MMcELROYAD AGENCY

THE BUSINESS BLUEPRINT / NOVEMBER 2026

A focused agency.
A compounding advantage.

A proposed operating plan for McElroy Ad Agency: win a narrow market, connect marketing to sales, and build repeatable delivery before expanding.

Explore the 90-day plan ↓

Planning assumptions and targets, not current client results or financial forecasts. No advertising, vendor purchase, or account integration is activated by this website.

01 / POSITIONING

Own a problem.
Then own a category.

Position: a revenue-focused marketing partner for established DFW service businesses that already deliver a valuable service but cannot reliably connect advertising, lead response, and sales results.

Beachhead: choose one home-service niche—such as HVAC replacement, remodeling, or roofing—through 12 owner interviews and a sample of actual business economics. Do not launch all three simultaneously. Favor repeatable offers, sufficient service capacity, documented margins, and a decision-maker willing to share sales outcomes.

The ideal first client

An owner-led company with roughly $1M–$10M in annual revenue, a working sales process, and $3,000–$6,000 per month available for media in addition to the pilot fee. These are prospecting filters, not claims about existing clients. Confirm affordable acquisition cost from contribution margin and close rates before recommending spend.

Separate real estate specialty: use David’s domain experience for brokerages and seller-service campaigns, with separate intake, creative review, and housing-ad policy checks. Do not use household age, children, or other protected traits as housing-audience shortcuts. Keep brokerage referrals and agency compensation separately documented.

Decline or defer: clients without fulfillment capacity, unreliable sales records, unrealistic guarantees, or a budget too small to test their economics. Offer the diagnostic rather than force an unsuitable retainer.

Why clients would stay

The defensible asset is the accumulated learning: permissioned first-party outcomes, niche-specific creative concepts, documented experiments, reliable operations, and a shared definition of a good customer. Access to the same AI tools as everyone else is not a moat.

02 / BUSINESS MODEL

Productize the work.
Keep ownership with the client.

The homepage’s three proposed packages create a progression from a $1,500 diagnostic, to a $2,000 setup plus $2,500/month 90-day pilot, to a growth partnership from $4,500/month. Clients own their ad accounts, domains, CRM records, and paid-for creative exports. Media is paid directly to platforms.

Commercial structure and boundaries
Revenue streamWhat the client buysScope control
DiagnosticEconomics, access/data audit, prioritized roadmapOne business unit; 10-business-day target after complete inputs
SetupOne landing page, tracking design, CRM stages, launch QAOne website and one CRM; major migrations quoted separately
Recurring retainerMedia operations, distinct creative concepts, sales feedback, decisionsExplicit channel and creative limits; two revision rounds per concept
Expansion projectsAdditional markets, integrations, video shoots, or advanced measurementSeparate written scope and price

Use a 90-day pilot with written deliverables and a review at completion. Bill setup and the first month before work begins; bill media through the client account. After the pilot, use monthly service with 30-day cancellation. Credit the diagnostic against setup only when the work directly replaces setup tasks.

Do not start with revenue-share pricing. Attribution disputes, sales-team performance, refunds, and long closing cycles make it hard to administer. Consider a capped bonus later, only against a pre-agreed baseline and verified incremental contribution after refunds. Define attribution windows, customer exclusions, reconciliation, and dispute resolution before signing.

The pilot’s cash requirement

At the lower suggested media level, 90 days cost approximately $19,100: $2,000 setup + $7,500 retainers + $9,000 media + an illustrative $600 in tools/data. At $2,500 contribution per additional customer, eight additional customers cover that cost before fixed overhead and tax. Sales lag can push cash recovery beyond 90 days.

03 / ACQUIRING AGENCY CLIENTS

Lead with a useful diagnosis.

Flagship offer: a “Lost Opportunity Review”—a brief, personalized review of the prospect’s publicly visible offer, landing-page friction, and follow-up questions. Use only observable evidence. Never claim to have seen their ad account, customer records, or revenue without access.

Warm distribution first

Invite introductions through David’s business relationships, local business groups, website partners, and CRM implementers. Run one practical workshop per month: “Where ad spend disappears between the lead and the sale.” Obtain permission before publishing client examples.

Owned proof every week

Publish one original service-business teardown or field lesson weekly, then repurpose it into a short video and an email for subscribers. Explain a concrete problem, show the fix, and offer the diagnostic. Use synthetic examples until real, approved evidence exists.

