You do not have a traffic problem

Almost every service business we audit is buying enough clicks. What they are not buying is enough qualified, contactable, in-territory demand. Between 20% and 45% of paid lead volume in the accounts we inherit is junk: bots, spam form fills, wrong-service inquiries, out-of-area requests, and duplicate submissions already in the CRM.

That junk does three kinds of damage. It burns budget directly, it burns CSR and estimator hours indirectly, and — most expensively — it corrupts the data your bidding algorithms optimize against. Google and Meta will happily buy you more of whatever you tell them a conversion is. If your conversion definition includes garbage, you are paying a platform to manufacture garbage at scale.

Fix the front end before you touch bids. Budget efficiency gains from clean conversion data routinely exceed anything you get from campaign restructuring.

The 10 front-end fixes, in build order

1. 1. Kill the generic contact form — Replace one open-text form with service-specific intake. Ask for service type, property type, and ZIP before name and email. Unqualified traffic self-selects out.
2. 2. Enforce real-time email verification — Validate at submit, not after. Syntax, domain MX, disposable-domain and catch-all detection. Reject hard bounces before they enter the CRM.
3. 3. Enforce phone validation — Line-type lookup: reject VoIP-only and invalid numbers, flag landlines for commercial, require a valid mobile for text-based follow-up.
4. 4. Add a honeypot plus timing gate — A hidden field no human sees, plus a minimum time-on-form threshold. This alone removes the majority of unsophisticated bot submissions with zero UX cost.
5. 5. Layer invisible bot scoring — reCAPTCHA v3 or Turnstile in score mode. Never a visual challenge on a paid landing page — challenges cost you real conversions.
6. 6. Geo-fence at the form level — ZIP or postal validation against your actual service map. Show an honest out-of-area message rather than capturing a lead you cannot serve.
7. 7. Dedupe against the CRM at submit — Match on phone and email. Existing customers get routed to service, not to new-customer acquisition reporting.
8. 8. Separate conversion actions by value — A booked appointment, a verified form, and a 60-second-plus call are three different conversion events with three different values. Feed values, not counts.
9. 9. Instrument call tracking properly — Dynamic number insertion by source, call recording, and duration thresholds. Sub-30-second calls should never count as conversions.
10. 10. Send offline conversions back — Push booked and closed revenue back into the ad platforms. This is the single highest-leverage change most accounts have never made.

Conversion values that reflect your economics

Set values from your own numbers, not the platform defaults. Start with average job value, multiply by the historical close rate for that source and service, then subtract fulfillment cost if margins vary widely by service line.

Weekly operating cadence

• Monday: review last week's leads flagged as junk; add negatives, block placements, tighten geo.
• Monday: pull close rate by campaign and service line, not by channel average.
• Mid-week: review call recordings for the 10 longest and 10 shortest calls from paid.
• Month-end: reconcile ad spend to closed revenue in the CRM; recompute conversion values.

What good looks like after 90 days

• Junk lead rate under 8% of total paid volume.
• Cost per booked appointment reported weekly and trending down.
• Every campaign tied to closed revenue, not lead count.
• Bidding optimized against offline conversions imported from the CRM.

FAQs

Why does my Google Ads account keep getting more junk leads?

Because the conversion signal is dirty. If you count every form fill and every call as a conversion, the algorithm learns to find more of them — regardless of quality. Switching to verified conversions, offline conversions, and value-based bidding fixes the root cause.

How much budget can these front-end fixes recover?

In accounts we inherit, the wasted-spend recovery is typically 15–35% of the paid budget, with a bigger downstream effect on CSR and estimator productivity. The exact number depends on how much of your current lead volume is unqualified.

Should we stop paid ads while we fix the front end?

No. Keep the ads running, but cap spend while you install verification, deduplication, and conversion values. Once the signal is clean, you can re-accelerate with confidence that the algorithm is learning the right thing.

Read the full article on Penrose Growth

Commercial demand is created, not captured

Residential demand is captured. Commercial demand is created. Property managers, facility directors, and general contractors already have an incumbent vendor; they are not searching. They switch when the incumbent fails, when a portfolio changes hands, or when a credible alternative is standing there at the moment of a failure.

Outbound's job is to be the credible alternative already known at the moment of failure. That reframes the goal: you are not closing on contact one, you are earning a place on a short list you will be pulled from six to eighteen months out.

Build the list at the building level, not the company level

Most outbound fails because the list is a company list. Commercial service work is bought per property. Build a building-level target list and attach the decision chain to each address.

• Source: county assessor records, CoStar or Reonomy, permit filings, chamber and BOMA directories.
• Qualify by square footage, asset class, building age, and roof or system age where relevant.
• Attach: owner entity, property management firm, on-site engineer, regional facility lead.
• Trigger events: ownership change, permit filed, new tenant, capital project announced, negative online review of the incumbent.

Sequence structure that gets replies

1. Touch 1 — Property-specific email — Reference the actual building. One sentence of relevance, one sentence of proof at a comparable property, one low-friction ask (a 15-minute walkthrough, not a meeting).
2. Touch 2 — Call, day 2 — Do not pitch. Confirm you have the right person for that asset and ask who handles emergency dispatch after hours.
3. Touch 3 — Value asset, day 5 — Send something they keep: a system life-expectancy table, a code-compliance checklist, a capital-planning template for their asset class.
4. Touch 4 — Call + voicemail, day 8 — Reference the asset you sent. Voicemails should end with a specific date offer, not 'call me back.'
5. Touch 5 — LinkedIn touch, day 10 — Connect and comment before messaging. Facility and PM communities are small and reputation-driven.
6. Touch 6 — Break-up + standing offer, day 14 — Explicitly offer to be the backup vendor for emergencies. This converts far more often than any pitch in the sequence.
7. Nurture — quarterly — Move non-responders to a quarterly value cadence. Most commercial wins in our programs come from touch 9 through 20, not the initial sequence.

