Commercialization Playbook · Part 3 of 3
I ended the last piece with a question: if CPE sits at the bottom of the stack and the carrier’s metric is locked at the top, what exactly is that metric?
It isn’t BOM. It isn’t throughput. It isn’t even unit margin.
It’s cost-per-subscriber over the device’s useful life. And almost nobody on the vendor side prices their proposals to it.
Eleven years of watching this from the inside, I’ve come to a flat conclusion: the proposal that wins isn’t the one with the cheapest BOM, the best Wi-Fi spec, or the lowest landed cost. It’s the one priced in the same arithmetic the carrier is already using to evaluate it.
The unit on the table is the device. The unit on the invoice is the device. The unit on the P&L is the subscriber.
The Formula Nobody Quotes Against
Every Tier-1 and most Tier-2 ISPs run the same calculation, in some form or another.
Cost-per-subscriber over a five-year horizon equals the landed cost of the device, plus truck rolls — the initial install plus any field intervention across the device’s life — plus warranty reserves at the carrier’s projected RMA frequency and mortality curve, plus customer support load costed at the fully-loaded call-center hourly rate, plus firmware lifecycle cost in carrier engineering hours, divided by the number of subscribers the device successfully serves before replacement.
Six inputs in the numerator. One number in the denominator. Both are levers.
Most vendor proposals optimize one input. They drive the BOM down — strip a chipset variant, requalify a memory module, find a cheaper power supply — and win the line item on a number that represents maybe a quarter of the actual cost-per-subscriber the carrier will absorb. The other seventy-five percent is invisible to the vendor, and completely visible to the carrier’s finance team.
The vendors who win consistently move both. They lower the BOM where they can. But they also extend the denominator — longer device life, lower RMA rate, fewer support calls, cleaner firmware updates — because they understand the denominator is the bigger lever in the carrier’s math.
A dollar off the BOM is a dollar off the numerator. A six-month extension on average device life is a five-to-ten-dollar reduction in cost-per-subscriber. The carrier’s CFO knows that arithmetic by heart. Most CPE BD managers have never written it down.
Why the Cheapest Bid Costs the Carrier More
I’ve watched a Tier-2 ISP run an RFP where the lowest-BOM proposal came in nearly twelve percent under the next bidder. On paper, the obvious win. Procurement’s spreadsheet, sorted by unit cost, ranked it first. The carrier’s finance team ran a different spreadsheet. Same dataset, different denominator. The cheapest device carried a projected RMA rate more than twice the next bidder’s, a higher support-call-per-thousand-subscriber number from a prior deployment, and a chipset roadmap that ended eighteen months earlier. Adjusted for five-year cost-per-subscriber, the “cheapest” device landed over eight percent more expensive than the second cheapest. The award went to the second cheapest. The lowest bidder never understood why.
I’ve also watched the inverse. A vendor whose BOM came in eleven percent above the lowest bid walked into a US Tier-1 evaluation and submitted not a unit price but a five-year subscriber economics statement — landed cost, warranty reserve at three RMA scenarios, expected support-call reduction versus the incumbent CPE, firmware lifecycle commitment. Same device the rest of the market had quoted in BOM terms. Priced in the carrier’s own language. The procurement memo justifying the award didn’t mention unit cost in the first three paragraphs. It mentioned cost-per-subscriber. That vendor wasn’t the cheapest. They were the only one who showed up speaking the carrier’s math.
Both stories are the same story. One vendor optimized the numerator. The other optimized the equation.
How to Actually Quote It
Three concrete moves, if you want to start pricing the way carriers actually evaluate.
Quote the five-year, not the unit. Build a one-page lifecycle cost statement alongside the BOM. Landed cost, plus truck-roll cost at the carrier’s published or regionally typical rate, plus warranty reserve at your committed RMA rate, plus a support-call delta versus a named comparable device, plus firmware support commitment. Sum it. Divide by subscribers served. That number is the one the carrier is already building internally. Hand it to them before they have to build it themselves.
Show the denominator, not just the numerator. Most proposals quote MTBF and stop. Carriers don’t pay for MTBF. They pay for the consequence of mortality — replacement device cost, second truck roll, customer churn risk during the outage window. Quote those consequences. A vendor who says “our RMA rate is one-point-two percent, which at your subscriber base translates to roughly X fewer truck rolls per year than the incumbent” is doing the carrier’s denominator work for them.
Match the carrier’s subscriber math. Find their ARPU. Find their churn rate. Find their depreciation horizon in the last earnings disclosure or capex deck. Price your CPE to fit inside that math, not outside it. If the carrier is depreciating CPE over sixty months at a target cost-per-subscriber of fifteen dollars, your proposal has to land inside that envelope. A device priced at a number the carrier’s accounting model cannot absorb isn’t a competitive proposal. It’s a rejection slip in advance.
What This Means If You’re on the Vendor Side
Three things, if you take nothing else from this.
First, the unit isn’t the device. It’s the subscriber. The carrier’s P&L is built per subscriber, depreciated over the lifecycle of every component that touches that subscriber. Until your proposal speaks in those units, it isn’t in the conversation.
Second, lifecycle math, not BOM math. The numerator is a quarter of the answer. The denominator is the rest of it. Vendors who quote only the numerator are bidding on twenty-five percent of the carrier’s actual question.
Third, price to the denominator. Lower RMA rates, longer device life, fewer support calls, cleaner firmware updates — these aren’t features. They’re price reductions in the carrier’s math. The vendor who treats them as such walks into procurement with a number nobody else can match.
The Playbook, in One Line
Three pieces in, and the map is on the table.
RFPs aren’t markets. They’re confirmations of decisions the carrier has already made. CPE comes last because the network is built top-down, and the envelope at the top is locked before the RFP is written. And the carrier’s metric was never going to be unit cost — it was always going to be cost-per-subscriber over the device lifecycle.
The vendors who treat the RFP as a market lose. The vendors who read where they sit in the stack, who quote into the carrier’s actual arithmetic, who fit the envelope instead of trying to redefine it — win.
That’s the entire commercialization playbook, in one line. Everything else is execution.
← Previously: Why CPE Comes Last in the RFP Stack