First 30 days: a proposed funnel

Research 40 named target businesses, seek 12 owner conversations, deliver six tailored reviews, and aim to sell two paid diagnostics. These are activity goals—not assumed conversion rates. Log source, conversation, diagnostic, proposal, win/loss, labor hours, and collected fees.

Use highly relevant B2B outreach only after checking source permissions, suppression, vendor terms, and applicable requirements. Commercial email must include accurate identity, a valid postal address, and a working opt-out. Honor suppression immediately under agency policy. A purchased or enriched contact is not consent to automated calls or texts.

Agency paid acquisition: begin only after the offer converts through direct conversations. Cap the initial test at $1,500 total, funded from available operating cash. Use one high-intent search offer and one landing page. Track qualified conversations and paid diagnostics; exclude branded traffic from new-demand claims. Do not add paid spend automatically when the cap is reached.

Funding rule: collect upfront fees, maintain at least two months of fixed operating expenses, and reinvest up to 20% of collected agency contribution into acquisition after that reserve is met. Never fund the agency’s growth from a client’s media balance.

04 / DELIVERY ARCHITECTURE

Buy data only when
it changes a decision.

  1. Define and clean

    Set the ICP and required fields. Normalize email, company domain, phone, and address; reconcile duplicates; preserve source and timestamps; check suppression and permission. Distinguish an unknown field from a failed verification.

  2. Enrich selectively

    Start with existing first-party records and one licensed provider. Add one fallback only where the first provider leaves a commercially useful gap. Stop after a verified hit; cap the cost per accepted usable record. Use PropertyRadar for appropriately licensed property context, not as a universal B2B contact database.

  3. Research with evidence

    AI research may append public business facts such as service area, offer, or relevant hiring. Require a source URL, retrieval date, and confidence. Abstain when evidence conflicts. Treat MCP as a connection method, not a guarantee of data quality.

  4. Capture and route

    Store inquiry purpose, consent version, submission timestamp, source campaign, and CRM identity. Route to a named owner. Separate raw inquiry, qualified lead, appointment held, opportunity, won, and collected revenue.

  5. Return outcomes

    Send permitted conversion signals to ad platforms only after client configuration and testing. Reuse the same event ID/name for overlapping browser and server events; maintain idempotency in the agency’s own ledger. Reconcile refunds and updated values.

Lean stack first; complexity only when justified
LayerLaunch choiceUpgrade trigger
Site & captureNetlify landing pages + verified form receiptAdd a consent platform before non-essential ad tags where required
System of recordClient’s existing CRM; a standard stage and event dictionaryMove from a small pilot sheet to a CRM before multi-user scale
Workflown8n with separate client credentials, receipts, retries, alertsQueue/warehouse when volume or recovery needs exceed simple workflows
EnrichmentOne tested source + email validation; capped fallbackWaterfall when measured coverage gain exceeds incremental cost
CreativeOriginal expertise and licensed assets; AI-assisted drafts/videoSpecialist production after concept validation
Conversion feedbackMeta Pixel/CAPI and Google enhanced conversions where suitableUse supported Google Data Manager routes for new offline uploads
Advanced analyticsCRM cohorts and financial reconciliationPower-qualified lift test; MMM with sufficient history and expertise

Provider bake-off: test the same randomly selected 200–500 target records with each candidate. Score fill rate, independently checked correctness, freshness, cost per usable record, and lawful usability separately. A found email and a deliverable email are different outcomes. Stop purchasing a provider that loses on your own niche, regardless of its global claims.

Refresh by risk: validate a contact before use; update high-change fields on relevant events; review stable business attributes less often. Do not trigger expensive enrichment on every page visit or every minor CRM edit.

05 / CREATIVE & AI DISCOVERY

Distinct ideas beat
an endless pile of variants.

Use four concept families in the first monthly pilot: a customer problem explained by an expert, a process demonstration, an objection answered with evidence, and a comparison that makes the decision clearer. Produce one core asset and the necessary placement adaptations for each. The four-concept cadence is a scope choice, not a Meta algorithm requirement.

