Tactics

The diagnosis

Right channel, wrong job

Meta is very good at making someone want a van and very bad at making them buy one this week. Those are different jobs and the same budget cannot do both.

Why Meta genuinely fits this product

The purchase is emotionalA Class B van is used about 21 days a year. Nobody does that arithmetic and concludes it is good value. It is bought on the life it represents, and that is exactly what Meta sells well.
It covers both halves of a split marketClass B ownership is barbell-shaped, 51% under 35 and 49% over 55. Facebook reaches the older half and Instagram the younger one. One platform family, two very different buyers.
The category is natively visualA finished build photographs and films well. Very few high-ticket products can say that.
Nobody is searching for usDemand here has to be created, not captured. That is a paid-social job before it is a search job.

Why it will still look like it is failing

The ask is wrongIf cold traffic is being asked for a quote on a $180,000 vehicle, the cost per lead will look terrible and the leads will be unqualified. That is the funnel, not the channel.
The landing experience contradicts itselfThree different prices are live for the same van right now, with a $30,929 gap on El Capitan. Every ad dollar currently pays to deliver someone to that.
Last-click will bury itWith a six to twelve month cycle, credit lands on whatever touch happened last, usually a branded search. Meta does the introduction and gets none of the credit.
There is nowhere to go but a formNo nurture means a lead who is nine months from buying is simply lost. The triggers here are calendar-driven, not campaign-driven.

The map

What each channel is actually for

In a six-to-twelve month considered purchase, no single channel does the whole job. The question is not which channel wins, it is which stage each one owns.

Channel	The job it owns	What it does well	Where it breaks
Meta — Facebookkeep · Most of the spend to date	Reach the 55+ half of the market with the life the van makes possible.	Cheap reach, strong targeting, and the half of this audience that actually lives on Facebook.	Nobody on Facebook is shopping for a $180,000 van today. Asking for a quote here buys expensive tire-kickers.
Meta — Instagramkeep · Most of the spend to date	Aesthetic shortlisting. This is where a build gets saved and sent to a spouse.	The category is natively visual and the under-35 half lives here. Reels carry a finished build better than any photo.	Saves and follows are not leads. Measured on last-click it will always look worse than it is.
YouTubeadd · Gap	The 40-minute build tour a buyer watches three times before calling anyone.	The research named this the dominant medium in the category. Content made once keeps working for years.	Slow. Nothing happens for months, then it compounds.
Platform forumsadd · Gap	Survive the vetting. Sprinter-Source, Ford Transit USA Forum, ProMaster Forum.	Where the skeptical high-ticket buyer checks whether a shop is real. Every competitor complaint in our research surfaced here. Costs nothing but attention.	Marketing that reads as marketing gets destroyed. Requires a real person answering technical questions.
Google Ads — non-brandfix first · Significant spend	Catch live intent: conversion cost, builders near me, chassis and layout queries.	The highest intent available anywhere, and Arizona is thin on supply. 13 builders statewide against 79 in California.	These are price-comparison searches and they land on the pages quoting three different numbers. This channel is hurt by that more than any other.
Google Ads — brandtest · Significant spend	Appear when someone already searching for Papago by name.	Cheap clicks, high close rate, and it blocks a competitor bidding on the name.	Much of it is often harvesting traffic that would have arrived free. Untested, it flatters itself and every channel upstream of it goes uncredited.
Email nurtureadd · Unknown	Stay present for the six to twelve months between first look and signing.	The buying triggers here are calendar-driven: a retirement date, a home sale, a vesting date. The job is to still be there when it arrives.	Requires something worth sending. A monthly build story, not a newsletter.
Build-your-van configuratoradd · Gap	Pick a plan, pick your options, watch the price, the weight and the delivery date move.	The full expression of the one thing we already do that nobody else does, which is publish prices. It outputs an all-in number, so the conversion-only comparison problem disappears. Anyone who finishes a configuration is the most qualified lead this business can generate.	Cannot be built until Papago Collection is defined, because you can only configure a product with finite options. It is also a real software project, not a page.
3D walkthroughsadd · Gap	Let someone stand inside the van from their sofa in Vermont.	Answers the objection that kills these deals for distant buyers: they have never been inside one. A scan is cheap, runs in a browser with no app, and gives cold traffic something to do that is not filling in a form.	Economically terrible for one-off custom builds, because scanning a van that is already sold markets nothing. It only compounds on a fixed product line.
Reviewsuse · Underused	Peer proof, which the research ranked above every marketing claim.	4.7 across 71 reviews. Several direct competitors have zero reachable reviews anywhere.	Currently sitting unused. It is an asset, not a channel, and nothing points at it.
Rental fleettest · Gap	Let someone sleep in one before committing six figures.	Boho runs 40+ vans this way and converts renters into buyers. It removes the objection no brochure answers.	A different business with its own insurance, damage and utilization problems. Test with two vans, not forty.

