Skip to content

Jumpstart Scaling · Operating system

Paid Acquisition Engineering

Performance media buying engineered for maximum ROAS across Google, Meta, TikTok, and LinkedIn for businesses dominating the market.

The Chaos of Traditional Media Buying

Moving from spray-and-pray to engineered scaling.

Most brands treat paid acquisition like gambling. They launch untested creatives, manually scale budgets, and pray for predictability. This approach leads to creative fatigue, auction competition, and diminishing returns.

Return on ad spend often drops when budgets scale because the variance was never tested.

”Traditional agencies operate on artistry, not engineering.”

The Scaling Protocol

Three rigorous phases to eliminate scaling variance.

1

Phase 1: Sandbox

Controlled multivariate testing with statistical thresholds. Server-side CAPI from day one. Winners advance only after MER thresholds are met.

Click to expand →

2

Phase 2: Extraction

Steady iteration on the angles that are working. Creative refresh systems for infinite variations and high-LTV audience expansion.

Click to expand →

3

Phase 3: Scale

Unlimited budget ramp with margin protection. Bidding strategies are set against your margin, not a fixed return target.

Click to expand →

Project Your ROAS

💰 ROAS Projection Engine

See how small improvements compound into massive profit.

Projected Revenue
$0.00
ROAS
0x
Below 1.5x
Net Profit (Ad Spend)
$0.00

0

Worked examples on this page

0

Business sectors covered

0

Calculator models behind the math

Infrastructure Wins Markets

Traditional agencies deliver campaigns. We deliver infrastructure. Campaigns produce temporary spikes. Infrastructure produces compounding returns.

When you engineer acquisition, every dollar spent builds on the previous dollar. Attribution is accurate. Scaling decisions are data-driven rather than hopeful.

Engineered acquisition turns paid media from an expense into a compounding asset.

Scalability Axiom

Ready to Engineer Your Acquisition?

Step 1 of 5: What type of business are you scaling?
01 / 05

What type of business are you scaling?

Modeled scenarios5 worked examples

How Paid Acquisition Engineering plays out for different kinds of business

Each example below starts from one kind of business and one problem. It shows the plan, the numbers we would track, and a short piece of arithmetic that uses assumed inputs. The inputs are ours, and you can change them in the matching calculator.

Modeled scenarioFast-growing startups and small software teams

A worked example for fast-growing startups and small software teams

Picture a 15-person software startup. A team pays a lot for developers and the product still breaks. Often the cause is no tests, no clear owner, and fixes stacked on fixes. A short audit finds the few things that break most.

Paid search and social are set up around one number: cost per qualified trial, not cost per click. Campaigns that bring sign-ups who never activate get cut.

Drafting tools, a small wooden block tower and a blank spiral notebook with a pencil.
Illustration for the software and digital companies examples. Not a photo of a client.
Where AI fitsAI assists

AI helps write ad variations and sort search terms. People set budgets and choose what to cut.

Modeled scenario

These are assumed inputs run through the same formula as our Google Ads ROI Calculator for Service Businesses. They are not a forecast and not a client result.

Assumed inputs
InputAssumed value
Monthly ad spend$8,000
Cost per click$6
Page conversion (%)4%
Close rate (%)12%
Average sale$4,800
Profit margin (%)70%
What the formula gives
ResultAs assumedIf page conversion is 5%
Leads per month53.366.7
Sales per month6.48
Revenue$30,720$38,400
Profit after ad spend$13,504$18,880
Cost per lead$150$120
Cost per sale$1,250$1,000

Moving “Page conversion” from 4% to 5% changes “profit after ad spend” from $13,504 to $18,880, a difference of +$5,376.

A clearer sign-up page is the cheapest lever, because it raises leads without raising spend.

Formula: clicks = spend / cost per click; leads = clicks x page conversion; jobs = leads x close rate; revenue = jobs x ticket; profit = revenue x margin - spend

Method: the inputs are our assumptions for a 15-person software startup, chosen to be plausible, not measured. Change them to your own numbers in the calculator and the same formula applies.

The plan

  1. List the three actions that predict a paying customer, and track them as events.
  2. Split campaigns by intent, such as brand, competitor, and problem terms, and give each its own budget.
  3. Review search terms weekly and add negatives. Pause any ad group with spend and no activated trials.

