NEMT is a phone business with a geography problem, and Google Ads punishes companies that get either wrong. The phone part: the overwhelming majority of bookable NEMT demand converts by call — a daughter arranging her father’s dialysis schedule, a discharge planner with a 2pm pickup window, a family whose broker-assigned ride just no-showed — which means an account measured on form fills is measuring the wrong species, and an account without call tracking is flying on rumor. The geography part: your revenue lives inside a service polygon your vehicles can profitably cover, demand density varies wildly across it (the corridor between the dialysis clusters and the hospital district is worth multiples of the exurban edge), and the default location settings quietly sell you clicks from people interested in your area rather than in it — the classic presence-vs-interest leak, which in a vehicles-on-roads business converts to literally undeliverable calls.
And then there’s the screening problem unique to this category’s query stream: “medical transportation” searches mix your bookable private-pay and facility demand with three populations you cannot serve through an ad click — emergencies that need 911 (a real and serious ad-copy responsibility), Medicaid members whose rides are booked through brokers (a huge share of the query volume in most states), and job seekers flooding “NEMT” terms with driver-application intent. An account that doesn’t screen these — in keywords, copy, and call handling — spends most of its budget buying conversations its dispatcher can’t convert.
This guide is the operating manual: campaign architecture split by audience and trip type (consumer, facility, and the failure-demand goldmine), area targeting done with polygon honesty (radius and location strategy, presence-only settings, bid layering by demand density), the negative and copy screens for the three unservable populations, call tracking as the account’s central nervous system (pools, recording, booking-confirmed conversions), dispatch as the final conversion step (answer rates and after-hours reality), and measurement that separates episodic bookings from the standing-order relationships that make the numbers work.
NEMT PPC is a call business inside a polygon — build it that way. Architecture: consumer trip campaigns split by service/trip type (wheelchair, stretcher, dialysis schedules, discharge pickups — different values, different urgency), the failure-demand campaign (“ride didn’t show,” “same-day wheelchair transport” — highest intent in the category), and the facility campaign (low volume, huge value — “NEMT provider [city]” on business-hours schedule, landing on the facility hub). Geography: target the real service polygon (cities/zips over lazy radii), presence-only location setting always, bid adjustments by demand density (the dialysis-corridor zips up, the profitable-but-thin edge down), and exclusions beyond the polygon — per the location-settings discipline. The three screens: emergency intent (negative ambulance/911/emergency terms; copy states “non-emergency” explicitly — a safety responsibility, not just hygiene), Medicaid-broker intent (negative the broker-booking phrasings or route them to the honest explainer — don’t pay to answer the broker’s phone), and job seekers (“driver,” “jobs,” “hiring,” “salary” — the category’s biggest junk stream). Call infrastructure: tracking numbers per campaign with recording, booking-confirmed calls (not raw calls) as the primary conversion via call-length proxies upgraded to CRM import, dispatch answer rate and after-hours coverage measured like media. Judge on cost per completed booking and per facility account — two funnels, two scoreboards.
Campaign Architecture: Audience × Trip Type × Urgency
| Campaign | Query territory | Build notes |
|---|---|---|
| Consumer — wheelchair/ambulatory | “wheelchair transportation [city],” “handicap van service,” “elderly transportation” | The volume workhorse: call-focused assets, family-framing copy (“book for a loved one”), landing on the matching service-type pages |
| Consumer — stretcher | “stretcher transport,” “gurney transportation,” “bed-bound transport” | Higher value per trip, lower volume — separate campaign so its bids and budget aren’t averaged away; capability honesty in copy (only advertise what the fleet actually does) |
| Trip-purpose — dialysis & recurring | “dialysis transportation,” “transportation to chemo,” “medical appointments rides” | The recurring-revenue consumer tier — a booked dialysis schedule is dozens of trips; copy sells reliability (“same driver, every visit”), and this campaign earns the aggressive bids its lifetime values justify |
| Failure-demand / same-day | “same day wheelchair transport,” “medical ride today,” “transportation didn’t show up” | The category’s highest-intent moment: broker no-shows and last-minute discharges — premium bids justified only if dispatch can actually deliver same-day; honest availability copy, extended-hours scheduling |
| Facility / B2B | “NEMT provider [city],” “patient transport company,” “medical transportation for facilities” | Low volume, account-sized value: business-hours schedule, provider-call CTA, the facility hub as landing page, and patience — per the B2B economics, one landed account carries the campaign’s year |
Area Targeting: The Polygon Is the Business
- Target the real service area, not a lazy radius: the cities, counties, and zips your fleet profitably covers — a radius from your yard usually includes territory you’d decline and excludes corridors you love; build the location list to match dispatch reality, and revisit it when the fleet or contracts change.
