How to Decide Which Rebate Programs Are Worth Running

Why this matters

Every rebate program you agree to administer is a standing office job you took on for free. Somebody in your shop now checks eligibility before the sale, chases a serial number after the install, fills the form, tracks a claim they cannot see inside, and fields the "where is my check" call for as long as the processor takes to pay. Shops rarely decide to do that. They drift into it one program at a time, because a supplier rep asked or a customer waved a flyer, and two seasons later a coordinator is spending most of a day a week on paperwork nobody ever measured. The decision worth making on purpose is which programs you run as a standing line, which ones you point at and hand off, and which ones you stay out of.

Know which direction the money runs before you screen anything

In nearly every utility and manufacturer program, the incentive belongs to the customer. You are not being paid to file. You are absorbing office labor so that somebody else's money arrives, and the only reason to do it is that it moves one of your own numbers:

• It lifts close rate on qualifying proposals. A credible incentive lowers the effective price of the option you actually want the customer to buy.
• It buys competitive parity. If every shop bidding the same job is a listed participant on the same program and you are not, you are the one explaining why your customer has to chase it alone.
• It produces goodwill you can trace. A customer whose claim paid without a fight refers; a customer whose claim died on your paperwork error tells people.

If a candidate program does none of those three, it is not a program, it is a chore with a logo.

There is a fourth category that never belongs in this screen at all, because there is nothing for you to administer: incentives the customer claims directly. Federal residential energy tax credits under 26 U.S.C. 25C and 25D, claimed by the homeowner on IRS Form 5695, are the common example. You file nothing, so there is no administrative line to evaluate. That does not make them work-free for you: 25C carries a qualified-manufacturer product identification number requirement for property placed in service after 2024, which means the customer needs a document carrying that number from you or the manufacturer even though you are not a party to the return. And because these credits have been created, amended and terminated by successive tax acts, their availability in any given tax year is a question for the current Form 5695 instructions, never for your sales script.

The four gates

Run every candidate through all four. They are AND, not a score. A program becomes a standing line only if it clears every one.

Gate 1, fit without distortion. Does a job type you already sell qualify as you already sell it? If qualifying means changing your standard specification, that is a much larger decision about what you install and it does not belong inside a paperwork screen. Answer it separately or the incentive will quietly redesign your offering.

Gate 2, volume floor. At least 10 qualifying jobs a year, counted off last year's actual sold work rather than off optimism. Below roughly 10, every claim is somebody's first claim: the form is unfamiliar, the current version has changed since last time, and handling time never comes down the learning curve.

Gate 3, handling ceiling. At or below 2.0 office hours per claim, measured as a median of at least 5 completed claims, counted from the first eligibility check to funds confirmed, and including one rejection-and-refile cycle. The unit is per claim, not per job. It is a median because a single ugly claim will drag a mean over the line on its own.

Gate 4, materiality to the customer. The incentive covers at least roughly a tenth of a typical ticket for that job type. Under that it does not move a decision, which usually means the close-rate payoff is not there either.

Gate 3 is the one everybody guesses at

Estimates of handling time run low by a wide margin because people time the wrong thing. They time filling the form. The form is the small part. What you are actually measuring is: reading the current terms, confirming the model is on the qualifying list as of the install date, collecting the customer's signature, chasing the nameplate photo the tech did not take, splitting the qualifying item onto its own invoice line, submitting, logging the confirmation number, two customer status calls, and, on one claim in several, a rejection and a refile.

Measure it once, deliberately, on the next 5 claims. Have whoever handles them note elapsed office time in tenths of an hour against the claim, not against the day. Do not average in the claim that went perfectly and do not throw out the one that went badly; take the median of the 5 and use it. If your program has been running long enough that you have 5 claims of history but nobody timed them, you do not have a gate 3 number. You have an opinion.

Case one: a program that clears

A replacement job type, last year: 40 qualifying proposals went out, 20 sold, a 50% close rate.

• Gate 1. The equipment qualifies exactly as the shop already specifies it. Pass.
• Gate 2. 20 qualifying jobs against a floor of 10. Pass.
• Gate 3. Handling hours on the 7 claims filed last season, sorted: 1.2, 1.4, 1.5, 1.5, 1.8, 2.0, 3.6. With 7 values the median is the 4th, so 1.5 office hours. Pass, at or below the 2.0 ceiling. Note the honest part: one of those 7 blew straight through the ceiling on its own at 3.6 hours, and it was a refile. One breach in 7 is what the ceiling is meant to tolerate. Three breaches in 7 would flip the read even with a passing median, so count them, do not just take the middle value.
• Gate 4. The incentive covers roughly a fifth of a typical ticket for that job type. Pass.

