From a bigger agency to a growth platform

Today HSD runs ads, Local Service Ads, lead generation and reviews for 36 home service companies, mostly tree service. That business stays exactly as it is. The 10X move is adding a $299 a month website product that far more contractors can buy, built and run mostly by our own AI, so we can grow to about 300 customers without growing the team at the same rate.

Keep the agency. Add one simple product, one way to get customers, and AI that does the routine work.

What stays and what we add

Live numbers on Where we are today. How the AI side works on The AI that runs it.

What we promise the owner

On the page we say: Get found. Book jobs. Grow repeat business. Inside the company the same three are called Acquisition, Conversion and Retention.

The path a contractor takes

Where we are today

Read from Stripe on October 28, 2026: every company we charged from September 28 to October 28. A flat fee every month = a retainer client. A charge that changes week to week = a pay-per-lead client. Some are both.

Rounded. A $20 one-time lead charge (Baja Hauling) isn't counted as a client. Alexa's Tree Service was charged in this window and stopped working with us in October.

What clients pay us for

Most common first: Facebook ads with Google review management, Local Service Ads ranking, Google Ads, a growth package (several channels for one fee), social media management, SEO, AI answering and voice AI, phone usage, and one website fee (Q&T Services).

What we already built

Most of the 10X machine exists. The plan is to connect it and sell it, not to start over.

"Have it today" = works and is in use. "Planned" = built or started, not yet doing the job at full speed. The Agent OS dashboard and the Miro boards (Agent OS map, the $299 front-door plan) have the full detail.

Redesign the company, don't just push harder

The idea from 10X Is Easier Than 2X (Dan Sullivan and Benjamin Hardy): growing 2X usually means doing more of the same. Growing 10X means deciding what to stop, hand off, automate or standardize, so we can serve far more customers without far more of David's time or far more staff.

What 10X means for us

Every quarter: what do we stop?

Look at every product, task, customer type and David's calendar, and ask: does it move a number that matters? Can it become a system? Can someone else, or an AI agent, own it? Would we start it today if we weren't already doing it?

Get Leads, Book Jobs, Grow Customers

Everything we sell fits in one of three buckets. The two entry offers get the customer in. The rest are add-ons once the first product is working for him.

1. Get Leads

2. Book Jobs

The biggest software opportunity: turning calls, forms and estimates the owner already has into booked jobs.

3. Grow Customers

What we will not build

Two ways in, one relationship

Start with two entry offers, not nine products at once. Every customer gets one app and one relationship with us, whatever he bought first.

Add-on prices are guesses to test with real customers, not set prices. Consider a simple "Growth" bundle instead of a long menu for first-time buyers.

How we sell it: build first, no credit card

We don't sell "live in an hour." The owner doesn't care how fast we build; he cares that it's his site, it looks right and it brings calls. Still to check: when the card is collected (at signing, held until the site is built, or a payment link after the build), and what happens if he doesn't like the site.

What the ladder can add up to

Examples to show the math, not predictions. Sebastian's figure: confirm the time period and the net amount.

Rules for the ladder

• Don't charge twice. Anything already in the $299 is never sold again as an add-on. A premium tier has to add something clear.
• Write down the limits. Revisions, launch time, support, text and phone usage, third-party fees, and who handles content, photos, domain and approvals.
• A low price must not hide work. $299 never turns into unlimited custom work.
• Sell the upgrade after the first product works, not before.

Who owns the website

One app, three tabs

The contractor sees one simple HSD app on his phone and computer. It shows him the next thing to do, not a giant CRM.

Home screen idea: new leads this week, leads waiting for a reply, estimates waiting for follow-up, estimates booked, missed calls recovered, newest reviews. Only say an automation "booked a job" when the data shows it.

What we have today

What the app should record

So we can prove value and spot problems: site launched, form submitted, first alert delivered, missed call recovered, reply sent, estimate follow-up started, estimate booked, review requested, past customer reactivated, and every subscription change. When we can't be sure what caused a result, label it an estimate.

AI does the routine work, people check it

Going from 36 to about 300 customers only works if new customers don't need new people at the same rate. So every routine job in the new business gets an AI agent, built on tools we already have. People check the work, handle the exceptions and keep customers happy.

Five agents, in this order: websites, sales follow-up, client reporting, reviews, onboarding.

How an agent earns trust

November 9: one funnel, one offer

Two channels go live together and are measured separately: Facebook and Instagram ads, and cold email. Both sell the same $299 offer to tree companies.

Channel 1: Facebook and Instagram ads

The five ad messages

No promises of lead counts, sales or rankings in any ad. Later test: let the video send people straight to checkout as well as to a call, once the call path works.

Channel 2: Cold email

• Clean, de-duplicate and verify the list first.
• Sort by website quality, fit, region and likely need.
• Short, personal and honest. Only mention something about his site if we really looked.
• Link to a short page and a booking link. Don't make every recipient sit through the long video.
• SPF, DKIM, DMARC set up, warm the inboxes, build up to 800 a day. Don't start new domains at full volume.
• Follow CAN-SPAM: real sender, honest subject, our mailing address, a working unsubscribe, honored fast. A scraped email is not permission to text or auto-call.
• Track replies, good replies, calls booked and held, customers and cancels for email on its own.

Next channels

The goal for the first 30 days

Teach the result, not the tech

Before: mostly teaching tree owners to close more jobs. Now: helping home service owners get found, turn inquiries into booked jobs, and get customers back. Every post shows the result for the owner in the first three seconds.

Promise: "Get found. Catch every lead. Book more jobs." Not website design tips, not another ad agency, not a Jobber replacement, never a guaranteed #1 on Google.

What we post about

Starting mix, not a weekly quota. Examples about 70% tree service, 30% other trades, reviewed after 30 days.