- Presence-only, always: the location option must be people in your locations — the default “presence or interest” setting buys searches about your area from anywhere (the out-of-state daughter researching is the one legitimate exception, and she’s searching “[city] wheelchair transport,” which presence-targeting on your metro still catches when she includes the city name — the leak the setting stops is the rest of the world’s noise, per the full treatment).
- Bid by demand density: location bid adjustments layered on the polygon — up on the zips dense with dialysis centers, hospitals, and senior communities (map your actual trip origins; dispatch data is the bid map), down on the thin edges where deadhead miles eat the margin; this is the cheapest optimization in the account and almost nobody does it.
- Exclude explicitly beyond the polygon: neighboring metros with confusable city names, and — where relevant — the long-distance-transfer intent you don’t serve (or route it to its own small campaign if you do interstate transfers, priced accordingly).
Export ninety days of completed trips with pickup zips and revenue, pivot by zip, and you hold the demand-density map no keyword tool can generate: which zips produce the recurring dialysis schedules, which produce one-off airport runs, which produce the cancellations and deadhead losers. Translate it directly into location bid adjustments (+15–30% on the top-decile zips, −20–50% on the money-losers), into the geo terms worth their own ad groups (‘wheelchair transportation [suburb]’ where the density justifies it, per the hyperlocal playbook), and into the honest answer to ‘should we expand the radius’ (the edge zips’ current margins answer it). Refresh quarterly — facility contracts and demographic shifts move the map, and the account should move with it.
The Three Screens: Keywords, Copy, and Call Handling in Depth
- The emergency screen — a safety responsibility first: negative every emergency phrasings (“ambulance,” “911,” “emergency transport,” “ER now” families), state “non-emergency medical transportation” explicitly in copy and landing headlines, and script dispatch’s first question to triage (“is this a medical emergency? call 911”) — because someone in a genuine emergency clicking your ad is a harm scenario, not just a wasted click, and the account should be built so it functionally cannot happen.
- The Medicaid-broker screen: in broker-model states, the large “Medicaid transportation” query stream mostly needs the broker’s booking line — negative the clearly-broker phrasings (“Medicaid ride number,” “free medical transportation Medicaid” where you can’t serve it) from consumer campaigns, and let the organic payer-navigation content serve that demand for free; the paid exception worth keeping: the failure-demand phrasings (“Medicaid ride didn’t come”) where private-pay conversion is real, with copy honest about what you are (“private wheelchair transport — when you can’t wait”).
- The job-seeker screen: “NEMT driver,” “jobs,” “hiring,” “salary,” “how to start a NEMT business,” “NEMT contracts” — the category’s single largest junk stream (industry terms attract industry aspirants), seeded at launch into the shared negative lists and re-mined weekly at first; if you are hiring, that’s a separate recruitment campaign with its own budget, never contamination of the demand campaigns.
Call Infrastructure: The Account’s Central Nervous System
- Tracking numbers per campaign (dynamic insertion on the site, dedicated numbers on assets), with recording enabled and disclosed — the attribution backbone that makes everything below possible, and the coaching library dispatch improvement runs on.
- Booking-confirmed as the primary conversion: raw calls lie (job seekers and broker-confused callers dial too) — start with a call-length threshold as the proxy (bookings run long; wrong numbers don’t), then graduate to the honest version: dispatch marks booked/not-booked (and the not-booked reason) per call in the booking system, imported as offline conversions per the primary/secondary design — at which point Smart Bidding is finally learning from trips, not telephony.
- Values by trip economics: the dialysis-schedule booking imported at schedule value, not single-trip value — the recurring tier’s bids are only defensible if the algorithm sees what a standing schedule is worth; facility provider-calls valued at pipeline expectation.
- Dispatch metrics on the marketing dashboard: answer rate by hour, time-to-answer, booking rate per answered call, after-hours volume — because the last conversion step is a human with a headset, and a 70% answer rate is a 30% media-budget bonfire no bid strategy can fix. Ad scheduling follows dispatch reality: if nights genuinely can’t book, stop paying for night clicks — or better, fix nights, because the failure-demand campaign’s best hours are exactly when competitors’ phones go dark.