What we would track

  • Cost per activated trial
  • Share of spend on terms that never activate
  • Trial to paid rate by campaign

Risks to plan for

  • Small trial counts swing wildly, so wait for enough data before you cut a campaign.
  • Platform reports and your own database often disagree. Pick one source of truth.
How much should a startup test before scaling spend?

Enough sign-ups per campaign that one good or bad week does not change the answer. Set the minimum number of conversions before you start, and write it down.

Should we rewrite?

Rarely first. Fix the top few failures, measure again, and only then decide.

Modeled scenarioMarketing and creative agencies

A worked example for marketing and creative agencies

Picture a boutique creative agency. A new client signs, then waits days for kickoff. The wait looks sloppy and invites doubt. A fixed onboarding path, with the same steps each time, shortens that gap and makes the agency look as careful as it is.

For an agency, paid media brings in a few large clients, not many small ones. The plan targets decision makers on LinkedIn and search, and scores each lead by company size.

A stack of draft letters, a magnifying glass and an unpressed stamp beside an approval tray: AI drafts, a person decides.
What AI does

Assist only: turn the intake answers into a first-draft project brief for the lead to edit.

Where AI fitsAI assists

AI drafts outreach copy and summarizes a prospect's site. A person reviews every message.

Modeled scenario

These are assumed inputs run through the same formula as our ServiceTitan Home Services ROI Calculator. They are not a forecast and not a client result.

Assumed inputs
InputAssumed value
Leads per month30
Booking rate (%)40%
Close rate (%)20%
Average sale$36,000
What the formula gives
ResultAs assumedIf booking rate is 50%
Booked per month1215
Signed per month2.43
Revenue per month$86,400$108,000

Moving “Booking rate” from 40% to 50% changes “revenue per month” from $86,400 to $108,000, a difference of +$21,600.

Booking a call fast after a lead arrives is where agencies lose most prospects.

Formula: booked = leads x booking rate; signed = booked x close rate; revenue = signed x ticket

Method: the inputs are our assumptions for a boutique creative agency, chosen to be plausible, not measured. Change them to your own numbers in the calculator and the same formula applies.

The plan

  1. Define the buyer: title, company size, and the problem they search for.
  2. Run a small LinkedIn and search test with one clear offer, such as a short audit.
  3. Send every lead a booking link within minutes and track which source leads to proposals.

What we would track

  • Cost per booked call
  • Calls to proposals to signed clients
  • Average first-year client value by source

Risks to plan for

  • Large deals mean few data points, so decide on pipeline stage, not just closed deals.
  • Broad targeting brings small, price-driven buyers.
Is paid media worth it for a business that closes few deals each year?

It can be, if one client is worth a lot. Do the arithmetic first: how many calls do you need to win one client, and what can you spend per call?

Does a template make us look generic?

The steps are the same, the content is not. Clients tend to notice speed and clarity first.

Modeled scenarioOnline stores and ecommerce brands

A worked example for online stores and ecommerce brands

Picture an online store with a mid-priced product. Stock counts on one channel do not match the warehouse, so an item sells that is not there. Refunds and bad reviews follow. A single inventory number, shared by every sales channel, prevents most overselling.

The first number for a store is break-even return on ad spend, set by margin after fees and shipping. Ads run only where the return clears that line.

A stack of draft letters, a magnifying glass and an unpressed stamp beside an approval tray: AI drafts, a person decides.
What AI does

Assist only: draft product copy or answer common questions. Inventory numbers never come from AI.

Where AI fitsAI assists

AI helps produce image and copy variants. Budget decisions use your margin numbers.

Modeled scenario

These are assumed inputs run through the same formula as our E-Commerce Break-Even ROAS Calculator. They are not a forecast and not a client result.

Assumed inputs
InputAssumed value
Selling price$79
Product cost$24
Shipping cost$9
What the formula gives
ResultAs assumedIf shipping cost is $6
Margin per order before ads$43.41$46.41
Margin before ads55%58.8%
Break-even return on ad spend1.82 to 11.7 to 1

Moving “Shipping cost” from $9 to $6 changes “break-even return on ad spend” from 1.82 to 1 to 1.7 to 1, a difference of -0.12 to 1.

