Section 1
The two cost curves
Set up the comparison honestly, because the platforms win the comparison operators usually run and lose the one that matters. The rented curve is flat and permanent. On Google Local Services Ads, cost per lead averaged around 53 dollars across 888 contractors and 126,650 leads in early 2026, and rented cost per lead does not fall with scale. It tends to rise, because you are bidding against every other tenant and the platform captures the surplus (Searchlight Digital / BlueGrid, 2026). Worse, that is cost per lead, not cost per job. On Angi, shared leads convert at 13 to 20 percent versus 27 to 30 percent for exclusive, so a shared lead's true cost per booked job is the lead price divided by your win rate (LeadTruffle, 2026). A 53-dollar lead that converts at 18 percent is a 294-dollar booking cost. Bark makes it starker: credits at 2.35 dollars each, 5 to 20 credits per lead, and from November 2025 those credits expire in 90 days whether you use them or not (Bark Help Center, 2026). The rented curve never bends down. Every job costs roughly the same forever, and forever is at the landlord's discretion. The owned curve is steep, then it bends. Building your own booking channel, a fast site with your own booking flow, a Google Business Profile you control, a review asset on your domain, a follow-up system, an email list, costs real money and time up front and converts almost nothing in month one. Cost per booked job through an owned channel starts absurdly high, because you are dividing a fixed build cost by a handful of bookings. But the denominator grows. Every review compounds. Every past customer becomes a repeat or a referrer. Referred customers cost less to acquire, convert at higher rates, and carry 16 to 25 percent higher lifetime value than customers from other channels (Wharton, Schmitt/Skiera/Van den Bulte). The owned cost curve falls as the asset accumulates, and at some point it crosses under the flat rented line and keeps falling. The crossover is the whole game. Before it, rent is cheaper and the spreadsheet tells you to stay a tenant. After it, owned is cheaper and getting cheaper, and you own the thing besides. The operators who win are the ones who fund the gap between month one and the crossover without flinching.
Section 2
The framework: four models on the build decision
Whether to fund that gap is a capital allocation decision under uncertainty, with a strategic counterparty who can move the goalposts. Four models, run together. Model 1: Solow growth, on why owned compounds and rented does not The Solow model separates two kinds of growth. Accumulation of a factor hits diminishing returns and stalls. Only technology, the thing that raises output per unit of input, sustains growth. Map it onto acquisition. Buying more rented leads is pure accumulation. Double the spend, roughly double the leads, no improvement in cost per job, and you plateau the moment you stop paying. An owned channel is the technology term. A review asset, a reputation, an email list, a referral loop raise output per dollar over time. The same marketing dollar books more jobs next year than this year because the asset underneath it grew. Rented spend is labor you rent by the hour. Owned spend is capital you build once and draw on repeatedly. Solow growth: the compounding lens • Assumes: owned assets genuinely raise productivity per dollar over time, not just once. • Fits because: reviews, lists and referral loops accumulate and pay forward. • Breaks when: the owned asset decays faster than it compounds (reputation damage, a stale list, an algorithm change that buries your profile). • Counteracts: the belief that more spend is the only growth lever. • May reinforce: overinvesting in "asset building" that never actually converts. Model 2: Mechanism design, on why rent rises The platform designed a system where your cost per job trends up and your ability to leave trends down. Shared leads maximize platform revenue per lead. Credit expiries and automated disputes minimize platform obligation. Reviews that live on the platform's domain rather than yours mean your reputation is collateral you cannot take with you. Read as design, the message is consistent: the rented curve is engineered to stay flat-to-rising and to keep you from pricing the owned alternative. Which is precisely the reason to build the owned alternative, so you can price it and walk. Mechanism design: the "why is rent structured this way" lens • Assumes: platform rules optimize platform take, not your cost per job. • Fits because: pricing, disputes and review ownership all tilt one direction over time. • Breaks when: a rule change is externally forced rather than chosen, so the intent read misfires. • Counteracts: modeling rent as a stable, fair market price. • May reinforce: assuming every platform move is hostile when some are neutral. Model 3: Real options, on the value of the build even before payback A real option is the value of the right, but not the obligation, to act later. Building an owned channel before you strictly need it is buying an option. The exercise value shows up the day a platform raises rent, changes terms, or cuts you off. If you already have owned demand, that day is a shrug. If you do not, it is a crisis with no cheap fix, because a channel