The job hasn’t changed much in twenty-five years. Loading a truck, protecting the goods, showing up when promised: none of that looks fundamentally different than it did in 2000. What has changed almost beyond recognition is the cost of finding the person who needs that job done.
Four separate channels now compete for that same job today, one added in each of the last four eras of demand generation. None of the earlier ones retired when the next one arrived. An operator today typically pays for all four at once. Their predecessor twenty years ago paid for one.
Four eras, one toll booth
2000–2007. Acquisition was cheap in this era because discovery itself was local, static and largely offline. A business paid for a Yellow Pages listing, negotiated once a year at a fixed rate, plus a sign on the truck that never needed updating. Referrals from satisfied customers did the rest, and cost nothing extra beyond doing the job well the first time. Discovery had a natural ceiling on price, since the places a customer could plausibly find a mover were few enough to count on one hand.
2008–2015. Websites, early SEO and early paid search became normal costs of doing business in this era. A business now needed a site that could be found, which meant a designer, hosting, and someone who understood basic on-page structure. Early Google AdWords auctions were cheap enough that a modestly funded local operator could bid on a competitive keyword and win some of those auctions against a much bigger competitor, because the barrier to entry on paid search was still low. Prices rose from the previous era, but direct acquisition was still the realistic default: a customer found a business and called it, with no platform in between.
2016–2024. Google Ads, SEO as a specialist discipline, Meta advertising, review platforms and lead marketplaces all became standard line items in this era. SEO stopped being something an owner could manage on a weekend and became a retainer relationship with a consultant, because the algorithms it competed against had grown too complex for a part-time effort. Review platforms went from a nice-to-have to something close to a precondition: a business with a thin or outdated review profile increasingly didn’t get considered at all, regardless of how good the work was. Lead marketplaces filled the remaining gap between search and sale. They sold the same household’s contact details to several competitors at once and charged each one for the introduction. Moving-specific cost-per-lead data for this whole span isn’t tracked, but the pattern shows up in adjacent trades: LocaliQ’s 2025 Search Ad Benchmarks report found cost per lead rose for 69% of home services businesses, up 10.51% year over year, outpacing the 5.13% rise across all industries. Cost per booked opportunity climbed through the era, before a single sale had closed and before any commission layered on top of it.
2025–2035. AI-driven search, aggregator platforms and personalised discovery layers look set to make direct acquisition harder again, not easier. The early signs are already measurable: Google’s own AI Mode passed 1 billion monthly users in May 2026, and Google has confirmed its AI can now call a local business directly to gather prices on a customer’s behalf. If more of discovery gets mediated by a platform standing between the operator and the customer, and the trajectory above suggests it will, that platform inherits the same control over visibility that lead marketplaces and review platforms already have. It decides who gets seen, on what terms, before the operator gets a say.
What four tolls add up to
The numbers below are a simplified model, not measured figures pulled from any real business’s books. Picture a local moving company booking twenty jobs a month at an average ticket of $1,800, and price its acquisition cost the way each era’s typical toolkit would have priced it.
In the 2000-2007 model, that acquisition spend behaves like a fixed cost. A Yellow Pages listing and truck signage might run $180 a month, whether the crew books twelve jobs that month or thirty. Spread across twenty bookings, $180 becomes $9 a job: half a percent of the average ticket.
Layer the 2008-2015 toolkit on top rather than swapping it in, since that’s what happened. Add a website, basic SEO work, and a modest AdWords budget: another $720 a month. The Yellow Pages listing doesn’t go anywhere; plenty of customers still check it out of habit. Total acquisition spend is now $900 a month, or $45 a job: two and a half percent of the ticket, five times the era-one figure for the same customer.
Layer on the 2016-2024 toolkit. The number jumps again, by more than either previous step. That era’s toolkit might run $2,700 a month: a specialist SEO retainer, a paid search budget sized to stay competitive locally, review-platform management, and a handful of marketplace leads bought to fill slow weeks. Nothing from the first two eras gets switched off. Total monthly spend: $3,600, or $180 a job, ten percent of the ticket before a truck moves, before a crew gets paid, before profit gets calculated.
The shape of the pattern sits in those three numbers: half a percent of revenue, then two and a half, then ten, just to win the job. The mechanism matters more than the exact dollar figures.
The job stayed the same size. The toll didn’t
Set side by side, those four eras show an unmistakable pattern. The cost of reaching the same customer, for fundamentally the same job, has climbed every decade. Each new era hasn’t replaced the last era’s costs so much as stacked a new toll on top of them. A modern operator often pays for SEO, paid search, review management, and marketplace commissions simultaneously, in a stack their predecessor twenty years ago never had to build at all.
The stacking isn’t an accident of bad negotiating or wasted spend. It’s a structural feature of the category, not a solvable execution problem. Why does the earlier, cheaper channel never get fully retired? The answer is competitive, not historical. Visibility is a relative position, not an absolute one: a business doesn’t need to rank well in some objective sense. It needs to rank better than the specific competitors a customer happens to be comparing it against in that moment. If every competitor in a local market still maintains a baseline SEO presence, dropping that spend doesn’t return a business to some earlier, cheaper equilibrium. It just drops the business behind everyone who kept paying, on a channel that’s now table stakes: the operational cost of staying in the category, not a source of advantage within it. The toll survives because opting out of it doesn’t save the money. It spends the money on lost position instead.