Now size the standing load. 20 claims at 1.5 hours is 30 office hours. Add a rejection allowance at the observed rate of about one in eight, which is 3 refiles, at 1.0 extra hour each, so 3 hours. Total 33 office hours a year. Against roughly 1,900 working hours in a coordinator's year, that is about 1.7% of one person.

The season the shop started presenting the program, close rate on those 40 qualifying proposals moved from 50% to 60%, so 20 sold became 24. Be careful how you read that. One season is not proof, and a 10-point close-rate move has several plausible causes besides a rebate flyer. Be careful in a second way too: those 4 extra jobs are sold field work that still costs technician hours, travel and materials to deliver, so they are not the same currency as the 33 office hours and you must not divide one into the other and call the result a return. The only question the screen has to answer is whether 1.7% of a coordinator's year is worth a plausible four additional jobs plus parity with every competitor quoting the same program. It plainly is.

Case two: a program that fails

Same shop, second program. 12 qualifying proposals last year, 6 sold.

• Gate 1. Passes. The work qualifies as specified.
• Gate 2. 6 qualifying jobs against a floor of 10. Fail.
• Gate 3. Handling hours on the 5 claims with history, sorted: 2.4, 2.8, 3.0, 3.3, 4.5. Median is the 3rd value, 3.0 office hours. Fail, above the 2.0 ceiling. Rejection rate ran about one in three, more than double the first program's one in eight.
• Gate 4. The incentive covers under a tenth of a typical ticket. Fail.

The annual load is 6 claims at 3.0 hours, so 18 hours, plus 2 refiles at 2.0 hours each, so 4 hours. 22 office hours, about 1.2% of a coordinator's year. Notice that the load is smaller than the program that passed. If you screened on hours alone you would keep the wrong one. What kills this program is that 22 hours buys nothing measurable: the incentive is too small to change a customer's decision, the volume is too low to ever get fluent, and the rejection rate is high precisely because nobody in the shop files it often enough to remember the fine print.

The third outcome: refer, do not administer

Gates 1 and 4 clearing while gate 2 or gate 3 fails is a real and common shape, and the answer is not binary. Put that program on the refer-only list. You tell the customer it exists, you hand them the program's own current terms page, and you give them every document they need from you: an itemized invoice with the qualifying item on its own line, the model and serial of what you installed, the install and completion dates, and your license number. You do not submit on their behalf and you do not track it.

Say it plainly at the sale so nobody is surprised later. "There is a program that covers part of this. I am not the one who files it and I cannot approve it, but I will hand you everything you need to apply and I will get it to you the day we finish." That is an honest position, not a dodge. A shop that files four claims a year is worse at it than the customer who reads the instructions once, carefully. See the sibling card on explaining incentives without overpromising for the language around what you can and cannot say about the amount.

Re-screening, and how a program moves

Re-screen annually, and immediately on any one of three triggers: the program publishes new terms, your qualifying volume drops below the floor for two consecutive quarters, or your measured median handling time on the last 5 claims crosses 2.0 hours.

Movement is one step per trigger event, in one direction: standing line moves to refer-only, refer-only moves to out. Do not jump from standing line to out on a single bad quarter, because seasonal volume swings will bounce you out of programs you should keep. Do not promote a refer-only program back to a standing line until it has cleared the volume floor for two consecutive quarters, because one good quarter of qualifying work does not rebuild fluency with a form.

The failure mode this whole screen exists to prevent is not a bad program. It is three mediocre programs running at once, none of them measured, with the coordinator's actual job quietly crowded out behind them. That failure is slow, it never announces itself, and the first hard evidence usually shows up somewhere else entirely, in collections aging or in a late invoice run.

References

• IRS, Instructions for Form 5695, Residential Energy Credits (26 U.S.C. 25C and 25D), including the qualified manufacturer product identification number requirement
• U.S. Department of Energy and EPA, ENERGY STAR program requirements and rebate finder
• The current published terms of the specific utility, manufacturer or state program, which are the only authority on eligibility, funding and deadlines
• See related: Explaining Rebates and Incentives Without Overpromising; The Paperwork Chain Behind a Rebate and Who Owns Each Link; The Rebate Season That Cost More Than It Returned