Territory and coverage model

Pipeline math: work backward from booked walkthroughs

Use realistic conversion rates and staff to the goal rather than to activity targets pulled from a SaaS blog.

• 1,000 building-level contacts touched → 60–90 conversations (6–9%).
• 60–90 conversations → 18–30 walkthroughs (30%).
• 18–30 walkthroughs → 8–14 proposals.
• 8–14 proposals → 3–5 wins, weighted toward months 4–9.

Judge outbound on walkthroughs booked and short-list placements in months 1–3, not revenue. Revenue judgments before month six kill programs that were working.

FAQs

How long does commercial outbound take to produce revenue?

Most wins come from touch 9 through 20, with revenue typically showing in months 4–9. The first 90 days should be measured on walkthroughs booked and short-list placements, not closed deals.

What is the right list size for a single commercial rep?

For a geographic hunter, 150–250 building-level targets. For vertical hunters, 100–200 accounts. For a farmer model focused on property management firms, 8–20 portfolio relationships.

Should we use AI to write outbound emails?

Use AI for list building and research, not for final cold emails. Buyers in the trades identify fully AI-written emails instantly; reply rates fall by roughly half. Keep human review on every first touch.

Read the full article on Penrose Growth

Where AI wins and where it loses

What kills reply rates

• Fully AI-written emails at volume. Buyers in the trades identify them instantly; reply rates in our tracked accounts fall by roughly half.
• Sending from the primary domain. Always use dedicated sending domains with matching branding.
• Volume without warm-up. New domains need 4–6 weeks of ramp; cap at 30–50 sends per mailbox per day.
• Missing authentication. SPF, DKIM, and DMARC are non-negotiable in 2026 — bulk senders without them are filtered by default.

Deliverability checklist

1. Domains — 2–4 dedicated domains, each with 2–3 mailboxes. Never send outbound from the domain your customers email.
2. Authentication — SPF, DKIM, DMARC with a policy of at least quarantine. Verify with a header check monthly.
3. Warm-up — Automated warm-up for 4–6 weeks before real sends; keep warm-up running underneath live campaigns.
4. List hygiene — Verify every address before send. Keep bounce rate under 2% and spam complaints under 0.1%.
5. Content — Plain text, no tracking pixels on cold touches, no images, at most one link after the first reply.

Benchmarks from active service-business programs

FAQs

Can AI completely replace outbound salespeople?

No. AI is excellent at list building, signal monitoring, and administrative follow-up. Cold calls, objection handling, and relationship building remain human. The best teams use AI to make each human touch more informed, not to replace it.

How many emails should a rep send per day?

Cap at 30–50 sends per mailbox per day on a properly warmed dedicated domain. Sending from the primary domain or blasting volume from a fresh domain destroys deliverability.

Why are our open rates falling even though the list is clean?

Usually authentication or sending-domain health. Check SPF, DKIM, and DMARC with a header analyzer. Then verify you are not sending from a domain that also handles customer service and billing.

Read the full article on Penrose Growth

Sequencing matters more than tooling

Automation built on an undefined process encodes the mess. Before any build, the process must be written down, owned by a named person, and stable for at least 30 days. Then automate — starting where the return is measurable in booked revenue rather than in hours saved.

Build order

1. Phase 1 — Speed to lead — Auto-response within 60 seconds by text and email, round-robin dispatch to the first available CSR, and escalation if unclaimed in 5 minutes. Response inside 5 minutes routinely doubles contact rates versus 30 minutes.
2. Phase 2 — After-hours intake — AI voice or chat intake that captures service type, address, urgency, and contact, then books into open capacity or escalates true emergencies. Target: zero missed calls.
3. Phase 3 — Estimate follow-up — Automated cadence on every open estimate: day 1, day 3, day 7, day 14, day 30, then quarterly. This is the highest-ROI automation in most service businesses because the demand is already qualified.
4. Phase 4 — Review and reputation — Trigger a review request on job completion plus payment, routed to the platform where you are weakest, with negative-sentiment interception before it posts.
5. Phase 5 — Reporting — Nightly sync from CRM and ad platforms into one dashboard: leads by source, booked rate, close rate, average ticket, revenue by service line.
6. Phase 6 — Back office — Invoice chasing, technician debrief summaries, permit and warranty tracking, and recurring-maintenance renewal reminders.

Roadmap by company size

Guardrails for AI touching customers

• Never let AI quote price or commit to a scope.
• Always disclose when a customer is speaking with an assistant.
• Hard escalation triggers: safety, water or gas, anger, insurance, warranty.
• Log and review 10 interactions weekly for the first 90 days.

FAQs

What is the first automation a small service business should build?

Speed to lead. A sub-60-second text and email response, routed to the first available CSR, routinely doubles contact rates. It is cheap to build and pays for itself immediately.

Should we use AI voice for after-hours calls?

Yes, once the intake process is documented. The AI should capture service type, address, urgency, and contact, then book into open capacity or escalate emergencies. Never let it quote price or dispatch without guardrails.

At what size does a reporting dashboard become worth it?

Once you have more than one service line or more than one location, a single source of truth becomes essential. Below $5M, the CRM and ad platform reports are usually enough if they are reviewed weekly.

Read the full article on Penrose Growth

The salesperson scorecard

• Outcome 1: hit $X in sold revenue at or above Y% gross margin by month 4.
• Outcome 2: maintain a close rate at or above the team median on scored, comparable leads.
• Outcome 3: same-day documentation of every call in the CRM, no exceptions.
• Competencies: coachability, follow-through, comfort with silence, technical curiosity, resilience after a no.
• Disqualifiers: blames lead quality in the interview, cannot explain a lost deal, will not role-play.

