Too few live calls
Keep collecting evidence. Do not kill a young test for being young.
FIELD REPORT 04 · DECISION INFRASTRUCTURE · B2B HIGH TICKET SALES CLIENT
The data lake
behind Cut,
Scale & Kill
For a B2B high ticket sales client, I joined paid media to live calls, revenue and cash, corrected the numbers everyone trusted, and built the decision system that tells every ad to Watch, Cut, Scale or Kill.
$500K+monthly paid media
under decision control
under decision control
622duplicate meetings
found and removed
found and removed
● DECISION LEDGER · ONE AD · 14 DAY WINDOWILLUSTRATIVE AD
DAY 14 · VERDICTSCALE
Mature and under target. Raise the budget in controlled steps, with the sales team's capacity in view.
SPEND$3,500$250 a day
LIVE CALLS74 to act
COST PER LIVE CALL$5000.63x target
COST PER BOOKING$23315 bookings
Illustrative ads, target and thresholds. The rule shape is the real one: rank on trailing cost per live call, act only once there is enough live call volume.
01 THE BRIEF
The dashboard was not the system.
The client sells a high ticket offer through booked sales calls and puts more than $500K a month into paid media to fill the calendar. Since March 2026 I have run lead generation on the account and built the data layer under it.
The job was to connect the complete path from spend to cash, then turn that evidence into a repeatable operating decision. Meta knew the spend. HubSpot held the contacts, the meetings, the deals and the payments. Each system answered a different part of the decision.
Bookings were too early and show blind. Closes were the true value signal, but sparse and often five to ten days late. The reporting needed one definition of performance that survived every source.
ROLEData architecture and lead generation
SPEND$500K+ a month in paid media
OUTPUTDaily health, weekly decisions
WHENMarch 2026 to now
One number had to survive five systems and still arrive in time to move a budget: cost per live call.
02 THE DECISION METRIC
Cost per live call sat in the useful middle.
Same four illustrative ads as the ledger above. Pick a metric, move the read day, and see which ad each one would put first.
● RANK FOUR ADS BY ONE METRICILLUSTRATIVE ADS · 4 EVENTS TO RANK
DECISION METRIC
WHEN EACH SIGNAL LANDS · ONE LEAD
#1AD ASCALE
#2AD BCUT
#3AD CKILL
TOO FEWAD DWATCH
On day 14, cost per live call puts AD A first and AD C last. That is the order the business would want.
SPEEDEARLY ENOUGH
QUALITYQUALIFIED BEHAVIOR
USEUSE TO RANK
Live calls reflect whether booked leads actually showed up and qualified, early enough to manage budget before close data matures.
Rank on trailing cost per live call. Act only when enough live call volume makes the signal meaningful.
03 THE SYSTEM MAP
Five sources, one layer, one set of definitions.
Daily, weekly and monthly views read from the same definitions. No metric changed meaning between reports.
Join keys simplified for the page. The real layer carries more fields per record.Stack: Google Sheets, HubSpot, Meta Ads, Coefficient, Python, Google APIs, Meta Marketing API.
04 INTEGRITY AUDIT 01 · MEETING RECORDS
622 meetings were counted twice.
HubSpot's calendar behavior left a second, cancellation prefixed record for the same meeting. Both counted. Toggle the audit to see what one fix did to the show rate.
● TWIN DEDUPLICATIONMATCH ON SAME OWNER + SAME START TIME
ONE PAIR, SCHEMATIC
ALL 622 TWINS · ONE SQUARE EACH
622extra meeting records
in the booked denominator
in the booked denominator
APRIL SHOW RATE
19.0%23.7%+4.7 pts
MAY SHOW RATE
20.3%22.9%+2.6 pts
Show rate is live calls over booked meetings. Every twin added one booked meeting that could never show, so booked volume read high and the show rate read low. The pair above is a schematic; the 622 and both show rates are the audit's real figures.
The correction changed no calls. It made the show rate true.
05 INTEGRITY AUDIT 02 · CASH AND REVENUE
$95,557 hid behind one naming edge case.
The evergreen report excluded webinar sales by label. The rule matched one exact label, so every dated webinar variant flowed into evergreen: closes, revenue and cash.
● WHERE DATED WEBINAR SALES LANDEDLABEL NAMES ILLUSTRATIVE
IN THE WRONG FUNNEL
$95,557
Closes, revenue and cash from dated webinar variants, credited to evergreen. Evergreen looked better than it was.
BEFORE
exclude: "webinar"ONE LABELAFTER
exclude: "webinar*"THE WHOLE FAMILY1,008formulas repaired
0new formula errors
Validation: the cash removed from evergreen matched the full dated webinar payment total exactly, and every protected metric came out of the repair unchanged.
06 THE DECISION ENGINE
Every call starts with one question: what is proven?
Maturity came before action. A young test was watched. A stale close could not protect recent inefficiency. A low show rate did not condemn a test without enough live call volume.
Map drawn with illustrative thresholds: four live calls to act, an $800 target, Kill past 1.5x. Same four ads as the ledger at the top.
Mature but drifting
Reduce exposure while preserving enough volume to confirm the direction.
Efficient live call volume
Increase budget in controlled steps and keep the sales capacity ceiling visible.
Mature and consistently inefficient
Stop funding a narrative the downstream outcomes no longer support.
07 OPERATING RHYTHM
Faster reporting mattered. Harder to fake decisions mattered more.
Two weeks on the account. A health check every morning, a creative review every Friday, and a decision window that always looks back 14 days.
07:15Daily health
Check source freshness, filters and pipeline failures before anyone reads the report.
FRICreative review
Read hooks and formats against qualified applications, live calls and revenue.
14DDecision window
Rank meaningful volume by cost per live call, then Watch, Cut, Scale or Kill.
15 SECOND SYSTEM SUMMARY
From source data to a budget decision, in one square.
The motion piece compresses the architecture, the integrity audits, the decision signal and the weekly rhythm into a format built for a LinkedIn feed. The PDF has the same story in seven pages.
08 THE RESULT
Growth decisions got harder to fake.
Both audits found errors that flattered someone: extra bookings in the calendar, extra revenue in evergreen. The fixes made the numbers true before any budget moved on them.
622cancellation prefixed duplicate meetings removed
23.7%April show rate, corrected from 19.0%. May: 20.3% to 22.9%
$95,557moved out of the wrong funnel by one wildcard
1,008formulas repaired, zero new errors
Figures are from my own audits of the client's reporting, March 2026 to now. Ad names, targets and thresholds in the diagrams are illustrative. Raw volumes, revenue and the client's name stay with the client.
09 WHAT CARRIES OVER
What I would build first on any account with a sales team.
01
Pick the decision metric on timing and quality together. The earliest number is usually show blind, and the truest one lands after the budget call.
02
Audit the joins before trusting the trend. Both errors on this account sat in the plumbing: a calendar habit and a naming rule.
03
Write the rule for young data first. Watch is a decision too, and it saves more tests than any other.