None of that spend was wasted, in the narrow sense that each layer responded rationally to how customers were finding businesses when it arrived. Together, the layers add up to something worse: an industry spending more every year to stand still, with margin quietly transferring from the people doing the physical work to the platforms sitting between them and the customer they’re serving. No operator agreed to that trade. It just happened, one rational decision at a time.
The same undifferentiated spending shows up one level up, in how agencies serve this market: a template is not a moat for the operator paying for it either.
“Nobody’s forcing you to pay for all four”
That specific objection holds up when a single channel gets judged in isolation, and it’s worth taking seriously. Nobody is forcing an operator to run Google Ads. Nobody is forcing an operator to buy leads from a marketplace charging four competitors for the same household’s name. Cutting an underperforming channel and keeping only what pays for itself is good business discipline, not a mistake.
The objection runs out at the level of the whole toll stack, not any single toll inside it. An operator who cuts one channel doesn’t get to keep that channel’s dollar figure as pure savings, because competitors still running it absorb the visibility that channel used to buy. What actually happens is narrower and worse: total acquisition spend falls a little, and so does the operator’s share of local bookings, roughly in proportion. The toll on the whole stack doesn’t drop to zero when one lane closes. It settles at whatever floor the local competitive set has collectively decided to pay, and that floor has risen every era covered above, not fallen.
A real decision sits inside that constraint, just not the one the objection assumes. An operator can’t opt out of the toll system by refusing to pay any single toll inside it. What an operator can influence is which dollars inside that system buy something that keeps paying off after the transaction closes, and which dollars have to be spent again in full the next time a customer needs finding.
What comes next won’t be gentler
The next decade’s discovery layer is built from AI systems that compare providers, gather quotes, and recommend one on a customer’s behalf. That layer doesn’t look likely to reverse this trend. If anything, it adds another intermediary between the operator and the person deciding who gets the job, with its own rules about who gets surfaced, rules the operator doesn’t write and mostly won’t be able to see.
Someone typing a moving request into an AI assistant today usually gets back a short list, not a directory: three or four names, drawn from whatever the model considers well-documented and trustworthy, not every operator capable of doing the job in that zip code. A business with a thin, inconsistent, or outdated presence across the web doesn’t get ranked lower on that list. It often doesn’t make the list at all, because the model has nothing reliable to summarize about it. BrightLocal’s 2026 Local Consumer Review Survey found 45% of consumers had used an AI tool such as ChatGPT for a local business recommendation in the past year, up from 6% a year earlier, already ahead of Yelp and TripAdvisor as a discovery channel. That share will keep moving; the mechanism it exposes doesn’t depend on where it lands next: a discovery layer deciding who gets summarized holds more control than a search results page ever did, since a curated shortlist offers no scroll-down option and no page two, only whatever the algorithm decided to hand over.
Every era added a new toll and kept the old ones. Nobody’s job description got easier; every operator’s cost structure did the opposite.
The operators who’ll do best from here
Adding a fifth toll on top of four won’t separate anyone from the pack. It only accelerates the same losing game at a faster pace, one more subscription competing for the same finite pool of local demand.
The operators who do best from here will invest differently: in something that gets cheaper to run per job over time, instead of resetting to full price every time a customer needs finding. The distinction is concrete, not aspirational. A purchased lead costs roughly the same the hundredth time an operator buys one as it did the first time, since a marketplace transaction carries no learning curve and no discount for experience. When a qualification process gets faster at telling a ready buyer from a tire-kicker in the first two minutes of a call, it gets cheaper to run every time it’s used, because the crew spends less unpaid time on quotes that were never going to close. When a follow-up sequence converts a slightly larger share of the leads already paid for, it does the same thing without buying a single additional lead. Neither improvement costs anything to a marketplace or a platform. Both compound instead of resetting.
The real choice sits underneath the toll booth metaphor: not whether to pay, since every operator pays some version of it, but whether the money spent inside that system builds something that keeps paying back after the transaction that created it closes, or has to be repurchased in full the next time around.
Movaros replaces the toll stack with infrastructure you keep.
Building on shared infrastructure means qualification and follow-up compound instead of resetting to zero every renewal.
No toll-free lane exists
None of the four eras above offered an operator a way to opt out of paying to be found. What changed, era to era, wasn’t whether the toll existed. It was how many lanes fed into it, and how much each lane cost to use.
The pattern is worth sitting with plainly, because it cuts against a comforting assumption some operators make: that the acquisition problem is temporary, a phase to survive until the market settles back into something cheaper. That’s a story operators tell themselves, not a forecast. Nothing in the last twenty-five years supports that read, and nothing in the AI-mediated decade ahead points toward it either. The toll has never gone down. It has only ever added a lane.
An operator with twenty jobs booked this month isn’t facing a problem that gets solved by finding a fifth channel, or by working harder inside the four channels that already exist. Four different acquisition costs are already stacked into every one of those twenty jobs. The gap gets closed, if it gets closed at all, by building infrastructure underneath those four channels that makes each dollar spent on them work harder than it did the year before: sharper qualification, higher conversion. Together, they add up to a system that turns a bought lead into a closed job at a rate the toll stack alone would never predict. The tolls aren’t going away. Nothing has ever moved the number except getting more out of every dollar already paid into them.
Fortifications have followed the same arms-race logic since the first city wall went up: every improvement to the wall provoked an equal improvement to the siege engine trying to breach it, until both sides had spent a fortune achieving exactly the stalemate they started with. Advertising spend runs the identical race today. Rivals bidding up the same keyword are building taller walls and bigger siege towers at the same time, and the only guaranteed outcome is that everyone’s costs go up while nobody’s relative position moves.
