Section 1
Why this works, in one note
Two models justify the sequence. Mechanism design says: stop trying to win the game the platform built, and build your own game with rules that pay you. The aggregator engineered an arrangement where you rent access per job. You engineer an arrangement where a booked job feeds a review, a referral and a repeat visit you own. Colonel Blotto (a resource-allocation model from game theory) says: you have a finite dollar and finite hours to spread across fronts, and you lose fronts you underfund. The mistake is fighting on the aggregator's front at full strength while leaving the owned fronts (map pack, reviews, database, referrals) at zero. The wean is a Blotto reallocation done slowly enough that no front collapses while it is still your only source of work. Neither model tells you the exact week your owned channel overtakes rent. That is why the schedule is gated on real booking data, not on a calendar promise.
Section 2
Before you start: three numbers on an index card
You cannot manage this transition on feel. Get three numbers first. 1. Rented share. Of last month's booked jobs, how many came from paid aggregators. If it is 80 percent or more, you are deep in the trap and the wean runs the full 90 days. If it is under 40 percent, you can move faster. 2. Cost per booked job from the aggregator. Not cost per lead. Total aggregator spend last month divided by jobs actually booked from it. This is the number the owned channel has to beat, and for most trades it is brutal. Reported Angi effective acquisition costs reach 1,000 to 1,400 dollars per booked job once you factor close rates and wasted spend, against roughly 290 to 310 for the same job won through SEO (LeadTruffle, 2026). 3. Weeks of cash. How many weeks of fixed costs you can cover if new-lead flow dropped 30 percent tomorrow. This sets how aggressive you can be. Under six weeks of cash, you hold aggregator spend flat until Week 5 no matter what. Write those three on a card. They are the dashboard for the whole plan.
Section 3
The artifact: the 90-day de-platforming schedule
The plan runs in three 30-day phases. Aggregator spend never drops on a hunch. It drops only when the weekly gate clears. Phase 1, Days 1 to 30: Build the owned rails, hold the rent flat You do not cut a dollar of aggregator spend this month. You keep the rented pipeline at full strength while you lay the track the future work will run on. Cutting now, before anything owned exists, is how operators end up back at the landlord's door. • Week 1. Claim and complete the deed. Google Business Profile fully filled out (services, service area, hours, 15-plus photos), because the map pack is the last channel you rank in for free. Set up a review-request system (a saved text and email you send after every completed job). Stand up a simple booking path on your own site or a booking link, so a lead has somewhere to land that you own. • Week 2. Turn every current job into an owned asset. Start asking every completed-job customer for a review the same day. Load your last two years of customers into one list (a spreadsheet is fine to start) with name, address, job, date. This database is the cheapest lead source you will ever own and most operators have never built it. • Week 3. Point demand at the deed. Add your GBP link and review link to your invoices, email signature, truck, and job-site signage. Ask three suppliers or adjacent trades whether they will trade referrals. The goal is to open owned inlets while the rented one still runs. • Week 4. Measure the baseline. Count booked jobs by source. You now have Week-4 owned bookings, probably small. That number is the thing every later phase watches. Gate to Phase 2: you have a working review system producing at least a handful of new reviews, a customer list built, and at least one owned booking that did not come from an aggregator. If not, run Phase 1 another two weeks. Do not cut spend yet. Phase 2, Days 31 to 60: First cut, tied to owned proof Now the wean begins, and it is small on purpose. You cut aggregator spend only in proportion to owned bookings, and only after they show up. • The cut rule. For every owned booked job you win in a week (map pack, review-driven, referral, repeat), reduce next week's aggregator budget by that job's aggregator cost, not the whole month. If your aggregator books a job for 300 dollars of spend and you won one owned job this week, take 300 off next week's aggregator budget. You are literally letting owned work buy down the rent, one job at a time. • Week 5. Turn on a light local presence push: ask for reviews harder, post to GBP weekly, send one reactivation message to twenty of your best past customers ("we have openings this month"). Repeat-customer work is the fastest owned channel to fire because those people already trust you. • Week 6 to 7. Formalize the referral ask into a script you use on every job (see the companion piece on the referral engine). Keep the aggregator running, but stop upgrading to premium placements or