Shipping is a cost you can renegotiate, and it moves the break-even line directly.

Formula: fees = price x 2.9% + $0.30; margin per order = price - product cost - shipping - fees; break-even return on ad spend = price / margin per order

Method: the inputs are our assumptions for an online store with a mid-priced product, chosen to be plausible, not measured. Change them to your own numbers in the calculator and the same formula applies.

The plan

  1. Work out margin per order after product, shipping, and payment fees.
  2. Set a break-even return target and a goal above it. Track each campaign against both.
  3. Cut or fix any campaign that stays under break-even once it has enough spend to judge.

What we would track

  • Return on ad spend by campaign
  • Margin per order after fees
  • Repeat purchase rate within 90 days

Risks to plan for

  • Break-even ignores repeat buyers. If your repeat rate is high, a lower first-order return can still pay.
  • Discounts change margin. Recalculate when you run a sale.
What return on ad spend should I aim for?

It depends on your margin. Use the break-even formula on this page as the floor, then add the profit you want.

Do we need a custom system?

Often an existing sync tool is enough. Custom code makes sense when your channels or rules do not fit a standard tool.

Modeled scenarioSaaS and subscription software companies

A worked example for saaS and subscription software companies

Picture a self-serve software product. People sign up for a trial and never use the app. The first session is where most of the loss happens. A clear first task, a nudge when someone stalls, and a short check-in for larger accounts can improve it.

Spend is judged on lifetime value against cost to win an account. Ads that bring people who churn in month two are cut, even when sign-ups look cheap.

A stack of draft letters, a magnifying glass and an unpressed stamp beside an approval tray: AI drafts, a person decides.
What AI does

Assist only: draft answers to common support questions from your own docs. A person reviews the answers that matter.

Where AI fitsAI assists

AI helps cluster search terms and write test ads. People decide channel mix.

Modeled scenario

These are assumed inputs run through the same formula as our SaaS LTV to CAC Ratio Calculator. They are not a forecast and not a client result.

Assumed inputs
InputAssumed value
Monthly revenue per account$120
Gross margin (%)80%
Monthly churn (%)5%
Cost to win one account$900
What the formula gives
ResultAs assumedIf monthly churn is 4%
Lifetime value per account$1,920$2,400
Lifetime value to cost ratio2.13 to 12.67 to 1
Months to earn back the cost9.4 months9.4 months

Moving “Monthly churn” from 5% to 4% changes “lifetime value per account” from $1,920 to $2,400, a difference of +$480.

Lower churn raises lifetime value without any change in ad spend.

Formula: lifetime value = monthly revenue per account x gross margin / monthly churn; ratio = lifetime value / acquisition cost; payback months = acquisition cost / (monthly revenue x margin)

Method: the inputs are our assumptions for a self-serve software product, chosen to be plausible, not measured. Change them to your own numbers in the calculator and the same formula applies.

The plan

  1. Calculate lifetime value and payback months from your own billing data.
  2. Tag every sign-up with its source, then compare churn by source after three months.
  3. Move budget to sources with the best lifetime value to cost ratio.

What we would track

  • Cost to win a paying account
  • Payback months
  • Churn by acquisition source

Risks to plan for

  • Churn needs months of history. Early numbers are estimates.
  • Annual plans can hide churn for a year.
What is a good lifetime value to cost ratio?

Rules of thumb vary and depend on your cash. Watch payback months too, since a good ratio with a slow payback can still strain cash.

What should we measure first?

The share of new sign-ups who finish one key action in the first week. Pick the action that best predicts a paying customer.

Modeled scenarioOnline course creators

A worked example for online course creators

Picture a solo course creator with one flagship course. People buy a course and then cannot log in, so the creator becomes tech support. Most of those emails have the same few causes. A clear welcome email and a self-serve reset flow remove most of them.

Ads send people to a free lesson first, and only then to the course page. This lowers the cost of testing and builds a list you own.

A stack of draft letters, a magnifying glass and an unpressed stamp beside an approval tray: AI drafts, a person decides.
What AI does

Assist only: draft answers to common support questions from your own help page.

