← → / space to navigate
SystemsGoCase study 01 / 13
RevOps Case Study

Rebuilding a B2B SaaS funnel around unit economics

A sales-led company was booking inbound demos on availability rather than fit. We rebuilt qualification and routing around lifetime value and cost to serve.

SystemsGo·Illustrative engagement, figures anonymized·2026
ScopeDiagnosis, system design, and build over ten weeks.
APPROACH02 / 13
Approach

Three stages: truth, diagnosis, change.

The company was hitting bookings and quota while profit stayed flat. Rather than add pipeline, we reconciled customer value across its systems, located where margin and selling time were lost, and rebuilt the qualification responsible for the loss. The rest of this document follows those three stages.

Truthone reconciled view Diagnosisfind the constraint Changerebuild the system ↻
TRUTH · DATA RECONCILIATION03 / 13
Stage 1 · Data

Fourteen sources, reconciled into one customer record.

LTV
subscriptions
billing
plan changes
CAC
marketing spend
sales overhead
commissions
Cost to serve
support tickets
onboarding
support spend
server cost
ICP-fit
firmographics
usage
integration depth
Rep load
meetings
calendar
↓deduplicated and reconciled to one record per customer
customer_value_model1,840 customers · net of cost to serve
CustomerLTVCACCost/serveGross marginICP-fit
Northwind$210k$61k$38k82%strong
Vantage$96k$24k$18k81%strong
Harbor Co$21k$12k$10k52%weak
Dunmore$14k$13k$9k36%weak
1,836 more rows

LTV came from subscription and billing history, CAC from marketing and sales spend, cost to serve from support, onboarding, and infrastructure. Fourteen sources, reconciled to one record across 1,840 accounts.

DIAGNOSIS · LTV CONCENTRATION04 / 13
Value concentration
0%

of lifetime value came from the top 18% of customers.

Those accounts retained at 95% a year against 69% for the rest. The pattern stayed hidden until value, cost, and usage were joined into one model.

Cumulative lifetime value by customern = 1,840 · net of cost to serve
0100% 050%100% customers, ranked by lifetime value → 18% of customers 80% of value
DIAGNOSIS · COST TO SERVE05 / 13
Cost to serve

Cost to serve scaled far slower than contract value.

Enterprise contracts ran 10 to 20 times a small account but only 3 to 5 times the cost to deliver and support. After cost to serve, gross margin was 82% at the top of the book and 36% at the bottom.

$0$25k$50k $0$130k$260k annual contract value → annual cost to serve → 82% margin 36% margin
enterprisemid-marketsmall
DIAGNOSIS · RETENTION vs PAYBACK06 / 13
Cohort retention

Small accounts churned before recovering CAC.

Net revenue retained, by deal-size cohortmonthly cohorts, trailing 24 months
050%100% CAC payback ~14mo Enterprise · 106% Small · 46%

Enterprise cohorts retained 106% of revenue at 24 months. Small-deal cohorts fell to 46% and crossed below CAC payback near month 14. About 40% of small accounts churned before breaking even.

DIAGNOSIS · SELLING TIME07 / 13
Selling time

Reps spent most selling time on accounts that would not return CAC.

With no SDR layer, AEs qualified inbound during the demo itself. Of every 100 hours of demo time, 25 reached strong-fit accounts, 55 went to small, churn-prone deals, and 20 to leads disqualified on the call.

25 · strong ICP 55 · small, churn-prone 20 · disqualified on the call
DIAGNOSIS · ACCOUNT TEAM08 / 13
From the account team

"My calendar is full of low-value junk I have to disqualify myself."

Account executive, sales team

The data located the loss. Interviews explained why it persisted: reps had no way to filter inbound before it reached their calendar.

CHANGE · INTERVENTION09 / 13
Intervention

Every inbound is scored and routed before it can book.

Every inbound lead Score every lead Enrich → Classify → Route before the calendar loads Strong ICP fitreaches a rep faster than beforeBOOK A DEMO On the fenceworth a conversation, not a full demoDISCOVERY CALL Good fit, low-touchself-serve + onboarding · rescued if it growsSELF-SERVE Partner / consultantown page, never touches salesDEFLECT Out of market / no fitunsupported region or languageDECLINE

Each inbound is enriched and scored on fit in the time the form takes to submit. An LLM classifies the lead; deterministic rules handle the edges. Strong-fit accounts reach a rep faster than before. The rest route to self-serve, partners, or a decline.

CHANGE · DELIVERY11 / 13
Delivery

Built and deployed in ten weeks, with no pause in selling.

✓ServiceThe classifier serviceReal-time enrichment plus an LLM-and-rules ICP classifier, behind the form.
✓RoutingSix-tier routingWired to the calendar, the CRM, and the scheduling tool, live.
✓LaneThe self-serve laneTrial, guided onboarding, and a self-billing portal. No contract, no rep.
✓RescueThe rescue pathEnrichment keeps re-scoring; a growing self-serve account gets pulled to sales.
✓DeflectPartner deflectionA separate enrollment page so partnership asks never reach a calendar.
✓OpsAutomation + alertsLifecycle automation per path, and one clean Slack alert per lead.
DeliveryBuilt end to endTen weeksNo pause in selling
RESULTS12 / 13
Results

Outcomes six months after rollout.

LTV : CACpayback under 12 months
before
2.8×
after
4.6×
Quota attainmentper rep
before
64%
after
82%
AE time on ICP accountsshare of demo hours
before
25%
after
70%
Revenue per support FTEsame headcount
before
base
after
+40%

Acquisition efficiency and selling time both improved while support headcount held flat.

NEXT STEP13 / 13
Work with us

Run the same analysis on your funnel.

If your team is busy but margin is flat, the cause is often in data that has never been joined. We reconcile it, locate the constraint, and propose the change before any larger engagement.

Start here
Book a working sessionhello@systemsgo.co

Reconcile the data, find the constraint, change the system.

SystemsGo