Example for HVAC replacement: “Repair or replace?” decision explainer; a walkthrough of an estimate; a truthful account of installation-day disruption; and a comparison of total ownership cost. Price claims must match the client’s actual offer and disclose material conditions. Use approved real proof; never fabricate reviews, before-and-after results, or customer stories.

Test offer and concept before button colors. Consolidate unnecessary campaign fragmentation, but retain meaningful differences in geography, economics, and regulated categories. Judge creative after the planned minimum spend and conversion-lag window, not a universal 5–10-day rule. Pause immediately for broken destinations, incorrect claims, or budget breaches.

Discovery after AI

Create service and location pages that answer real buying questions, show named expertise, and make prices or decision criteria understandable. Keep essential information in crawlable text, support it with original photos or demonstrations, and maintain accurate business listings. Google says ordinary SEO fundamentals apply to its AI features; there is no special AI schema or guaranteed inclusion.

Track qualified inquiries, referral sources, branded demand, and assisted discovery over time. Test a small, repeatable set of customer questions in AI products as qualitative research; do not sell an “AI citation score” as proof of revenue. Paid placements in AI products are an optional future experiment only when access, reporting, and economics are verified.

06 / MEASUREMENT

Optimize for the business
you actually want.

Primary scorecard: collected revenue from new customers, contribution after fulfillment and acquisition, all-in customer acquisition cost, qualified-lead cost, appointment-held rate, close rate, and payback time. Keep raw leads, ad clicks, and platform ROAS as diagnostic measures.

Measurement maturity gates
StageUse nowEvidence required to advance
FoundationEvent QA, CRM stages, campaign tags, sales receiptsTest submission and sale reconcile without duplicates; named sales owner
Outcome biddingOptimize to a meaningful, frequent qualified eventStable definitions, timely uploads, sufficient current platform volume
Value biddingActual contribution or calibrated expected contributionReliable differentiated values, adequate volume, lag checks, account eligibility
IncrementalityRandomized holdout or matched-market experimentPower analysis, pre-registered outcome/window, limited spillover, feasible cost
MMM / clean roomsSpecialist modeling or privacy-controlled joinsEnough independent variation/history, resolved identifiers, governance, a specific decision worth the cost

Do not add “lead,” “qualified,” and “sale” values as if they were three separate purchases. Choose the primary optimization outcome; use a documented value adjustment or stage strategy. Keep predicted value separate from realized value and recalibrate against mature cohorts. Early pLTV models are often less useful than a reliable service-margin table.

Report platform-attributed outcomes, CRM-observed outcomes, and experimentally estimated incremental outcomes separately. Exclude or isolate brand capture and existing-customer activity when evaluating new demand. A high match score or server-side event count is not proof of incremental sales.

Small pilots may be statistically underpowered for causal claims. Use them to validate operations and directional economics; state uncertainty. Do not prescribe an enterprise clean room merely because spend crosses an arbitrary threshold. No universal $879,000 setup cost is assumed in this plan.

07 / AGENCY ECONOMICS

Build a business
that can deliver its promise.

The following steady-state scenario is a planning model for ten clients, not a first-quarter forecast. Setup and diagnostic revenue are excluded. All delivery labor—including founder delivery—is costed to avoid overstating margin.

Illustrative monthly agency P&L · 10 clients
ItemAssumptionMonthly amount
Pilot retainers7 × $2,500$17,500
Growth retainers3 × $4,500$13,500
Service revenueExcludes client media spend$31,000
Delivery labor(7 × 18 hours + 3 × 28 hours) × $60($12,600)
Client-specific tools/data10 × $100($1,000)
Shared delivery softwareBudget assumption($1,200)
Gross contribution52.3% service margin$16,200
Founder management/salesExcludes costed delivery hours($5,000)
Agency acquisitionIllustrative monthly cap($1,500)
Administration/insuranceBudget assumption($1,000)
Operating surplusBefore tax, financing, and reserves$8,700

At this mix, contribution before shared software is approximately $1,740 per client. Shared software plus operating overhead totals $8,700, so roughly five similarly mixed clients cover the modeled recurring fixed costs. This simple break-even ignores ramp-up, churn, one-time costs, and cash collection timing.

Capacity: 210 monthly delivery hours require more than a solo founder. Plan two part-time specialists with a combined 240-hour monthly delivery capacity and 30 hours of buffer, plus separate founder management time. Measure actual hours by client weekly. Raise price or reduce scope if steady-state gross margin stays below 50%.