Interview process

1. Screen (20 min) — Comp expectations, ticket size history, and one question: walk me through your last three losses.
2. Structured interview (60 min) — Chronological work history with numbers at every stop. Ask for quota, attainment, and manager name each time.
3. Live role-play (45 min) — Real scenario with a real objection. Score on discovery depth, options presented, and how they handle 'I need to think about it.'
4. Ride-along — Half a day in the field with your best rep. The candidate's questions during that ride tell you more than the interview.
5. References — Two former managers, not peers. Ask: would you hire them again, and at what number?

30/60/90 ramp

Comp that rewards margin

Volume-only commission trains discounting. Pay on gross profit, with accelerators above a margin threshold and a clawback on cancellations and unpaid invoices.

• Base at 35–45% of on-target earnings for in-home residential; 45–55% for commercial.
• Commission on gross profit, tiered: below target margin at half rate, at target at full rate, above target at 1.25–1.5x.
• Monthly payout, quarterly true-up against collections.
• Clawback on any job cancelled or uncollected within 60 days.

The call framework

1. Frame — Set the agenda and the decision point at the door. 'By the end I'll show you options and you can tell me yes, no, or not yet.'
2. Diagnose — Longer than feels comfortable. Symptoms, history, what they've already tried, who else is involved, timeline, and budget reality.
3. Teach — Explain the cause, not just the fix. Show the customer something about their property they did not know.
4. Present options — Good, better, best — always three, always priced, always with the difference stated in outcomes rather than parts.
5. Ask — Direct, once, then silence. 'Which of these do you want to move forward with?'
6. Handle and close — Isolate the objection, confirm it's the only one, resolve it, re-ask. Never leave without a scheduled next step.

FAQs

How do you avoid hiring discount-happy salespeople?

Pay on gross profit, not revenue. Add clawbacks for cancellations and uncollected invoices. During the interview, ask the candidate to walk through their last three losses and listen for whether they blame price, lead quality, or themselves.

What is a realistic ramp timeline for a new salesperson?

Day 30: pass pricing and scope exam. Day 60: 50% of team median close rate. Day 90: 80% of median and at target margin. Day 120: independent and at or above median.

Why does the in-home call framework start with a frame?

Because the biggest cause of no-decision is ambiguity. Setting the agenda and the decision point upfront removes the 'let me think about it' escape route and lets the customer focus on choosing the right option.

Read the full article on Penrose Growth

Estimator variance is the silent margin killer

In most service businesses, close rate spread between the best and worst estimator is 20 to 30 points, and margin spread is 10 to 15 points, on identical work. That variance is not talent; it is the absence of a standard. Standardization does not remove judgment — it removes the parts of the estimate that should never have been improvised.

Standardized scope template

• Site conditions: access, staging, protection, disposal, permits.
• Scope of work in customer language, then in trade language.
• Explicit exclusions — the single biggest source of margin loss.
• Assumptions and unknowns with a stated allowance.
• Schedule, crew size, and duration.
• Warranty terms and what voids them.

Pricing structure

Good / better / best that actually works

• Best is presented first and anchors the range.
• The difference between tiers must be an outcome — lifespan, efficiency, downtime, warranty — never just a brand.
• Middle option should be designed to win 50–60% of the time.
• Every tier priced on the page. Never 'starting at.'

Follow-up cadence on open estimates

1. Hour 1 — Written estimate delivered while still on site or same day. Delay past 24 hours costs roughly a third of the close rate.
2. Day 2 — Call, confirm received, answer the one question they thought of after you left.
3. Day 5 — Value touch — photo, spec sheet, or a comparable job.
4. Day 10 — Direct ask with a scheduling incentive tied to crew availability, not a discount.
5. Day 21 and quarterly — Move to nurture. Aged estimates re-close at 5–12% with a systematic cadence and near zero without one.

Change-order discipline

• No verbal change orders. Ever.
• Signed and priced before work proceeds, including on time-and-material.
• Change orders carry the same or higher margin than base scope.
• Weekly review of jobs with more than two change orders — that is a scoping problem, not a customer problem.

FAQs

How do you close the gap between best and worst estimators?

Standardize scope, pricing, and presentation so every estimator starts from the same foundation. Then measure close rate, margin, and follow-up compliance by individual. Variance that remains after standardization is coaching and talent, not process.

Why do 'good, better, best' options increase close rate?

They shift the conversation from whether to buy to which option to buy. The best option anchors the range, the middle option is designed to win, and the good option provides a clear budget fallback. Customers choose more confidently when the decision is framed this way.

What is the biggest source of margin loss in estimates?

Exclusions that were not written down and verbal change orders that were never priced. Standardize exclusions in every estimate and require written, signed change orders before work proceeds.

Read the full article on Penrose Growth

The core metric set

Adjust for lead quality or you will lose your best reps

Raw close rate punishes whoever gets the hardest leads. Score every lead at creation — before assignment — on source, service type, urgency, and property or contract value. Then compare each rep against the expected close rate for their actual lead mix.

• Score leads A/B/C at creation, never retroactively.
• Report close rate by tier as well as blended.
• Rotate lead assignment so mixes are comparable across reps.
• Track and publish assignment mix monthly to keep the system honest.

Review cadence

1. Weekly, 30 minutes, one-on-one — Three numbers, two call recordings, one commitment for the coming week. Never a pipeline read-out.
2. Weekly, team, 20 minutes — Leaderboard on margin and close rate, one win teardown, one loss teardown.
3. Monthly — Full scorecard against target, ramp status for new reps, comp reconciliation.
4. Quarterly — Territory and lead-mix rebalance, target reset, and formal performance conversation.

Rollout plan

• Weeks 1–2: instrument the CRM so every metric is captured automatically. Manual scorecards die.
• Weeks 3–4: publish baselines with no consequences attached. Let reps challenge the data.
• Weeks 5–8: start weekly one-on-ones on the new format.
• Week 9+: tie comp accelerators to margin and follow-up compliance.