takes many months to build and you needed it yesterday. The option has value even in the scenario where the platforms behave, because the build lowers your cost per job regardless. You are paying a premium now for the right to not be a hostage later, and the premium is cheap relative to the payoff. Real options: the "value of the right to walk" lens • Assumes: owned channels take time to build, so pre-building has option value. • Fits because: platform terms can break overnight and the fix is slow. • Breaks when: the build cost is so high it exceeds the insurance value, over-hedging a risk that never fires. • Counteracts: waiting until you are forced to move, when it is too late and too expensive. • May reinforce: premature building that starves the current, working channel of cash. The structure-break flag Every payback model assumes the inputs hold long enough to reach payback. The aggregator can break the inputs before you get there. If Google reprices LSA, if Bark expires your credits, if Angi widens lead-sharing, your rented cost per job jumps and your crossover math changes underneath you. This is why the model is not a static spreadsheet. It is a plan that has to survive the landlord rewriting a term mid-build. Any 18-month payback that assumes 18 months of stable rent is fiction. Build the owned asset partly because the rented curve itself is not stable.
Section 3
The solution: GEER, RADAR, CHAIN
GEER: rank the levers on the crossover Net exposure: your cost per booked job is high, rising, and controlled by someone else. Dominant channels are price and rule risk. Levers, cheapest and most reversible first. 1. Measure cost per booked job, not cost per lead. You cannot manage the crossover you cannot see. Divide every channel's spend by jobs actually won, not leads received. Most operators discover their rented channel costs two to five times what they thought. Free, and it reframes every decision below. 2. Capture every rented customer into an owned asset. You already paid rent to acquire them. Get them onto your list, your booking link, your review request before the job ends. This is the single highest-return move, because it converts sunk rent into a compounding asset at near-zero marginal cost and bends the owned curve down faster. 3. Stand up the owned booking flow. Your own site, your own calendar, your own Google Business Profile. This is the capital asset. It carries the Solow compounding and it is the thing a buyer pays a multiple for. 4. Engineer the referral and repeat loop. Highest-LTV, lowest-CAC customers, and they build the asset for free. A deliberate ask after every completed job. This is the steepest bend in the owned curve. 5. Demote rented channels to overflow financing. Keep them as the marginal buyer of idle capacity while the owned channel ramps. They fund the build. Do not fire the thing paying for the transition. Financing flag: levers 1, 2 and 4 are time, not cash. Lever 3 needs modest capital, funded by the rented channel you keep running as overflow. The build is self-financing if you sequence it right. RADAR: schedule against the crossover • DO NOW: Instrument cost per booked job this week. Turn on contact capture for every job. Start the review-and-referral ask. Zero regret, they lower cost and build the asset under every scenario. • HEDGE: Export platform customer data now, before terms change and lock it. Keep reviews mirrored off-platform. Run two rented channels so no single landlord controls your ramp. Cheap insurance against a structure break mid-build. • DEFER + TRIGGER: Do not cut rented spend on a schedule. Cut it on a signal. Pre-commit: when owned bookings cross the point where owned cost per job drops below rented cost per job, step rented spend down by a fixed increment and redirect it into the owned asset. Also trigger a cut if a platform's effective cost per booked job breaches your ceiling or it changes terms against you. Decide the numbers now. CHAIN: check the payback against history Reference class by structure: operators funding a fixed-cost owned asset to escape a variable rented one, in a business where reputation and repeat custom compound locally. The base rate is encouraging when the operator finishes the build and brutal when they quit early. Franchise operators who built independent local reputations kept margin; those who stayed pure to the franchisor's lead flow did not. E-commerce sellers who built owned email and repeat purchase survived platform fee hikes; pure marketplace sellers got squeezed. The pattern: the crossover is real, it takes longer than people expect, roughly a year and a half for a service operator building from a standing start, and the failure mode is almost always abandonment before the curve bends, not the curve failing to bend. Present-state modifier: local service reputation is stickier and more referral-driven than e-commerce, which shortens the payback and raises the odds. Your customers are geographically concentrated and talk to each other, so the owned asset compounds faster than the base rate suggests. Matrix-break flag: if a platform genuinely becomes a net-new demand engine reaching customers you never could, the "escape the tollbooth" reference class no longer fits and you would keep more rented spend than the model says. Rare, but watch for it.
