch: a call, text or email a staff member made. Machine sends excluded. A CRM note is not a contact; a dial is not a connect.
  • Machine: our AI's first-touch, cadence, auto-reply, confirm and check-in sends. AI-drafted, person-approved counted separately.
  • Active lead: open in VinSolutions (not Sold, Lost, Bad, Duplicate). Fresh: notes synced in the last 24h (syncs ~6am, 10am, 2pm, 7pm).
  • Real leads: VS leads minus the July Ford ICI mailing import (2,078) and cross-store transfers.
  • Mature cohort: leads 30+ days old; 96% of sales land within 30 days. Mature 90 = leads 6/16–9/13.
  • Sold (any): VS Sold flag, Axcessa delivery tied by name, or VS sold feed. Delivered: Axcessa-tied only. Units: Axcessa retail + lease by date sold.
  • Show rate: came ÷ (came + no-show); cancels excluded.
  • GL gross: Axcessa accounting view, net of chargebacks. Deal-log: front + F&I per deal row.
  • Two-way conversation: genuine customer reply and a person's text within 48h.
  • rho: rank correlation, −1 to 1; 0 = no link. p90: the slowest 10% wait at least this long.
  • FCTP: Ford's loaner program (In-Service); CTV: Courtesy Transportation Vehicle.
  • Targets marked proposed are ours, from the store's best recent numbers; they need Gary's sign-off.

Sources (all files in the gary-audit folder)

TagFileAs of
Agary-A-followup.md / .json · recovery list gary-C-recovery.csv10/15 10:34 CT
Bgary-B-activity.md / .jsonOct 1–14; VS 10/14 7:09pm
Cgary-C-funnel.md / .json · gary-C2-recoverable-deals.csvleads 10/13 20:10Z; appts/Axcessa 10/15
Dgary-D-digital.md / .json · gary-E1-inventory-exceptions.csvVinCue 10/15 6:45am; site 10/15
Egary-E-ai.md / .json · gary-E2-ai-exceptions.csv10/15 ~11:00 CT
Fgary-F-financial-people.md / .json (RESTRICTED)Axcessa 10/15 7:22am
D2gary-D2-coaching.md (RESTRICTED, one-on-ones)10/15
F2gary-F2-training-plan.md10/15, proposed
Xgary-X-crossstore.md / .json (Manor / Austin / Buda)census Sep 1–Oct 12; read 10/15
APManor-Online-Action-Plan-2026-10-14-corrected.pdf10/14

Systems: VinSolutions (leads, notes, appointments, activity), Axcessa (units, gross, F&I), VinCue (inventory, pricing), CallRevu (calls), the Mini's send logs. Store notes keep the newest 20 lines per lead; 12 backups were merged to recover history. Closed Sundays.