If a number cannot be pulled without asking a person, it will not survive month three. Automate capture first.

FAQs

Should we fire our lowest close-rate rep?

Not before you adjust for lead quality. A rep with a 30% close rate on C-grade commercial leads may be outperforming a rep with 45% on A-grade residential leads. Score leads before assignment, then compare expected vs. actual close rate.

What is the most important weekly sales metric?

Gross margin per presented estimate combined with follow-up compliance. Revenue without margin is a loss, and even the best presenter fails if open estimates are not nurtured.

How do you make salespeople trust the numbers?

Let them challenge the data for the first four weeks with no consequences. Fix the CRM capture, then publish baselines. Once reps agree the data is accurate, they will compete against it.

Read the full article on Penrose Growth

The chain

Four handoffs, each of which breaks attribution if unmanaged: click to lead, lead to CRM record, CRM record to estimate, estimate to collected revenue. Most service businesses instrument the first and guess at the rest, which is why marketing debates are argued with opinions.

1. Click → Lead — GCLID/WBRAID/FBCLID captured in a hidden form field and on dynamic call tracking numbers. Persist UTMs in a first-party cookie for at least 90 days.
2. Lead → CRM — Every lead enters the CRM with source, campaign, and click ID populated automatically. No free-text 'how did you hear about us' as a source of truth.
3. CRM → Estimate — Estimate record inherits the lead's attribution fields. Never let a re-created record orphan the source.
4. Estimate → Revenue — Invoiced and collected amounts write back to the same record. Report on collected, not sold, for channel decisions.

Call attribution done properly

• Dynamic number insertion with a pool sized for concurrent sessions, not total sessions.
• Separate static numbers for GBP, direct mail, vehicle wraps, and yard signs.
• Duration threshold of 60 seconds minimum before a call counts.
• Automated call outcome tagging: booked, quoted, existing customer, spam, wrong number.

Reporting views

Feed it back to the platforms

Offline conversion import is the payoff. Once booked and closed revenue flows back into Google and Meta, bidding optimizes toward customers rather than form fills. Expect a 4–8 week relearning period and a meaningful improvement in cost per booked job after it.

Rule: no channel decision gets made on lead volume. Only on cost per booked job and closed revenue.

FAQs

What is the most common attribution mistake in service businesses?

Reporting on leads instead of booked jobs or closed revenue. A channel with a low cost per lead but a low close rate is more expensive than a channel with a high cost per lead and a high close rate.

Do we need a new CRM to track attribution?

Usually not. ServiceTitan, Housecall Pro, and most modern CRMs can hold custom fields and estimates. The fix is usually process and field mapping, not a new platform.

How long until offline conversion import improves performance?

Expect 4–8 weeks of relearning once the platform receives booked and closed signals. The improvement is meaningful but not instant; the algorithm needs time to learn which clicks produce customers.

Read the full article on Penrose Growth

A single-location painting company with strong craft and repeat clients but no repeatable playbook for opening new markets, no attributed demand system, and no infrastructure to maintain quality at distance.

Growth

Started Exquisite Painterz from a single location and expanded into New York, Dallas, and South Florida — building a multi-market luxury painting brand.

Operations

Built the systems, pricing, and project management infrastructure so a high-end painting business could replicate its quality and sales motion across three markets.

Biggest Wins

• Built the brand and demand engine from one location into New York, Dallas, and South Florida.
• Created local authority, project galleries, and review velocity in each new market from a standing start.
• Implemented AI-assisted estimating and project intake so each location could quote and schedule consistently.
• Developed a designer and builder partnership channel to feed high-margin commercial and luxury residential work.
• Established a repeatable market-launch playbook that can be redeployed in future expansion.

Channels Built

• Paid search built around luxury residential, commercial, and designer-builder intent
• Local and organic search — new market pages, project galleries, and review velocity
• Paid social and portfolio retargeting for high-end residential and commercial projects
• Designer, builder, and contractor partnership channel
• Outbound sales and relationship nurture for commercial and property accounts
• Referral and repeat-client engine

Outcomes

• Expanded from one location to three major markets: New York, Dallas, and South Florida.
• Multiple independently producing lead and referral channels across residential, commercial, and designer-builder segments.
• A scalable operating backbone that preserves quality as the company grows geographically.

Results

• 1→3 — Markets
• NY · TX · FL — Expansion
• +5 — Lead channels

Penrose is not like any marketing or consulting firm we have ever worked with. They took us from a single-location painting business and built the brand, demand, and operating systems needed to expand into New York, Dallas, and South Florida. They did not just run ads — they created local authority, project galleries, and review velocity in each market; built designer and builder partnerships that feed high-margin work; and used AI to make estimating and intake consistent across every location. They operate like an internal growth and operations team, not a vendor, and they gave us a playbook we can use to keep expanding.

Read the full case study

Commercial projects were won or lost inside the owner's inbox. Paid spend was generating proposals, but no one knew which campaigns produced signed contracts — or where deals were dying in follow-up.

Growth

Scaled from a single-market design-build firm to a multi-market operation with four profitable revenue channels.

Operations

Removed the owner from daily estimating and business development by building AI-assisted bidding, intake systems, and a recruited back-office.

Biggest Wins

• Built attribution from click to signed contract across every channel — call tracking, CRM stages, and closed-won contracts tied back to campaign and keyword.
• Ran a true multi-channel revenue engine — paid search, paid social, outbound sales, automated direct mail, gifting, referral, and partnerships — into one tracked pipeline.
• Built an outbound sales team with playbooks for following up on proposals, target accounts, and open opportunities.
• Automated outbound based on real triggers: permits, project starts, property manager turnover, and new construction activity.
• Compressed bid turnaround from days to same-week with AI-assisted estimating and enforced follow-up.
• Brought in recruiting, training, and EOS partners to staff and run each new market on the same playbook.