The series we repeat

How every Reel is built

Profile and the path to a sale

Post → profile or DM → short offer explanation → sales video → booked call → $299 customer → he uses it → the right add-on.

The weekly routine

Roles: David picks topics and sometimes films; an editor cuts; the operations partner posts and checks; Sebastian (or a CSM) feeds client questions and hands leads to sales. David must not become the daily posting bottleneck.

Ten Reel scripts and a 30-day calendar

Drafts. Check every claim and every on-screen product step before posting. Film these five first: #1, #2, #4, #6 and #9.

30-day calendar

One post a day is the stretch goal. If that's too much, post five a week and keep the order. Skewed toward websites on purpose during the $299 launch.

What to test in the first 30 days

• Tree only vs. all trades: which brings more good calls?
• Site review vs. straight pitch: which starts more real conversations per 1,000 views?
• Short tip vs. full demo: 20 to 30 seconds vs. 45 to 60 seconds.
• DM a keyword vs. "book a call".
• On camera vs. screen recording.

Keep topics that bring booked and held calls from the right companies. Views alone don't count.

Build it so it runs without David on every call

Sales, in four stages

Delivery: the launch checklist

One trained person can build about five sites a day (David's estimate). Five a day over 20 workdays is 100 builds a month at most, before revisions, approvals, domains, support and checks. That's capacity, not 100 new paying customers.

Steps 1 to 4 are the website agent's job; step 5 and the first 30 days are the onboarding agent's (see The AI that runs it). A person checks anything a customer sees, important messages, and exceptions. A full site with a page for every service; town pages are the SEO package, not the $299.

Customer success

Who does what

The goal: David manages a few leaders with clear scorecards, not every account. Current clients keep the same people and the same service.

The numbers that must work

What subscribers add up to

Simple multiplication. Ignores cancels, discounts, failed payments, costs and upgrades. A customer count alone doesn't tell you what the company is worth.

What it really costs to win a customer

Does each customer pay for himself?

What one $299 customer costs us each month

Fill in the "to measure" rows from the first 20 customers. Until then, the margin on $299 is a guess.

The scorecard

Everything we track, by stage

First milestone

Work toward the first 25 to 50 new subscription customers in the first 60 to 90 days, as a learning milestone, not a promise or a deadline.

Find the bottleneck, fix that one

From Benjamin Hardy's The Science of Scaling: find the one thing holding growth back right now, put the effort there, then look again. Don't try to improve everything at once.

Every week: name the one number holding back new monthly revenue. Give it one owner and the biggest experiment. Don't let side projects eat the time it needs.

The likely order of bottlenecks

A starting guess, not a fixed order. Check which one is real each week.

What to fix, by what you see

Early signs vs. results

Six phases, each with a gate

Each phase has a test it must pass before we spend more or widen out. The real number for each gate comes from our margins, cash and the results we see, not from a guess made today.

Twelve ways this fails, and the guard for each

Rules we don't break

• Product first, not cheap custom agency work: fixed scope, templates, predictable delivery.
• No guaranteed rankings, lead counts or revenue where we don't control them.
• Clear ownership, cancel, data and transfer terms, checked by a lawyer.
• Legal by design: cold email (CAN-SPAM), calls, call recording, text consent and opt-outs (TCPA, A2P), platform rules.
• Google review rules: no filtering out unhappy customers, no rewards for 5 stars, no fake reviews.
• Permission before showing any contractor's site, messages or results, or blur them.
• One owner for every number, reported by offer, channel, trade and signup month.
• Keep customers by value, never by making it hard to leave.
• Don't build our own platform early. Prove demand with what we already have.
• Protect David's time. Every new feature, channel and hire should help hand off work.

Topline Pro: what to learn, what not to copy

What's decided, what's still open

The next 90 days for David

Every quarter: the 10X questions

1. What does the 10X HSD look like: revenue quality, David's calendar, the customer's experience, the company structure?
2. What one bottleneck blocks it today, and what shows that's the one?
3. Which 80% of tasks, features, services or exceptions can we cut, hand off or automate?
4. What is the one experiment that would fix the bottleneck this month?
5. Who owns it, how much time does it get, and what result means keep going, stop or change?
6. Which two good ideas will we deliberately not do this quarter?
7. Did the work add freedom, margin and customer results, or just activity?

Review rhythm: weekly bottleneck review, monthly numbers and product review, quarterly 10X reset. Update the guesses on this site with real data and write down why each decision was made.

Where this came from

Four strategy documents, drafted with ChatGPT from David's conversations, dated October 27 and 28, 2026. They repeat each other a lot; this site keeps each idea once and puts it in plain words.

Notes from merging

• Client count and revenue come from Stripe, read October 28, 2026: every company charged September 28 to October 28. Same fee each month = retainer; a changing weekly charge = pay-per-lead. Paused subscriptions are not counted. The documents' "about 45 clients" was replaced with this.
• Two different ad cost examples ($272 and $381 per customer) came from different booking rates; both are on the numbers page.
• "Local service pages" in the documents: the $299 is a full site with a page for every service; town pages are the SEO package.
• Added by David on October 28: keep every current client as is, 10X = about 300 customers, build the site first with no credit card (from Topline Pro), the agent order, the owners of each area.
• Founder-reported figures (five sites a day, Sebastian's ~$3,500, the 8,000-company list) are as David described them, not checked against the books.
• Topline Pro comparisons are a strategy comparison, not verified facts about them.
• The 10 Reel scripts are ChatGPT drafts; they still need our own script pass before filming.
• The books (10X Is Easier Than 2X, The Science of Scaling) are applied as ideas, not quoted.

Next to this site: the HSD playbook (what we run today) and the $299 website page.