Where AI fitsAI assists

AI drafts ad copy and email subject lines. You choose what goes live.

Modeled scenario

These are assumed inputs run through the same formula as our Google Ads ROI Calculator for Service Businesses. They are not a forecast and not a client result.

Assumed inputs
InputAssumed value
Monthly ad spend$3,000
Cost per click$1.8
Page conversion (%)25%
Close rate (%)3%
Average sale$497
Profit margin (%)90%
What the formula gives
ResultAs assumedIf close rate is 4%
Leads per month417417
Sales per month12.516.7
Revenue$6,213$8,283
Profit after ad spend$2,591$4,455
Cost per lead$7.2$7.2
Cost per sale$240$180

Moving “Close rate” from 3% to 4% changes “profit after ad spend” from $2,591 to $4,455, a difference of +$1,864.

A short email series after the free lesson is the main lever on the buy rate.

Formula: clicks = spend / cost per click; leads = clicks x page conversion; jobs = leads x close rate; revenue = jobs x ticket; profit = revenue x margin - spend

Method: the inputs are our assumptions for a solo course creator with one flagship course, chosen to be plausible, not measured. Change them to your own numbers in the calculator and the same formula applies.

The plan

  1. Make one free lesson that solves a small, real problem and links to the course.
  2. Send ads to the lesson page and follow up with a five-email sequence.
  3. Track cost per email address and cost per sale separately.

What we would track

  • Cost per email sign-up
  • Email to buyer rate
  • Refund rate

Risks to plan for

  • Income claims in ads draw scrutiny. Say only what you can document.
  • A small email list gives noisy results, so judge over several weeks.
Should I run ads to a sales page or a free lesson?

Test both with small budgets. A free lesson costs less per lead, while a sales page costs less per sale when the offer is well known.

How do I find the top support causes?

Tag the last 50 support emails by reason. Fix the top two.

Hypothetical case studies2 planning examples

How you would use Paid Acquisition Engineering in a business like yours

Each study below takes one kind of business and shows how this service would be put to work: who does each step, what the first 90 days look like, what to measure, and the arithmetic with assumed inputs. A planning example, not a client result.

Hypothetical case studyOnline course creators

A planning example, not a client result.

How a course creator could bring more email subscribers into the live launch session

The situation

Picture an online course creator with a free guide, an email list, and a launch twice a year. The creator teaches a skill to beginners and sells a course at a moderate price. Most buyers come from the email list during a launch.

The list is large and quiet between launches. During a launch, only a small share attend the live session. Sales emails go to everyone alike. A few buyers ask for help in ways the course does not cover, and the creator answers one by one.

The plan, step by step

  1. Person Creator. Write the launch plan: the free session topic, the date, the offer, and the closing date.
  2. Software Sign-up and reminders. Let subscribers pick a session time and send reminders a day before, an hour before, and at start.
  3. Software Segmented emails. Send different follow-up to people who attended, who registered and missed it, and who never registered.
  4. AI FAQ drafter. Draft answers to common course questions from the creator's own materials. The creator approves each one before it goes into the launch emails.
  5. Person Creator. Run the live session, take questions in person, and open the offer with clear terms.

The first 30 days

  • Write the launch plan and the free session.
  • Clean the email list and remove addresses with no consent.

Days 31 to 60

  • Send the invitation and reminders.
  • Run the live session and take questions.

Days 61 to 90

  • Close the offer and review results by segment.
  • Write the notes for the next launch.

What to measure, and how

  • Subscribers who register for the session. Divide registrations by subscribers who received the invitation.
  • Registered people who attend. Divide attendees by registrations.
  • Attendees who buy. Divide purchases by attendees, and compare with non-attendees.

The worked calculation

These are assumed inputs run through the same formula as our ServiceTitan Home Services ROI Calculator. Put your own numbers in the calculator to try it. Nothing here is a forecast or a client result.

Assumed inputs
InputAssumed value
Leads per month500
Booking rate (%)30%
Close rate (%)8%
Average sale$400
The arithmetic, one step at a time
StepCalculationAs assumedIf booking rate is 40%
Booked per month 500 × 30% 150 200
Signed per month 150 × 8% 12 16
Revenue per month 12 × $400 $4,800 $6,400

Moving booking rate from 30% to 40% changes revenue per month from $4,800 to $6,400, a difference of +$1,600.