Transportation-adjacent categories have appeared unevenly in Local Services Ads’ vertical expansions, and availability, screening requirements, and category definitions shift — so the standing instruction is: check LSA’s current category list for your market rather than assuming either way. If available: the pay-per-lead model and the screening badge suit this trust-sensitive category well (the same logic as the LSA-vs-search comparison — badge trust plus lead pricing insulation from CPC inflation), with the standard disciplines: dispute unqualified leads promptly per the dispute playbook, and remember LSA’s coarse targeting can’t do trip-type precision — it complements rather than replaces the search campaigns’ surgical tiers. If unavailable: the search-ads architecture above is the whole paid game, and the budget that would have gone to LSA belongs in the failure-demand and dialysis tiers where intent density is highest. Either way, the GBP layer (service-area setup, categories, reviews) underpins both surfaces — paid performance in this category visibly correlates with profile strength, because callers check the profile between the ad click and the dial.
Measurement: Two Funnels, Priced Honestly
The consumer funnel prices in cost per completed booking (not per call — the booking-confirmed import is what makes this readable) with the recurring tier additionally read in schedules booked and their projected trip volume; the facility funnel prices in provider conversations and accounts landed, on quarterly patience, with pipeline value logged as recurring per the recurring-revenue lens; and the shared reads: search-terms hygiene trend (the junk share should fall monthly for the first quarter), polygon efficiency (bookings by zip vs spend by zip — the dispatch-data loop closing), dispatch answer rate trend, and the blended sanity check — total ad spend against total trips originated, the number the owner actually feels. The quarterly reallocation logic: failure-demand and dialysis tiers usually win on cost-per-booking and earn budget; generic terms usually lose to them and surrender it; the facility campaign is judged on a different clock entirely — and the source-of-truth accounting keeps all three honest when the booking system, GA4, and Google Ads inevitably disagree.
5 Common NEMT PPC Mistakes
- Raw calls as the conversion. Smart Bidding optimizes toward job seekers and wrong numbers with mechanical enthusiasm — booking-confirmed or bust.
- Presence-or-interest location settings. Buying the world’s curiosity about your city while your vans serve a polygon — the one-checkbox fix.
- No job-seeker wall. The category’s biggest junk stream, discoverable in any search-terms report, funded month after month.
- Advertising same-day service dispatch can’t deliver. The failure-demand tier’s premium clicks converting to apologies — capability honesty before bid courage.
- One blended campaign, one blended number. Stretcher economics averaged with ambulatory, facility patience judged on consumer clocks — split or misread everything.
Frequently Asked Questions
What's a realistic cost per booking for NEMT Google Ads?
Build the expectation from the chain rather than an industry average, because the spread across metros and campaign tiers is enormous. The components: NEMT CPCs typically run modest by service-industry standards (transportation terms price well below legal or home-services auctions in most metros), call-through rates from well-matched ads and landing pages run strong (this audience calls readily — the phone-first design pays here), and booking rates per answered call range widely (40–70% for well-screened traffic with competent dispatch; far lower when the three screens are missing — which is the diagnostic: a low booking rate is almost always a screening problem wearing a dispatch costume). Multiply through and well-run consumer campaigns commonly land cost-per-completed-booking in the low-to-mid double digits — against single-trip revenues that make one-off bookings marginally profitable and recurring-schedule bookings extremely so, which is why the tier structure matters: the dialysis campaign’s cost-per-booking buys dozens of trips, the same-day tier’s premium buys the category’s hottest intent, and the generic tier’s cheaper bookings buy the least value — averaging them into one number hides the reallocation the data is trying to suggest. The honest first-quarter expectation: costs run high while the negative walls build (the junk share falls week over week under disciplined mining), then settle as screening matures — judge the trend at ninety days, per tier, in bookings, and let schedules-booked carry the extra weight it deserves.
Should we advertise for Medicaid members at all if the broker controls the rides?
Mostly no for direct booking intent, selectively yes at the edges — and the edges are where the strategy lives. The core rule: in broker-model states, paying for clicks from members who must book through the broker line buys conversations dispatch can’t convert — those phrasings (‘Medicaid ride phone number,’ ‘free Medicaid transportation’) belong in the negative walls, with the organic payer-navigation content serving that demand at zero marginal cost (and building the goodwill that pays later). The profitable edges: failure demand — ‘Medicaid ride didn’t show,’ ‘broker transport late for dialysis’ searches are the moment families consider private-pay for reliability, and honest copy (‘private wheelchair transport when the assigned ride fails — same-day where available’) converts a real fraction at real margins; network-participation leverage — where you’re in the broker network and the state’s system supports member preference or provider requests, awareness advertising has a mechanism to convert through (know your state’s actual mechanics before spending on this theory); and the payer-adjacent discoveries — the ‘does insurance cover medical rides’ stream includes Medicare Advantage members with transport benefits and facility-arranged situations where you are directly bookable, which the landing content can route correctly. The design principle across all of it: never let ad copy imply you’re the Medicaid booking line — the short-term calls it generates cost you the dispatcher time and the long-term trust the honest version compounds.