The configurator is not a marketing project bolted onto Papago Collection. It is the same work. Three plans, a good-better-best ladder on each subsystem, a price against every option and a build time behind it: that list is the Papago Collection product specification. You cannot build the tool without making the product decisions, and once the product decisions are made the tool is mostly data entry. So it is worth proposing as the artifact that proves Papago Collection is real rather than as a website feature to be funded separately.

Two things it must do that no competitor's would. First, show a delivery date, not just a price. A running total is useful; "configured today, keys on 4 March" is unanswerable, and it is the twelve-week promise made concrete on the exact screen where someone decides. Second, track payload. A Sprinter 2500 carries roughly 3,800 pounds and a full build eats up to 2,500 of it, so a buyer who ticks every heavy option can configure a van that is illegal before it leaves the lot. Every other configurator in every industry exists to sell you more options. Ours should show a weight bar filling up and tell you when to stop. That is the Payload argument made functional, it prevents us building something unsafe, and it is the single most trust-building thing a tool like this could do.

The floor plan does not move, and that is the whole point. The buyer picks one of the three engineered plans and then works the option ladders inside it. Nothing gets rearranged. That constraint is what keeps the cut lists repeatable, the parts stocked and the twelve-week promise honest, and it is also what makes the tool buildable at all: a fixed plan has a finite number of states to price, weigh and schedule. Moving the galley is a Papago Custom conversation with a human, and it should stay one.

Why a 3D walkthrough is really a Papago Collection asset. Most of our buyers are on the East Coast and will never stand in the Mesa shop before they sign. Boho answers that objection with forty rental vans, which is expensive and physical. A scan answers it from a sofa in Vermont for a few hundred dollars. But the economics only work on a fixed product line. Scanning a one-off custom build markets a van that is already sold. Scanning each of three Papago Collection plans once sells every future unit of that plan, indefinitely. That is why nobody in a custom-only industry bothers, and it is an argument for the second line that has nothing to do with bay-months. It also answers a question this page raises and leaves open: if we stop asking cold traffic for a quote, what do we ask for instead. "Walk through it" is something a stranger will actually click, and four minutes spent inside one qualifies them better than any form.

The two gaps worth the most. YouTube and the platform forums are where the research found this buyer actually doing their homework, and Papago appears in neither. A long, technical, unedited build tour is the single highest-value asset this business could make, because it does the work of a salesperson for years and costs one videographer and a van that is already in the shop. The forums cost nothing but a real person answering real questions under their own name.

The trap

Google is harvesting what Meta planted

This is the specific failure mode of running paid social and paid search together on a long sales cycle, and it is worth naming before anyone looks at a channel report.

The sequence almost always runs like this. Someone sees a build on Instagram in March and does nothing. In July their retirement date firms up, they Google "sprinter van conversion Arizona" or type our name directly, click, and convert. Last-click attribution hands that sale entirely to Google Ads and gives Meta nothing. The report then says Google is efficient and Meta is not, the budget moves, the thing that created the demand gets defunded, and six months later total volume falls with nobody able to say why. With a six to twelve month cycle, search is a harvester. Something upstream has to plant.

Three tests that settle it

Pause branded search for two weeksIf total conversions hold steady, that spend was largely buying traffic that would have arrived free, and it is the cheapest budget Papago will ever find. If they fall, the spend is defending the name and worth keeping. Either answer is useful and the test costs nothing.
Report first touch and last touch side by sideThe gap between the two columns is Meta's real contribution. One number alone cannot show it.
Segment non-brand search from brandThey are different businesses inside one account. Blending them hides a weak non-brand campaign behind a flattering brand one, which is the most common way a search account looks healthier than it is.

What search should be doing here

Follow the East Coast, not the zip code we sit inMost current sales come from the East Coast, and the builder directory shows this industry clustered in the Mountain West and Pacific Northwest. That is a supply gap Papago is already filling from 2,000 miles away. Arizona local search is cheap and worth owning, but it is not where the money is.
Geography is a bid modifier here, not a targetPeople fly in for a $180,000 van. On search, layering income and location bid adjustments over national coverage is the right shape. On Meta it is the wrong shape entirely, for the reason set out below.
Bid the money terms deliberatelyCost, price and comparison queries are where search earns its keep in this category, and they are exactly the queries currently landing on our worst pages.
Competitor terms are legitimate and cheapSomeone searching a rival by name has intent and no loyalty yet. Worth a small, carefully written test.