More reminders and a choice of session times are aimed at the share of subscribers who show up live.

Formula: booked = leads x booking rate; signed = booked x close rate; revenue = signed x ticket

Leads are subscribers who see the invitation. Booked means they attended the live session. Signed means they bought. The inputs are invented.

Risks and limits

  • Too many emails cause unsubscribes and spam flags. Keep the frequency fair.
  • A live session can fail from technical trouble. Run a test session first.
  • Do not promise results from taking the course.

What would be needed from the owner

  • A cleaned email list with consent.
  • A date for the session and the offer.
  • Course materials for the FAQ.
  • A backup plan for tech problems.

Hypothetical case studySaaS and subscription software companies

A planning example, not a client result.

How a growing subscription company could measure revenue kept from existing customers

The situation

Picture a subscription software company with a customer success lead, support staff, and a monthly recurring revenue base. The company has hundreds of paying accounts, from small teams to a few large ones. Growth comes from new sales and from existing accounts adding seats.

Leadership watches new revenue and total revenue. Nobody splits revenue lost from cancellations, revenue lost from downgrades, and revenue gained from upgrades. Customer success reacts to angry tickets. The company does not know if its base is growing or leaking.

The plan, step by step

  1. Person Customer success lead. Define the groups for retention reporting: monthly or yearly plan, plan size, and customer age.
  2. Software Revenue table. Build a monthly table of start revenue, upgrades, downgrades, and cancellations for the same set of customers.
  3. Software Use signals. Combine use data with the revenue table to flag accounts that are using the product less and have a renewal coming.
  4. Person Customer success manager. Contact flagged accounts before renewal, learn what is missing, and report the pattern to product.
  5. AI Ticket summarizer. Summarize a flagged account's recent tickets for the manager. The manager reads the original tickets when it matters.

The first 30 days

  • Define the groups and build the revenue table.
  • Compute net revenue retention for the last year.

Days 31 to 60

  • Add use signals and start the pre-renewal calls.
  • Log the answer from each call.

Days 61 to 90

  • Review the table for each group.
  • Pick one cause of cancellations to fix with product or support.

What to measure, and how

  • Net revenue retention. Divide revenue from the same customers now by their revenue at the start of the period.
  • Cancelled revenue by cause. Tag each cancellation with a cause and sum the lost revenue by cause.
  • Pre-renewal contacts made. Divide flagged accounts with a call before renewal by all flagged accounts.

The worked calculation

These are assumed inputs run through the same formula as our SaaS Net Revenue Retention (NRR). Put your own numbers in the calculator to try it. Nothing here is a forecast or a client result.

Assumed inputs
InputAssumed value
Starting monthly revenue$50,000
Expansion$2,500
Downgrades$1,200
Cancellations$2,000
The arithmetic, one step at a time
StepCalculationAs assumedIf cancellations is $1,500
Start plus expansion $50,000 + $2,500 $52,500 $52,500
Less downgrades $52,500 − $1,200 $51,300 $51,300
Revenue kept from the same customers $51,300 − $2,000 $49,300 $49,800
Net revenue retention $49,300 ÷ $50,000, as a percent 98.6% 99.6%

Moving cancellations from $2,000 to $1,500 changes net revenue retention (%) from 98.6% to 99.6%, a difference of +1%.

Pre-renewal calls and fixes to the top cause of cancellation are aimed at the revenue lost to cancellations.

Formula: retention = (starting revenue + expansion - downgrades - cancellations) / starting revenue

The figures are monthly recurring revenue from one set of customers. They are assumptions for the example, not data from any company.

Risks and limits

  • Retention by revenue hides churn among small accounts. Report by count as well.
  • A short period can mislead because of annual renewals bunched in a few months.
  • Customer success cannot fix a product that does not fit the buyer.

What would be needed from the owner

  • A revenue history by customer.
  • A way to join billing with product use.
  • A tagging rule for cancellation reasons.
  • Time for customer success to make renewal calls.

Hypothetical case studys

A planning example, not a client result.

