CHANNEL INTELLIGENCE REPORT · 2026
Where vendor back-end money is won or lost, and why no portal shows it
Your vendors give you portals with your numbers in them. The rules that decide what those numbers are worth live somewhere else, and months later a check arrives. Working out what happened in between is left to you, vendor by vendor.
That matters more each year because the transaction itself is moving away from the channel. Omdia puts the channel’s share of global IT spend at 69% in 2025, 65% in 2026 and an expected 63% in 2027, while Gartner expects worldwide IT spending to reach $6.37 trillion in 2026, up 14.2% from 2025. The falling share is of a growing market, and Omdia also estimates that “well over 90% of tech purchases are assisted by partners in some way.” IT solution providers, whether VARs, systems integrators or resellers, are keeping the influence while their share of the invoice shrinks.
What’s left is the back end: rebates, incentives and program funding. Your vendors spent 2026 rebuilding it. In late January, as its fiscal third quarter began, Cisco retired VIP, the Cisco Services Partner Program, Lifecycle Incentives and Perform Plus into one incentive, the Cisco Partner Incentive.
In May, Dell announced that its August program enhancements would include a new base rebate for named focus products. Microsoft’s new growth margin went live on October 1. And from November 1, 2026, HPE and Juniper partners work under one program, HPE Partner Ready Vantage.
Ask where a solution provider checks what it has earned, and the answer is the vendor’s partner portal. It belongs to the vendor, it’s accurate, and it’s maintained by the people who pay you. Treating it as the record is the reasonable conclusion.
The portal shows the numbers the vendor chooses to share, mostly bookings and revenue. The rules that turn those numbers into money live elsewhere, in program guides, eligibility spreadsheets, SKU lists and quarterly updates, sometimes across several vendor systems with different data. Then a check arrives. Everything in between, including what your bookings are worth, how close you are to the next line and whether you have already hit a cap, is left to you to work out, for every vendor you sell.
Microsoft put this in writing when its growth margin went live. Eligibility “is evaluated at the customer tenant level across Microsoft channels and partners and is locked when the transaction finishes,” the announcement says, so “validating eligibility before every quote remains an important step.” The product-level details and rates sit in a “gated growth margin guide.”
Microsoft gives partners a check for it in Partner Center. Running that check, on every quote, is the partner’s job.
That’s one vendor’s program. The same gap shows up across the vendors you carry.
Portals work as designed. They report status, and status is a summary. The rules that turn status into money run on the vendor’s side, in systems a portal was never built to expose. So the portal is right about what it shows and silent about what it doesn’t, and silence looks exactly like nothing being wrong.
Imagine your money spread across ten banks. One charges a fee if your average daily balance falls below $10,000, so a low week can still be fixed. Another charges if you dip below the minimum on any single day, so once that happens, extra money there is wasted and better moved elsewhere.
One counts checking and savings together; another counts only direct deposits. One pays a higher rate on your whole balance once you cross a line, and another stops paying above a cap. Each runs its own statement month, and each sends its rules in a separate document that changes during the year.
Every bank shows you an accurate balance. None of them tells you where you stand against its rules, what the next deposit is worth, or which bank it should go to. That is a solution provider’s vendor portfolio.
Each one is a mechanism, and each ends with what it does to you.
A published partner-tier label can stay displayed and unchanged after the requirements behind it have lapsed. You keep quoting and forecasting against a status you no longer hold, and the first hard evidence is a payment that lands under the number finance already booked.
A program can appear live in your vendor portal and the vendor can later disclaim it as a display error rather than a live commitment, with no warning beforehand. Every deal you steered toward it in the meantime earns at a rate you weren’t aiming for.
It can sit held on the vendor’s side, and the portal doesn’t show you that it’s stuck. Nothing was rejected, so nothing entered a queue, and the money stays where it is until somebody outside that system happens to ask about it.
Rebate activation can require a specific tier held separately in each sub-category. A category sitting under that threshold earns nothing for as long as it sits there, while you read your top-level status and reasonably assume it’s covered.
The incentive number a program displays can be absorbed in large part by customer-side offsets and credits before it reaches your account. You accrue against the headline and reconcile against the remainder, and the gap surfaces after the period has closed.
Exclusion rules can pull a deal out of rebate eligibility because it falls into a coded category: a renewal rather than new business, an attached service line, a particular product family. Your sales team sees a won deal, your rebate line sees nothing, and no report you currently run puts those two facts on the same page.
