KW / KYLESCOTTWILLIAMS.COM — PANEL SCHEDULE
SYSTEMS SHOWN BELOW ARE IN PRODUCTION
Kyle Williams — Marketing Operations & AI Systems

I build the systems lean teams run on.

Attribution you can trust, AI workflows that eliminate manual work, and the reporting a business steers by — designed, built, and shipped by one person, inside a real operating company. Before that: a SaaS startup founded, scaled, and sold, and twelve years of P&L-accountable growth marketing.

Inspect the work ↓ kyle@imurgent.com
17 yrs
Marketing & systems
1 exit
SaaS founded & acquired
$50–60K
Monthly ad spend managed
200+
SaaS clients served
10,000+
Devices shipped, 3 countries

The work, circuit by circuit

Five systems from the last twelve months as Operations Director at RD Electric, an electrical contractor where I own the entire marketing and operations stack. Customer data and company financials are redacted; the architecture is shown in full.

CKT-01

Making attribution trustworthy again

INTERACTIVE

Fault found

Reported conversions didn't match the phone ringing. The audit found the primary Google Ads conversion — phone calls — firing on bot traffic, plus three redundant analytics tag loaders left by prior vendors, each double-counting differently.

Rewired

  • Consolidated three competing loaders into one Google Tag Manager container
  • Rebuilt conversion definitions to count only verified calls and forms; instrumented UTM governance
  • Stood up Windsor.ai + Search Console cross-checks against booked work
  • Killed a display campaign the clean numbers proved was negative-ROI
Try the discovery yourself
reported conversions — inflated by a call event firing on bot traffic
Illustrative data. The real audit found the primary phone-call conversion counting bots; filtering restored numbers leadership could trust.
ResultConversion counts reconcile with reality; spend decisions run on numbers leadership trusts.
CKT-02

An AI back office

INTERACTIVE

Fault found

A small office drowning in repetitive work — retyping vendor invoices, compiling daily numbers, chasing crew hours — with no budget for admin headcount.

Rewired

  • AI invoice pipeline: reads invoices emailed by three suppliers, extracts line items with an LLM, posts structured records — zero manual entry
  • Daily digest: automated morning email assembling yesterday's jobs, hours, and exceptions from the CRM API
  • Crew-hours reconciliation and document/photo tagging, generated on schedule
ResultHours of weekly manual work eliminated — and a standing pattern: repetitive process → agent-style workflow, built proactively.
Run the invoice workflow
// press Run — this simulates the production pipeline, step for step
Simulation of the live workflow. The digest below is a real (redacted) output from the production system.
Redacted example of the automated Daily Field Digest email: per-job crew hours and plain-language work summaries, generated from the CRM API.
CKT-03

20-step Zapier chain → one native workflow

INTERACTIVE

Fault found

Service-job intake ran through a brittle 20-step Zapier automation — slow, costly, and broken whenever an upstream field changed. Classic no-code sprawl.

Rewired

  • Collapsed the chain into one maintainable, webhook-driven workflow against the CRM's API
  • Added enrichment Zapier never had: automatic Street View imagery of the job site on every new service job
  • Documented the flow so it isn't tribal knowledge
ResultFaster intake, near-zero maintenance, lower tooling cost — and proof I know when no-code is the right tool and when it isn't.
The rewire, animated
Trigger: job createdPath splitPath checkFilterFormatter: dateFormatter: startFormatter: endFind cost groupCode: JSAdd cost groupCode: JSPath splitPath checkLoop textFind recordCreate taskPath checkCopy: find IDCopy: create taskPath checkFilterCreate contract jobPath checkFind folderFilterCreate folderUpdate job
27 chained steps · 4 path splits · breaks when any upstream field changes
The before picture: the 27-step Zapier chain this workflow replaced.
A service job record with the auto-attached Google Street View photo of the job site (customer details redacted).
CKT-04

SpeedySparky — a self-service quoting funnel

Fault found

Quoting small electrical jobs was a bottleneck: every inquiry waited on a callback before hearing a number, and slow quotes lose jobs.

Rewired

  • React application with a 10-step guided intake, from job type to a usable estimate
  • Every question structured to emit clean, routable lead data — not a free-text form
  • Shipped as a standalone funnel alongside paid and organic acquisition
ResultA bottleneck became a conversion asset that qualifies leads while it quotes them. Gated behind a customer login — I walk it through end to end in a live demo.
CKT-05

The scoreboard the business runs by

INTERACTIVE

Fault found

Revenue without an operating model: no breakeven, no capacity math, no monthly target grounded in the cost base — and marketing decisions floating free of finance.

Rewired

  • Capacity-based financial model: fixed-cost base, breakeven, monthly revenue targets from billable capacity
  • A live scoreboard leadership reviews monthly, tying job volume and revenue mix to targets
  • Year-over-year revenue by job type, so spend follows the work worth winning
ResultMonthly business reviews run on a shared, current picture — the rhythm-of-business function, built from scratch. (Structure shown; figures redacted.)
The scoreboard, rebuilt with illustrative data structure real · values synthetic
$0
Revenue YTD ($K)
0
Jobs won
0%
vs. breakeven
0%
Proposal win rate
Service 46%
Residential 25%
Commercial 16%
Other 13%
The live sales scoreboard structure — revenue trend, mix by job type, win rate, and per-PM pipeline (all figures redacted).

How I got here

One continuous line: build a marketing system, run it on a P&L, then build the next one bigger.

2009 – 2013

QpINs.com — Founder / Owner

Founded an SMS loyalty-marketing SaaS. Permission-based lifecycle marketing before the term existed: 200+ clients, 10,000+ loyalty kiosks shipped across the US, Canada, and Mexico, two vertical software lines licensed. Acquired by Inland Cellular.

2013 – 2025

Inland Cellular — Agency Marketing Director

Joined via the acquisition; grew into leading the agency marketing practice and enterprise portfolio. $50–60K/month in managed ad spend, 20–25% year-over-year growth on a recurring-revenue base for eight consecutive years, cross-functional GTM launches, and an M&A integration that doubled the client base.

2025 – present

RD Electric — Operations Director

Recruited to modernize a licensed electrical contractor. Own the full GTM engine, the financial reporting, and every system on this page — the connective tissue between field operations, marketing, and finance.