More planning examples that use Paid Acquisition Engineering

These hypothetical case studies use this service as one part of a wider plan. Each opens on the service page that holds the full version.

Local service catalog

Explore Paid Acquisition Engineering by published market—only when you choose to.

General pages explain the system. The Local Intel catalog is where visitors can intentionally select a published market variation—without being redirected by IP location.

Browse Local Intel
Regional Control Plane Deployments

Localized Service Architectures & Market Landers

Verified revenue control planes tailored to the regulatory, market density, and unit economic constraints of active regional metropolitan markets.

Browse all 50 state directories →
Wichita, KS Data Core

High-Concurrency Database Systems in Wichita

Engineered revenue infrastructure for scaling operators in Sedgwick County. Deploying sub-second routing and closed-loop attribution near Keeper of the Plains.

Local Invariant Transaction Integrity and Tested Recovery
Inspect Wichita Architecture →
Dallas, TX Pipeline State Machine

CRM Pipeline & Lead Routing in Dallas

Engineered revenue infrastructure for scaling operators in Dallas County. Deploying sub-second routing and closed-loop attribution near Reunion Tower.

Local Invariant Pipeline Loss Detection and Recovery
Inspect Dallas Architecture →
Bridgeport, CT Vector Intelligence

Qdrant Vector Search Engines in Bridgeport

Engineered revenue infrastructure for scaling operators in Fairfield County. Deploying sub-second routing and closed-loop attribution near Seaside Park.

Local Invariant Sub-50ms HNSW Recall
Inspect Bridgeport Architecture →
Raleigh, NC Conversion Rate Ops

High-Conversion Funnel Systems in Raleigh

Engineered revenue infrastructure for scaling operators in Wake County. Deploying sub-second routing and closed-loop attribution near North Carolina State Capitol.

Local Invariant Sub-Second Lead Response
Inspect Raleigh Architecture →
Miami, FL Vector Intelligence

Qdrant Vector Search Engines in Miami

Engineered revenue infrastructure for scaling operators in Miami-Dade County. Deploying sub-second routing and closed-loop attribution near South Beach.

Local Invariant Sub-50ms HNSW Recall
Inspect Miami Architecture →
Fort Worth, TX Autonomous Runtimes

Rust AI Orchestration Services in Fort Worth

Engineered revenue infrastructure for scaling operators in Tarrant County. Deploying sub-second routing and closed-loop attribution near Fort Worth Stockyards.

Local Invariant Workload-Specific Latency Targets Tokio RTT
Inspect Fort Worth Architecture →
Global Architecture Index

Engineered AI & Control Plane Infrastructure

Autonomous workflows, vector intelligence, and memory-safe systems built to scale business operations without fragility.

Refactoring & Hardening
Code Stabilization
For: Founders with Broken MVP Code

Vibe Code Repair

Refactor fragile prototypes into production-grade, typed architectures.

High-Dimensional Indexing
Vector Engine
For: AI Platform Architects & Engineers

Qdrant Vector Search Infrastructure

HNSW vector indexes and multi-tenant collection clustering at scale.

Problem Addressed

Search results feel random even though embeddings are being stored.

State Machine Orchestration
Control Plane
For: Distributed Infrastructure Operators

Rust AI Orchestration Services

Deterministic Tokio worker pools and event-driven control plane loops.

Problem Addressed

The orchestration logic is hidden inside prompts or scattered helpers.

Technical Specifications
Architecture RFCs
For: Product Leads & Enterprise Buyers

Technical Documentation and Writeups

Definitive engineering scope blueprints that eliminate developer confusion.

Data Infrastructure
Database Core
For: High-Volume SaaS & Logistics Leaders

Database Design and Scaling

Schema indexing, write-path replication, and sub-second analytical queries.

Retrieval Augmented Generation
Hybrid RAG
For: Enterprise Knowledge & AI Teams

Qdrant RAG Pipeline Engineering

Dense + sparse hybrid search with cross-encoder reranking and payload filters.

Problem Addressed

The model answers confidently but pulls the wrong context.

Deterministic Architecture Discovery: Showing specialized subsystems suited to your active workflow context.

6 Clusters Active • Distinct Service Discovery Paths

In motion

See the system move.

A six-second look at a growth system in motion.

6 sec