Where the money is

The buyers are on the East Coast

Papago builds in Mesa and sells mostly to the other side of the country. That is a finding, not a problem, and it should shape both the targeting and the story.

Why it is happening

The category clusters in the westThe builder directory found 79 shops in California, 44 in Colorado, 30 in Oregon, 13 in Arizona. The Mountain West and Pacific Northwest are saturated with premium builders. The East Coast is comparatively underserved.
Distance is not the barrier it looks likeAt this price a buyer will fly. They already are. The purchase is rare enough and considered enough that a plane ticket does not register.
It is a real competitive positionAgainst Outside Van in Portland or Vansmith in Boulder, a western buyer has local options. An East Coast buyer has far fewer, and that is the market where Papago is genuinely needed rather than merely present.
It reframes the Boho questionIf the money is East Coast, a Tempe competitor fifteen minutes away matters less than the Boho analysis assumed. The closed-lost data settles it.

What to do with it

Make the delivery trip the first tripFly in, tour the shop, take delivery in person, drive it home across the country. That is not a logistics cost to apologise for, it is the single best first-trip story this product has and nobody is telling it.
Film that tripA customer driving a new build Arizona to Vermont is the build tour, the proof and the aspiration in one asset.
Weight search spend eastNot exclusive, but the bid adjustments should follow where deals actually close rather than treating the country as flat.
Check the freight assumptionIf Papago is currently paying to transport vans east, that cost belongs in the pro forma and it is not there. If customers collect, that is a marketing asset being treated as an inconvenience.

Andromeda

Avatars belong in the creative, not the targeting

Meta rebuilt its ranking and retrieval system in 2026. The practical consequence is that narrow targeting now costs performance rather than buying it, and creative variety is where the leverage moved.

This is the part of the plan I would change. Hand-picking affluent neighbourhoods across the country splits the budget into many small audiences, and under Andromeda that starves the system of the conversion data it needs to learn. Broad targeting is reported to outperform lookalikes substantially under the new system, and fragmenting spend across narrowly targeted ad sets is specifically called out as the failure mode. The instinct is right and the mechanism is backwards. Meta will find affluent buyers better than a zip list will, provided it is fed real conversion signal and enough creative to match different people with different angles. Give it the avatars as creative, not as audiences.

What this changes on Meta

Fewer ad sets, broaderNational, minimal interest layering, one or two ad sets rather than a dozen neighbourhood-level ones.
Far more creativeThe guidance runs to 20 or more genuinely different creatives per ad set, and brands testing 20+ new ads a month materially outperform those testing under 10. Meta's own data science team puts creative at over half of performance outcomes.
Different angles, not different editsA colour change is not a variant. The retired couple's version, the remote worker's version, the ski-and-bike version and the full-timer's version are variants.
Let the creative do the targetingThe person who stops on a shot of a mobile office is a different buyer from the person who stops on a bike rack, and the system now sorts that out per impression.

Where narrow targeting still works

Google AdsSearch is intent-driven, not inference-driven. Location and household income bid adjustments are legitimate and useful there. Do the affluent-geography work in the search account, not the Meta account.
RetargetingSmall, defined audiences are exactly right for people who already visited, and that is where objection-handling creative belongs.
ExclusionsStill worth keeping tight. Existing customers and recent buyers should not be paid for twice.
The honest caveatAndromeda guidance is a year old and the sources are agencies rather than Meta. The principle, broad plus creative volume plus conversion signal, is consistent across all of them, but treat specific percentages as directional.

The keystone

HubSpot is what makes any of it work

Tracking through the sales process is not admin. On a six-to-twelve month cycle with two paid channels and an algorithm that learns from conversions, it is the thing everything else depends on.

The specific unlock is offline conversion upload. Both Meta and Google accept closed-won deals pushed back from a CRM. Do that and the platforms stop optimising for form fills and start optimising for people who actually bought. Under Andromeda that is precisely the signal the system is hungriest for, and it is the reason broad targeting works: the algorithm can only find more buyers if it is told who the buyers were. Without it, Meta is optimising toward whoever fills in a form, which on a $180,000 product is a materially different and much larger group of people.

What has to be true in HubSpot

Source captured on every contactAnd preserved. First touch overwritten by last touch is the most common way this breaks.
Closed-lost reason and competitor, requiredItem 10 of the data request. Without it we can measure what we won and stay blind to what we lost.
Deal stages that match the real cycleEnquiry, consultation, design deposit, build deposit, delivered. Stages that reflect how the money actually arrives.
Closed-won pushed back to both ad platformsThe loop that turns a CRM from a filing cabinet into the thing that improves the ad buying.