extra lead types. Freeze the meter where it is, then let the cut rule chip at it. • Week 8. Read the numbers. What share of this month's bookings came from owned channels versus rented. If owned share rose at all versus Week 4, the rails work. Gate to Phase 3: owned channels produced at least 20 to 30 percent of the month's booked jobs, and your weeks-of-cash number did not fall. If owned share stalled under 20 percent, hold aggregator spend flat and stay in Phase 2 another two to four weeks. This is the gate that protects payroll. Phase 3, Days 61 to 90: Step down to a floor, do not slam to zero The target is not zero aggregator spend. The target is a deliberate floor: a small, watched aggregator budget you keep as overflow insurance while owned channels carry the base load. • Week 9 to 10. Accelerate the cut rule. Now reduce aggregator spend by 1.5 times each owned job's cost, because momentum in reviews and referrals compounds and you want to lean into it. Keep watching weeks-of-cash weekly. • Week 11. Handle the contract, carefully. Angi contracts commonly carry early-cancellation costs of 30 to 35 percent of the remaining term and require up to 60 days notice to cancel at term end (LeadTruffle / 7ten, 2026). Bark credits bought from November 2025 expire three months after purchase, so stop buying credits you will not burn (Bark Help Center, 2026). Do not pay a penalty to leave a month early. Time your exit to the renewal date and give notice inside the window. • Week 12. Set the floor. Most operators land at 20 to 40 percent of their original aggregator spend as a permanent overflow channel, switched on only in slow weeks. You are now a customer of the platform, not a tenant of it. The difference is that you can walk, so the pricing stops dictating your business. Final gate: owned channels carry the majority of booked jobs and weeks-of-cash is stable or higher. If yes, hold the floor. If a channel wobbles, the aggregator is still there at reduced spend to smooth it. That optionality is the whole point.
Section 4
The ordered lever list (GEER)
If you want the levers ranked by cheap-and-reversible first, pull them in this order. Each one is reversible except the last, which is why it comes last. 1. Review-request system (Week 1). Free, reversible, compounds. The single highest-return move because reviews feed both the map pack and referrals. 2. Customer database (Week 2). Free, reversible. Turns past work into a repeat and reactivation channel you fully own. 3. GBP completion and weekly posts (Week 1, ongoing). Free, reversible. Your only free ranking surface. 4. Referral ask on every job (Week 6). Free, reversible. Referral leads close at 30 to 50 percent against 8 to 15 percent for marketplace leads (Signpost / ServiceTitan, 2026). 5. Reactivation messages to past customers (Week 5). Near-free, reversible. Fastest owned channel to produce a booking. 6. The cut rule on aggregator spend (Week 5 on). Reversible: you can raise the budget back any week if owned bookings dip. 7. Contract cancellation (Week 11). Irreversible and penalty-loaded, so it comes last and only at the renewal window. The order is not arbitrary. Levers 1 through 5 build the owned pipeline at almost no cash cost while the rent still runs. Only after they produce do you touch spend (lever 6), and only at the end, from a position of strength, do you touch the contract (lever 7).
Section 5
What this plan cannot see
The schedule assumes your market still has a free map-pack channel and that owned demand is reachable in your area. In a few saturated metros, organic and map-pack visibility is so contested that the owned channel takes longer than 90 days to carry the base load, and the floor you hold on aggregators stays higher than 40 percent. The gates protect you here: if owned share will not clear 20 percent after Phase 2, the plan tells you to hold, not to jump. The other blind spot is a demand shock. If your whole market softens during the wean, both pipelines shrink at once, and the right move is to freeze the cut and keep the rented overflow until demand returns. The plan is a throttle, not a switch. Used as a throttle, it gets you off the meter without a single quiet week you did not choose.
Section 6
Are you ready for this
You are ready to run the 90-day plan if you have at least six weeks of cash, a stream of completed jobs to turn into reviews and referrals, and the discipline to cut spend on booking data instead of frustration. You are not ready if you have under six weeks of cash and 90 percent aggregator dependency at the same time. In that case, run Phase 1 only, build the owned rails for a full 60 days before you cut a dollar, and let the wean start once the base is thicker. Either way, the rule holds: never cut the rent faster than the owned pipeline fills the room. Sources: LeadTruffle, Angi Leads Cost 2026; 7ten, Angi Leads Costs & Cancellation; Bark Help Centre, Understanding lead pricing; Signpost, Referral Marketing Statistics 2026; ServiceTitan, Home Services Industry Statistics.