Section 4
The exit-value case, because this is where it pays off most
The crossover lowers your cost per job. The exit is where owning the channel pays off in a lump. Home service businesses traded roughly 3.5 to 7 times EBITDA in 2026, but the range inside that band is enormous and it is driven by exactly the thing an owned channel produces (Sunbelt, CT Acquisitions, 2026). Recurring and repeat revenue commands premium multiples, roughly 6.5 to 7.5 times for recurring maintenance against 4.5 to 5.5 for one-off work, and businesses with high recurring revenue can reach well into double-digit multiples. Retention drives it directly: 80 percent-plus annual customer retention adds a 1.0 to 1.5 times multiple premium over a business retaining 50 percent. And customer concentration or dependence triggers a haircut. A buyer performing diligence on a business whose pipeline is 80 percent rented from Angi sees a single point of failure the seller does not control, and prices it accordingly. A business that owns its demand, its list, its reviews and its repeat customers is buying the top of the range. Same revenue, same profit, materially different check, because one of them owns the asset and the other rents it. Put concrete: a business at, say, 500,000 dollars of EBITDA priced at 4 times because its demand is rented and concentrated sells for 2 million. The same EBITDA at 6.5 times because demand is owned, recurring and diversified sells for 3.25 million. The owned channel did not just lower your cost per job for eighteen months. It added over a million dollars to the exit. That is the real payback, and it never shows up in a cost-per-lead comparison.
Section 5
What this framework cannot see
First, it cannot time your exit or guarantee the multiple. Valuation multiples move with rates, buyer appetite and the cycle. The owned channel improves your position within whatever market exists; it does not control the market. A great asset in a bad M&A year still sells for less than the same asset in a hot one. Second, it cannot price your build competence. The Solow term only compounds if the asset is actually good. A neglected list, ignored reviews, a slow site that loses bookings, these decay instead of compounding. The model assumes you can execute the build. Many operators fund the gap and build something that never bends the curve, then conclude owned channels do not work. Third, it says nothing about the demand-creation question for a brand-new operator. If you have no reputation and no book, rented leads are seed capital, not rent, and the crossover math does not apply until you have something to compound. This is a model for operators with a business to convert into an asset, not for week one.
Section 6
The fitness test
Two numbers, one ratio. Cost per booked job through your owned channel, and cost per booked job through your rented channels. Track both every month. The day the owned number crosses under the rented number, you have reached payback, and from there the gap widens in your favor while the asset keeps building toward your exit. If the owned number is not falling month over month, your build is decorative, not compounding, and you should fix the follow-up before you spend another dollar on it. Watch the two lines cross. That crossing is the moment you stopped renting and started owning. --- Sources: Searchlight Digital and BlueGrid Media, Google LSA cost and statistics 2026; LeadTruffle, Angi lead conversion data 2026; Bark.com Help Center, credit pricing and expiry 2026; Sunbelt Atlanta and CT Acquisitions, home services valuation multiples 2026; Schmitt, Skiera, Van den Bulte, "Referral Programs and Customer Value" (Wharton, 2011). Multiples, conversion rates and lead costs are trade- and market-specific and move with the cycle; the 18-month crossover is a typical-operator estimate, not a guarantee. Treat all figures as order-of-magnitude planning inputs.