Channels Built

• Paid search rebuilt around GCs, owners, and property groups
• Paid social retargeting on completed commercial projects
• Dedicated inside sales reps following up on proposals and target-account outreach
• Automated outbound sequences triggered by permit, project, and property data
• Direct mailers, leave-behinds, and gifting for high-value GCs and property managers
• Referral and strategic partnership channel
• Local and organic search visibility per market

Outcomes

• 5x revenue growth across the portfolio.
• Single market expanded into multiple markets on one shared operating system.
• Owner removed from daily estimating and one-to-one BD.

Results

• 5x — Revenue growth
• 3+ — Markets expanded
• +4 — Revenue channels

Penrose isn’t an agency and they aren’t consultants — they’re an operating partner embedded inside our business. They rebuilt the entire revenue engine from the click all the way through to the signed contract, then went further and fixed the parts nobody else would touch: our estimating, our follow-up, our bid turnaround, and the way we chase target accounts. They brought AI into our bidding so proposals go out the same week instead of sitting for days, and they stood up an outbound sales team, trigger-based outreach, and direct mail that actually reaches the GCs and property groups we want. When we opened new markets, they handed us a playbook and the recruiting and training partners to run it. I have never seen another firm operate this way — they act like owners, not vendors. — Joe Belleli, Owner

Read the full case study

Key Takeaways

• Role, incentive, and reporting changes that reduce key-person risk
• What acquirers actually diligence in a services platform
• A 12-month sequence for making the business transferable

Find the dependency

• Which pricing decisions above a threshold require the owner?
• Which customers would call the owner directly if something went wrong?
• Who signs, who hires, who fires, who approves spend — by dollar band?
• What is not written down: pricing logic, key vendor terms, warranty policy, escalation paths?
• If the owner were unreachable for 30 days, what stops?

What acquirers actually diligence

12-month transfer sequence

1. Months 1–3 — Document decision rights and pricing logic. Move to accrual accounting with job-level costing.
2. Months 4–6 — Hire or promote a GM and a sales leader. Owner steps out of dispatch and pricing approvals under the threshold.
3. Months 7–9 — Install the weekly operating cadence — one scorecard, one meeting, named owners on every number.
4. Months 10–12 — Owner takes a deliberate 30-day absence. Whatever breaks is the remaining diligence risk; fix it before going to market.

Transferability is the multiple. Two businesses with identical EBITDA can trade three turns apart based on how much of the operation lives in the owner's head.

Access the full checklist on Penrose Growth

Key Takeaways

• Stage definitions and exit criteria that survive real-world use
• The KPI set that predicts next quarter's revenue
• Common ServiceTitan and HubSpot configuration mistakes

Stages with exit criteria

Data model rules

• One record per opportunity, never per conversation.
• Source fields write-once and system-populated; no rep-editable source.
• Loss reasons from a fixed list of 6–8 — free text makes loss analysis impossible.
• Every property is an object with history, so repeat and referral revenue is visible.

The KPI set that predicts next quarter

• Estimates delivered per week (leading indicator, 30–60 day lag to revenue).
• Average sold ticket trend.
• Close rate by lead tier.
• Open estimate value with an active next step.
• Backlog in weeks of crew capacity.

Common configuration mistakes

• ServiceTitan: job types not mapped to business units, so revenue by service line is unreportable.
• ServiceTitan: campaigns created per ad rather than per channel, fragmenting attribution.
• HubSpot: using deals for both service calls and projects with one pipeline.
• Any CRM: required fields added without training — reps enter junk and the reporting layer dies quietly.
• Any CRM: reporting built on 'created date' rather than cohort, which flatters recent months.

Access the full blueprint on Penrose Growth

Key Takeaways

• What healthy blended CAC looks like by trade and ticket size
• When local lead platforms outperform search, and when they destroy margin
• A simple channel profitability model you can run monthly

Benchmarks by trade

Ranges reflect metro competitiveness. Treat the low end as dense-suburban with strong organic presence and the high end as major metro with heavy PE-backed competition.

Channel selection

• Local Services Ads: best cost per booked job for emergency and small-ticket residential; caps out on volume.
• Search: highest intent, best for replacement and high-ticket; requires strong landing pages and negatives.
• Performance Max: use only with clean conversion values and brand exclusions, otherwise it cannibalizes brand.
• Meta: works for replacement, financing offers, and commercial retargeting — not for emergency demand.
• Lead aggregators: acceptable as capacity fill, dangerous as a primary channel; margin is structurally lower and the customer is not yours.

The monthly channel profitability model

1. Pull — Spend, leads, booked jobs, sold revenue, and gross profit by channel.
2. Compute — Cost per booked job, CAC as a share of gross profit, and contribution margin after CAC.
3. Decide — Scale any channel where CAC is under 25% of gross profit and capacity exists. Fix or cut anything above 40%.
4. Recheck — Re-run after the ad platforms' learning periods, not during them.

Access the full benchmark report on Penrose Growth

Key Takeaways

• Scorecard for grading each stage of the funnel against live benchmarks
• The five most common places service businesses lose booked revenue
• How to quantify the dollar value of each leak before spending on more leads

Grade each stage before buying more leads

The five most common leaks

• Unanswered and abandoned calls — usually 8–15% of inbound, almost always at lunch, after 5pm, and Saturdays.
• Slow first response — anything past 30 minutes halves contact rate.
• No structured follow-up on open estimates — the single largest dollar leak in most businesses.
• Unqualified or out-of-area leads consuming estimator capacity.
• Existing customers routed through new-customer flows, inflating CAC and depressing measured close rates.