What it produces once it is running

Cost per sale by channelThe only number that settles the Meta versus Google argument, and it does not exist today.
Real first-touch attributionWhich is where the case for keeping Meta funded will live.
Actual geography of buyersConfirming or killing the East Coast read, and telling us where to weight the search bids.
Which avatar actually convertsTurning the creative angles from four guesses into a ranked list.

The plan

First ninety days

Sequenced so the cheapest and most certain work happens first, and no budget decision gets made before the numbers to make it arrive.

Weeks 1–2

Stop the bleeding

Reconcile the three published price lists and kill /rainier-old/ and the stale FAQ figure.
Server-render the build tiers hub so crawlers stop reading every tier as $0.
Publish a van-plus-build total next to every conversion price, so a shopper can compare us to Outside Van without doing arithmetic.
Audit HubSpot: are closed-lost reasons and competitor names being recorded at all.
Pull the baseline for both paid channels: spend, leads, cost per lead, and what share of closed deals ever touched each one.
Split the Google Ads account reporting into brand and non-brand. They are different businesses and blending them hides which one is working.
Check whether HubSpot is preserving first-touch source, and whether closed-won is being pushed back to Meta and Google at all.

Weeks 3–6

Build the proof

Low

Film two build tours on vans already in the shop. Long, unedited, technical. Not a commercial.
Scan two finished builds as 3D walkthroughs. A few hundred dollars each, and it gives the ads somewhere to send people that is not a quote form.
Spec the configurator on paper, in parallel with the Papago Collection plan definition. Options, prices, weights and build times per subsystem. No code yet, because the spec is the hard part and it is the same document as the product definition.
Put the reviews to work. 71 at 4.7 while several direct competitors have none reachable anywhere. Surface them on the tier pages and cite the count in ad creative. Free, and nothing points at them today.
Draft the milestone payment structure and the twelve-week Papago Collection guarantee, and take both to ownership together. They answer the two things that actually stop these deals: deposit risk and the wait.
Open a real account on Sprinter-Source and the Transit forum. Answer technical questions, sign posts with a name, sell nothing.
Stand up a nurture sequence so a lead who is nine months out does not go cold.

Weeks 7–12

Rebuild the spend

Reallocation, not increase

Consolidate Meta into fewer, broader ad sets and stop splitting spend across narrow geographies. Under Andromeda that starves the learning.
Wire HubSpot closed-won deals back into Meta and Google as offline conversions, so both platforms optimise for buyers instead of form fills.
Build out four creative angles from the avatars and run them against one broad audience, not four narrow ones.
Move the Meta call to action from 'get a quote' to something a stranger will actually do: walk through the van in 3D, watch the tour, see the price, join the list.
Run the two-week branded search holdout and find out what that spend is actually buying.
Point non-brand search at a pricing page that tells the truth, with a van-plus-build total on it.
Measure against a window that matches the sales cycle. Last-click over 30 days will understate everything upstream.
Decide on the two-van rental pilot using the cost per sale numbers by then in hand.

Notice what is not in the first ninety days: a budget increase. Every item above is either free, a reallocation, or one videographer. That is deliberate. Asking ownership for more money before producing a cost-per-sale number is how a marketing hire spends their credibility in month two. The first ninety days are for establishing what a sale currently costs, fixing what is visibly broken, and building assets that keep working after the spend stops.

Measurement

The scoreboard, and why the obvious one is wrong

What to measure

Cost per sale, not cost per leadA cheap lead on a $180,000 product means nothing. The only number that survives a conversation with ownership is what a signed deal costs to acquire.
Marketing as a percent of revenueThe second number I will be measured against, and neither exists today.
First touch and last touch, bothReport them side by side. The gap between them is where Meta's real contribution hides.
Pipeline created, not just closedWith a nine-month cycle, this quarter's closes came from last year's spend. Judging current spend on current revenue punishes the wrong thing.

What will mislead

A 30-day attribution windowShorter than the sales cycle, so it will credit branded search and bury everything that created the demand.
Meta's own reported conversionsThe platform grades its own homework and counts view-through. Useful directionally, never as the number reported to ownership.
Lead volume as a headlineEasy to move and easy to fake by lowering the ask. Volume rising while close rate falls is a worse business.
Any of it, until HubSpot is recording lossesWithout closed-lost reasons and competitor names, we can measure what we won and stay blind to what we lost, which is the half that tells us what to change.

What unblocks the rest of this page. Items 07 and 08 of the data request: 2025 marketing spend broken out by channel, and lead volume by source through to close. Those two together produce cost per sale by channel, which is the number that decides whether Meta gets more money, the same money spent differently, or less. Until then everything here is sequencing and judgment, not arithmetic.

Two paid channels.No way to tell them apart.