Launch reserve: budget $15,000 of agency operating cash as a proposed starting reserve for contracts/insurance ($2,000), website and sales material ($1,000), tools ($1,000), contractor delivery deposits ($5,000), acquisition test ($1,500), and contingency ($4,500). Founder living costs and client media are additional. Obtain quotes before spending; these figures are not vendor prices.

08 / FIRST 90 DAYS

Proof before expansion.

DAYS 01–15

Choose the niche. Set the baseline.

Complete 12 owner interviews; score margin, urgency, reachability, and capacity. Select one niche and offer. Finalize contracts, operating entity details, privacy/data roles, client access checklist, and package scope.

Exit gate: a clear ICP, sample client economics, named delivery owners, and a costed offer.
DAYS 16–30

Sell the diagnostic. Build the first system.

Research 40 accounts, deliver six tailored reviews, aim for two paid diagnostics, and sign one suitable pilot. Set up the client’s source-to-sale ledger and test one complete inquiry-to-receipt flow.

Exit gate: collected first fees and verified capture, routing, consent, and measurement.
DAYS 31–60

Run one controlled acquisition test.

Launch the agreed channel and four concept families. Review response handling daily, spend and lead quality weekly, and matured cohort economics monthly. Reconcile receipts before retrying failed actions.

Exit gate: reliable operations and enough matured outcomes to make a defensible continue/change/stop decision.
DAYS 61–90

Document what works. Replicate carefully.

Publish one client-approved case study only if supported by real evidence. Seek pilot renewal, improve the niche playbook, and aim for two to three retained clients in total. Add the next client only within delivery capacity.

Exit gate: contribution and service quality justify renewal; any expanded budget has written approval.

Two to three retained clients is an execution goal. If sales cycles are longer or early tests miss the economics, extend the evidence window or revise the offer instead of presenting a 90-day success claim.

09 / TEAM & OPERATIONS

Automation with
a responsible owner.

David: positioning, sales, client relationship, commercial approval, and quality of claims. Paid media/creative specialist: budgets, campaigns, concept production, weekly decisions. Data/automation specialist: CRM, consent, events, deduplication, integrations, and recovery. Fractional accountant and counsel: financial controls and contract/privacy review. These are proposed roles, not claims of staff already hired.

AI agents assist research, draft creative, classify leads, summarize results, and detect anomalies. Require evidence for factual enrichment and human review for new claims, spending changes, public case studies, and sensitive targeting. Do not give every agent every client credential.

  • Daily: review spend pacing, form receipt, routing, event freshness, and critical failures.
  • Weekly: reconcile sales stages, qualify feedback, inspect actual labor, and record one prioritized experiment decision.
  • Monthly: review collected contribution, channel overlap, renewals, pipeline, and client profitability.
  • Recovery: record a durable action receipt; use bounded retries; check provider state before replay. Isolate an affected workflow while healthy work continues.

Set per-client credential access, retention rules, deletion handling, encrypted transport, and access reviews. Hashing is not anonymity and does not remove consent or disclosure obligations. Do not send sensitive customer information into prompts or advertising systems simply because a connector allows it.

Outreach controls: central suppression; provider terms reviewed; no automated calls or texts without the required recorded permission. For housing, financial services, employment, or other restricted categories, verify current platform settings and legal requirements before activation.

10 / SCALE DECISIONS

Let evidence
set the pace.

Expand when

Lead capture and sales reconciliation are reliable; mature cohorts support the target contribution; delivery margin is at least 50%; the client can handle more demand; and the next spend increment is approved. Increase budgets in measured steps and evaluate marginal economics.

Repair or stop when

Routing breaks, spending exceeds its cap, data becomes unreliable, claims are wrong, fulfillment deteriorates, or the pre-agreed economic threshold fails after the measurement window. Fix the affected system before buying more traffic.

Months 4–12

At three retained clients, standardize the niche playbook and measure retention. At five, formalize account ownership and weekly capacity planning. Approach ten only after staffing, cash reserves, and renewal quality support it. Add a second niche or channel when the first has repeatable economics—not simply because AI can produce more content.

The operating promise: every client receives a short decision memo: what was spent, what happened, what is verified, what remains uncertain, and what will change next.

Discuss the plan