Quantify the leak in dollars

Work the arithmetic before spending another dollar on demand. Example: 800 inbound leads per year, 12% unanswered = 96 lost inquiries. At a 55% book rate, 90% show, and 45% close, that is roughly 21 lost jobs. At a $9,500 average ticket, the answer-rate leak alone is about $200,000 in lost revenue per year — recoverable for the cost of overflow answering.

Run this arithmetic on every stage. Fix the largest dollar leak first, not the easiest one.

90-day remediation sequence

1. Weeks 1–2 — Instrument: call tracking, CRM stage definitions, and a single source-of-truth report.
2. Weeks 3–4 — Fix answer rate — overflow answering, after-hours intake, and call-back SLA.
3. Weeks 5–8 — Fix speed to lead and install the estimate follow-up cadence.
4. Weeks 9–12 — Fix close rate — options presentation, role-play cadence, and margin coaching.

Access the full diagnostic on Penrose Growth

Key Takeaways

• Tracking architecture across ads, CRM, and phone
• Lead source integrity and call attribution
• Estimate and job-level revenue matching
• Reporting views for owners and marketing partners

The chain

Four handoffs, each of which breaks attribution if unmanaged: click to lead, lead to CRM record, CRM record to estimate, estimate to collected revenue. Most service businesses instrument the first and guess at the rest, which is why marketing debates are argued with opinions.

1. Click → Lead — GCLID/WBRAID/FBCLID captured in a hidden form field and on dynamic call tracking numbers. Persist UTMs in a first-party cookie for at least 90 days.
2. Lead → CRM — Every lead enters the CRM with source, campaign, and click ID populated automatically. No free-text 'how did you hear about us' as a source of truth.
3. CRM → Estimate — Estimate record inherits the lead's attribution fields. Never let a re-created record orphan the source.
4. Estimate → Revenue — Invoiced and collected amounts write back to the same record. Report on collected, not sold, for channel decisions.

Call attribution done properly

• Dynamic number insertion with a pool sized for concurrent sessions, not total sessions.
• Separate static numbers for GBP, direct mail, vehicle wraps, and yard signs.
• Duration threshold of 60 seconds minimum before a call counts.
• Automated call outcome tagging: booked, quoted, existing customer, spam, wrong number.

Reporting views

Feed it back to the platforms

Offline conversion import is the payoff. Once booked and closed revenue flows back into Google and Meta, bidding optimizes toward customers rather than form fills. Expect a 4–8 week relearning period and a meaningful improvement in cost per booked job after it.

Rule: no channel decision gets made on lead volume. Only on cost per booked job and closed revenue.

Access the full framework on Penrose Growth

Key Takeaways

• Core metrics: close rate, average ticket, margin, speed
• Lead-quality adjustment so reps aren't punished for bad leads
• Weekly and monthly review formats
• Scorecard template and rollout plan

The core metric set

Adjust for lead quality or you will lose your best reps

Raw close rate punishes whoever gets the hardest leads. Score every lead at creation — before assignment — on source, service type, urgency, and property or contract value. Then compare each rep against the expected close rate for their actual lead mix.

• Score leads A/B/C at creation, never retroactively.
• Report close rate by tier as well as blended.
• Rotate lead assignment so mixes are comparable across reps.
• Track and publish assignment mix monthly to keep the system honest.

Review cadence

1. Weekly, 30 minutes, one-on-one — Three numbers, two call recordings, one commitment for the coming week. Never a pipeline read-out.
2. Weekly, team, 20 minutes — Leaderboard on margin and close rate, one win teardown, one loss teardown.
3. Monthly — Full scorecard against target, ramp status for new reps, comp reconciliation.
4. Quarterly — Territory and lead-mix rebalance, target reset, and formal performance conversation.

Rollout plan

• Weeks 1–2: instrument the CRM so every metric is captured automatically. Manual scorecards die.
• Weeks 3–4: publish baselines with no consequences attached. Let reps challenge the data.
• Weeks 5–8: start weekly one-on-ones on the new format.
• Week 9+: tie comp accelerators to margin and follow-up compliance.

If a number cannot be pulled without asking a person, it will not survive month three. Automate capture first.

Access the full framework on Penrose Growth

Key Takeaways

• Scorecard and interview process
• 30/60/90 ramp plan
• Comp plans that reward margin, not just volume
• In-home and on-site call framework
• Standardized scoping and pricing structure
• Good/better/best presentation format
• Estimate follow-up cadence
• Margin protection and change-order discipline

The salesperson scorecard

• Outcome 1: hit $X in sold revenue at or above Y% gross margin by month 4.
• Outcome 2: maintain a close rate at or above the team median on scored, comparable leads.
• Outcome 3: same-day documentation of every call in the CRM, no exceptions.
• Competencies: coachability, follow-through, comfort with silence, technical curiosity, resilience after a no.
• Disqualifiers: blames lead quality in the interview, cannot explain a lost deal, will not role-play.

Interview process

1. Screen (20 min) — Comp expectations, ticket size history, and one question: walk me through your last three losses.
2. Structured interview (60 min) — Chronological work history with numbers at every stop. Ask for quota, attainment, and manager name each time.
3. Live role-play (45 min) — Real scenario with a real objection. Score on discovery depth, options presented, and how they handle 'I need to think about it.'
4. Ride-along — Half a day in the field with your best rep. The candidate's questions during that ride tell you more than the interview.
5. References — Two former managers, not peers. Ask: would you hire them again, and at what number?

30/60/90 ramp

Comp that rewards margin

Volume-only commission trains discounting. Pay on gross profit, with accelerators above a margin threshold and a clawback on cancellations and unpaid invoices.

• Base at 35–45% of on-target earnings for in-home residential; 45–55% for commercial.
• Commission on gross profit, tiered: below target margin at half rate, at target at full rate, above target at 1.25–1.5x.
• Monthly payout, quarterly true-up against collections.
• Clawback on any job cancelled or uncollected within 60 days.