How do we win the facility and discharge-planner searches with ads when volume is so low?
Accept the tier’s nature — a low-volume, high-value, always-on presence play — and build it to different rules than the consumer campaigns. The mechanics: exact and phrase match on the professional query set (‘NEMT provider [city],’ ‘patient transport company,’ ‘medical transportation for facilities/hospitals/dialysis centers’), business-hours-weighted scheduling (this searcher works a desk), modest daily budget that’s comfortable serving few clicks (the campaign’s job is to never miss the searcher, not to spend), copy speaking operational language (‘Credentialed drivers · standing-order scheduling · dedicated facility line’) — and landing on the facility hub, never the consumer homepage, because the message-match here is the whole conversion. Conversion design for the tier: the provider-call request and credentialing-packet download as its conversions (phone-heavy too — the facility line’s tracking number tells you when the campaign works), manual bidding indefinitely (the volume will never feed automation, and doesn’t need to), and the judgment clock set to quarters — the campaign that spends modestly for months and lands one dialysis-center account has outperformed every consumer tier in the account, which only the recurring-value accounting makes visible. The force multiplier: this tier works best when the searcher who clicks has also seen you exist — the organic facility hub ranking alongside the ad, the GBP profile with facility-flavored reviews — because professional buyers verify before they call; the ad buys the moment, the footprint closes it.
Our phones ring from the ads but half the calls aren't bookable. What's the fix sequence?
Half-unbookable is the unscreened-account signature, and the fix sequence is diagnosis by call disposition — a week of dispatch logging (every ad-attributed call tagged: booked / job seeker / broker-Medicaid / emergency-misroute / out-of-area / price-shop / other) turns the mystery into a pie chart, and each slice has its own fix. Job-seeker slice (usually the biggest): the negative wall (‘driver, jobs, hiring, salary, start a NEMT business’) plus a search-terms mining pass — this slice should approach zero within two weeks of discipline. Broker-Medicaid slice: the phrasing negatives, the copy honesty (‘private-pay and insurance-verified transport’ framing), and the landing page’s payer paragraph doing pre-call routing — plus the script’s graceful redirect (‘here’s your broker’s booking number’) that at least converts the call to goodwill in seconds instead of minutes. Out-of-area slice: the location-settings check (presence-only?), the polygon audit, and the landing page’s service-area map earning its keep. Emergency slice (should be rare but matters most): the negative terms, the ‘non-emergency’ copy prominence, and the script’s triage line — treat any volume here as a build defect to eliminate. Price-shop slice: not waste — a conversion-rate problem; the honest pricing content pre-frames, and dispatch scripting (quote structure plus value framing) converts a real share. Then re-run the disposition week monthly: the bookable share climbing from ~50% toward 75–85% is the whole account improving at once — same clicks, more trips — and the disposition log doubles as the offline-conversion feed that finally teaches the bidding what a good call is.
Is it worth running ads at night and weekends for NEMT?
Follow the demand and your dispatch reality — the category’s off-hours are asymmetric in a way flat scheduling wastes. The demand shape: discharge-driven and failure-demand searches spill well past business hours (hospitals discharge into evenings; broker no-shows for morning dialysis get discovered at 5:30am; the ‘same day transport’ panic peaks early morning and late afternoon), family research happens on evenings and weekends (the working daughter arranges Dad’s transport after her own workday), while facility searches concentrate in business hours — so the tiers deserve different clocks: facility campaign on business hours, consumer research tiers extending into evening, and the failure-demand/same-day tier earning the widest schedule your operation can honestly serve. The dispatch gate: every advertised hour needs an answering capability — a call at 9pm to voicemail is a premium click donated to the competitor who answers (the after-hours answering service with a booking protocol pays for itself out of the first few captured schedules, and its cost belongs in the media math); where genuine 24/7 dispatch doesn’t exist, honest ad scheduling that stops paying for unanswerable hours is the correct second-best — with the strategic note that off-hours coverage is one of the few structural advantages a local operator can build against bigger competitors, because the failure-demand tier’s best conversions happen exactly when most NEMT phones go dark. The measurement close: hour-of-day booking-rate data (from the call dispositions) makes this whole question empirical within a quarter — let it set the schedule, then re-check seasonally as facility contracts shift the demand map.
Buying calls your dispatcher can’t turn into trips?
We’ll rebuild the account around the polygon and the phones — the trip-type tiers, the three screens, booking-confirmed tracking, and the dispatch loop — priced in completed bookings and standing schedules, not ring counts.
Get an NEMT PPC Audit Run the 200-Point Audit