The call framework

1. Frame — Set the agenda and the decision point at the door. 'By the end I'll show you options and you can tell me yes, no, or not yet.'
2. Diagnose — Longer than feels comfortable. Symptoms, history, what they've already tried, who else is involved, timeline, and budget reality.
3. Teach — Explain the cause, not just the fix. Show the customer something about their property they did not know.
4. Present options — Good, better, best — always three, always priced, always with the difference stated in outcomes rather than parts.
5. Ask — Direct, once, then silence. 'Which of these do you want to move forward with?'
6. Handle and close — Isolate the objection, confirm it's the only one, resolve it, re-ask. Never leave without a scheduled next step.

Estimator variance is the silent margin killer

In most service businesses, close rate spread between the best and worst estimator is 20 to 30 points, and margin spread is 10 to 15 points, on identical work. That variance is not talent; it is the absence of a standard. Standardization does not remove judgment — it removes the parts of the estimate that should never have been improvised.

Standardized scope template

• Site conditions: access, staging, protection, disposal, permits.
• Scope of work in customer language, then in trade language.
• Explicit exclusions — the single biggest source of margin loss.
• Assumptions and unknowns with a stated allowance.
• Schedule, crew size, and duration.
• Warranty terms and what voids them.

Pricing structure

Good / better / best that actually works

• Best is presented first and anchors the range.
• The difference between tiers must be an outcome — lifespan, efficiency, downtime, warranty — never just a brand.
• Middle option should be designed to win 50–60% of the time.
• Every tier priced on the page. Never 'starting at.'

Follow-up cadence on open estimates

1. Hour 1 — Written estimate delivered while still on site or same day. Delay past 24 hours costs roughly a third of the close rate.
2. Day 2 — Call, confirm received, answer the one question they thought of after you left.
3. Day 5 — Value touch — photo, spec sheet, or a comparable job.
4. Day 10 — Direct ask with a scheduling incentive tied to crew availability, not a discount.
5. Day 21 and quarterly — Move to nurture. Aged estimates re-close at 5–12% with a systematic cadence and near zero without one.

Change-order discipline

• No verbal change orders. Ever.
• Signed and priced before work proceeds, including on time-and-material.
• Change orders carry the same or higher margin than base scope.
• Weekly review of jobs with more than two change orders — that is a scoping problem, not a customer problem.

Access the full guidebook on Penrose Growth

Key Takeaways

• Speed-to-lead and after-hours intake automation
• Estimate follow-up and re-engagement flows
• Back-office and reporting automations
• Build-order roadmap by company size

Sequencing matters more than tooling

Automation built on an undefined process encodes the mess. Before any build, the process must be written down, owned by a named person, and stable for at least 30 days. Then automate — starting where the return is measurable in booked revenue rather than in hours saved.

Build order

1. Phase 1 — Speed to lead — Auto-response within 60 seconds by text and email, round-robin dispatch to the first available CSR, and escalation if unclaimed in 5 minutes. Response inside 5 minutes routinely doubles contact rates versus 30 minutes.
2. Phase 2 — After-hours intake — AI voice or chat intake that captures service type, address, urgency, and contact, then books into open capacity or escalates true emergencies. Target: zero missed calls.
3. Phase 3 — Estimate follow-up — Automated cadence on every open estimate: day 1, day 3, day 7, day 14, day 30, then quarterly. This is the highest-ROI automation in most service businesses because the demand is already qualified.
4. Phase 4 — Review and reputation — Trigger a review request on job completion plus payment, routed to the platform where you are weakest, with negative-sentiment interception before it posts.
5. Phase 5 — Reporting — Nightly sync from CRM and ad platforms into one dashboard: leads by source, booked rate, close rate, average ticket, revenue by service line.
6. Phase 6 — Back office — Invoice chasing, technician debrief summaries, permit and warranty tracking, and recurring-maintenance renewal reminders.

Roadmap by company size

Guardrails for AI touching customers

• Never let AI quote price or commit to a scope.
• Always disclose when a customer is speaking with an assistant.
• Hard escalation triggers: safety, water or gas, anger, insurance, warranty.
• Log and review 10 interactions weekly for the first 90 days.

Access the full playbook on Penrose Growth

Key Takeaways

• Human vs. AI split by outbound stage
• Deliverability and domain protection basics
• Research and personalization that scales
• Benchmarks from active service-business programs

Where AI wins and where it loses

What kills reply rates

• Fully AI-written emails at volume. Buyers in the trades identify them instantly; reply rates in our tracked accounts fall by roughly half.
• Sending from the primary domain. Always use dedicated sending domains with matching branding.
• Volume without warm-up. New domains need 4–6 weeks of ramp; cap at 30–50 sends per mailbox per day.
• Missing authentication. SPF, DKIM, and DMARC are non-negotiable in 2026 — bulk senders without them are filtered by default.

Deliverability checklist

1. Domains — 2–4 dedicated domains, each with 2–3 mailboxes. Never send outbound from the domain your customers email.
2. Authentication — SPF, DKIM, DMARC with a policy of at least quarantine. Verify with a header check monthly.
3. Warm-up — Automated warm-up for 4–6 weeks before real sends; keep warm-up running underneath live campaigns.
4. List hygiene — Verify every address before send. Keep bounce rate under 2% and spam complaints under 0.1%.
5. Content — Plain text, no tracking pixels on cold touches, no images, at most one link after the first reply.

Benchmarks from active service-business programs

Access the full field report on Penrose Growth

Key Takeaways

• Target account and building-list construction
• Sequences that get replies from PMs and facilities
• Territory and rep coverage models
• Pipeline math: activity to booked walkthroughs

Commercial demand does not arrive through a form

Residential demand is captured. Commercial demand is created. Property managers, facility directors, and general contractors already have an incumbent vendor; they are not searching. They switch when the incumbent fails, when a portfolio changes hands, or when a credible alternative is standing there at the moment of a failure.

Outbound's job is to be the credible alternative already known at the moment of failure. That reframes the goal: you are not closing on contact one, you are earning a place on a short list you will be pulled from six to eighteen months out.

Build the list at the building level, not the company level

Most outbound fails because the list is a company list. Commercial service work is bought per property. Build a building-level target list and attach the decision chain to each address.

• Source: county assessor records, CoStar or Reonomy, permit filings, chamber and BOMA directories.
• Qualify by square footage, asset class, building age, and roof or system age where relevant.
• Attach: owner entity, property management firm, on-site engineer, regional facility lead.
• Trigger events: ownership change, permit filed, new tenant, capital project announced, negative online review of the incumbent.

Sequence structure that gets replies

1. Touch 1 — Property-specific email — Reference the actual building. One sentence of relevance, one sentence of proof at a comparable property, one low-friction ask (a 15-minute walkthrough, not a meeting).
2. Touch 2 — Call, day 2 — Do not pitch. Confirm you have the right person for that asset and ask who handles emergency dispatch after hours.
3. Touch 3 — Value asset, day 5 — Send something they keep: a system life-expectancy table, a code-compliance checklist, a capital-planning template for their asset class.
4. Touch 4 — Call + voicemail, day 8 — Reference the asset you sent. Voicemails should end with a specific date offer, not 'call me back.'
5. Touch 5 — LinkedIn touch, day 10 — Connect and comment before messaging. Facility and PM communities are small and reputation-driven.
6. Touch 6 — Break-up + standing offer, day 14 — Explicitly offer to be the backup vendor for emergencies. This converts far more often than any pitch in the sequence.
7. Nurture — quarterly — Move non-responders to a quarterly value cadence. Most commercial wins in our programs come from touch 9 through 20, not the initial sequence.

Territory and coverage model

Pipeline math: work backward from booked walkthroughs

Use realistic conversion rates and staff to the goal rather than to activity targets pulled from a SaaS blog.

• 1,000 building-level contacts touched → 60–90 conversations (6–9%).
• 60–90 conversations → 18–30 walkthroughs (30%).
• 18–30 walkthroughs → 8–14 proposals.
• 8–14 proposals → 3–5 wins, weighted toward months 4–9.

Judge outbound on walkthroughs booked and short-list placements in months 1–3, not revenue. Revenue judgments before month six kill programs that were working.

Access the full playbook on Penrose Growth

Key Takeaways

• 10 front-end fixes that reduce wasted spend
• Spam prevention and bot filtering
• Number and email verification at form level
• Lead-quality tracking that ties back to closed revenue

The premise: you do not have a traffic problem

Almost every service business we audit is buying enough clicks. What they are not buying is enough qualified, contactable, in-territory demand. Between 20% and 45% of paid lead volume in the accounts we inherit is junk: bots, spam form fills, wrong-service inquiries, out-of-area requests, and duplicate submissions already in the CRM.

That junk does three kinds of damage. It burns budget directly, it burns CSR and estimator hours indirectly, and — most expensively — it corrupts the data your bidding algorithms optimize against. Google and Meta will happily buy you more of whatever you tell them a conversion is. If your conversion definition includes garbage, you are paying a platform to manufacture garbage at scale.

Fix the front end before you touch bids. Budget efficiency gains from clean conversion data routinely exceed anything you get from campaign restructuring.

The 10 front-end fixes, in build order

1. 1. Kill the generic contact form — Replace one open-text form with service-specific intake. Ask for service type, property type, and ZIP before name and email. Unqualified traffic self-selects out.
2. 2. Enforce real-time email verification — Validate at submit, not after. Syntax, domain MX, disposable-domain and catch-all detection. Reject hard bounces before they enter the CRM.
3. 3. Enforce phone validation — Line-type lookup: reject VoIP-only and invalid numbers, flag landlines for commercial, require a valid mobile for text-based follow-up.
4. 4. Add a honeypot plus timing gate — A hidden field no human sees, plus a minimum time-on-form threshold. This alone removes the majority of unsophisticated bot submissions with zero UX cost.
5. 5. Layer invisible bot scoring — reCAPTCHA v3 or Turnstile in score mode. Never a visual challenge on a paid landing page — challenges cost you real conversions.
6. 6. Geo-fence at the form level — ZIP or postal validation against your actual service map. Show an honest out-of-area message rather than capturing a lead you cannot serve.
7. 7. Dedupe against the CRM at submit — Match on phone and email. Existing customers get routed to service, not to new-customer acquisition reporting.
8. 8. Separate conversion actions by value — A booked appointment, a verified form, and a 60-second-plus call are three different conversion events with three different values. Feed values, not counts.
9. 9. Instrument call tracking properly — Dynamic number insertion by source, call recording, and duration thresholds. Sub-30-second calls should never count as conversions.
10. 10. Send offline conversions back — Push booked and closed revenue back into the ad platforms. This is the single highest-leverage change most accounts have never made.

Conversion values that actually reflect economics

Set values from your own numbers, not the platform defaults. Start with average job value, multiply by the historical close rate for that source and service, then subtract fulfillment cost if margins vary widely by service line.

Weekly operating cadence

• Monday: review last week's leads flagged as junk; add negatives, block placements, tighten geo.
• Monday: pull close rate by campaign and service line, not by channel average.
• Mid-week: review call recordings for the 10 longest and 10 shortest calls from paid.
• Month-end: reconcile ad spend to closed revenue in the CRM; recompute conversion values.

What good looks like after 90 days

• Junk lead rate under 8% of total paid volume.
• Cost per booked appointment reported weekly and trending down.
• Every campaign tied to closed revenue, not lead count.
• Bidding optimized against offline conversions imported from the CRM.

Access the full playbook on Penrose Growth