Some programs count a deal only if it ships inside the program’s window, whatever the booking date. In one case we worked, a Cisco shipment delay cost a partner more than $100,000 in rebate on a booking of more than $2 million, because the order shipped a few days after the “shipped” status window had closed. A request to escalate and expedite the order would have kept that money in the partner’s pocket.
None of these requires anyone to have made a mistake. In each one the portal and the money disagree, and the portal is the thing your team checks.
A note on the evidence, because you should ask. These mechanisms come from work inside the programs of real solution providers, with the partners’ names taken out. We can give you the mechanism and we can’t give you the prevalence, and anyone who hands you a percentage on this is guessing or selling. What makes each one worth your time is that you can check it in your own systems in minutes.
Ten major vendors, eight year-end dates, and eight of the ten land between late July and late January. HPE, HP Inc and Broadcom close within a day of each other. If you sell all three and sit near a line at each, the year-end decisions compete for the same customers, people and credit in the same week, and no single vendor’s portal shows you that trade-off. The table in Sources gives each vendor’s year end, taken from each company’s own published reports.
Four of the ten define the year end as a weekday rule rather than a fixed date, so it lands on a different date every year. A calendar copied forward from last year gets all four wrong every time: we computed the rules for 2020 to 2030, and a copied date never once comes out right. Broadcom’s “Sunday closest to October 31” ends its year in November in five of those eleven years, 2026 among them.
Dell’s “Friday nearest January 31” has ended its year in February, most recently on February 2, 2024. And Dell and NVIDIA name each fiscal year after the January it ends in, so their “Fiscal 2027” is mostly calendar 2026. Join that label to calendar-year sales data and your numbers are out by a year.
That gets filed under reconciliation, and it belongs under margin. Where a threshold is measured across the fiscal year, the year end is a reset line: when the period closes, the count starts again at zero. If you don’t know a vendor’s year end is ten days away, you can’t know you’re sitting just under a tier that’s about to reset. The small push that would have carried you over doesn’t get made, because nobody knew there was a decision to make.
So the deadline behind your largest margin decisions is invisible and moving at the same time: invisible because no vendor portal shows your distance to the line in a form you can act on, and moving because four of these dates land somewhere new every year.
Now put several on one page. One real solution provider’s table, anonymized: six vendor tier deadlines converging on the same year-end window, each row with its remaining distance to the next tier. One of the six was a single order away.
You can build that table yourself, and if you have one, you did, because a single vendor’s portal can’t see your other vendors’ data. That’s how the systems are built, not a choice anyone made.
And this is the half that gets forgotten. Money that didn’t arrive is the easier half to work, because chasing a missing payment has an obvious owner. The other half is the money a different decision would have created: a little more volume before a period closes, one more person certified before a status is measured, a deal registered before someone else claims it.
Nobody owes you that money today. It becomes yours only if somebody acts in time. Ask who owns that in your company.
Finance thinks in percentages, and percentages slide. A rebate threshold doesn’t.
A vendor rebate program can be built as a hard floor. Falling short of a minimum tier or annual purchase amount, by any margin, can forfeit the whole rebate rather than reduce it in proportion. On some programs a narrow miss produces nothing at all. The same shape shows up inside multi-part requirements: missing one component by its deadline can end eligibility for the whole program, not just the missed piece.
So we’re close isn’t the reassuring sentence it sounds like. Close is either over the line or it’s zero, and which one gets decided on a date that four of your vendors move every year.
Illustrative: a partner is at $4.7 million against a $5 million tier, three weeks before the period closes. On a program where the higher rate applies to all qualifying volume, say 1% rising to 2%, the next $300,000 of sales is worth about $53,000 in additional rebate, far more than the margin on the deal itself.
The arithmetic: at $4.7 million the rebate is $47,000, and at $5 million it’s $100,000. On a step, what a move is worth depends on which side of the line it puts you, and its size matters only as far as it gets you there.
A cap is the mirror image of a threshold. Illustrative: a partner reaches the cap on one vendor’s growth incentive in October. Further sales there earn no rebate, but the team keeps steering deals to that vendor because nothing tells it otherwise.
A competing vendor, where the partner is $200,000 from its next tier, would have paid on the same deal. No single vendor’s portal can show you that the next deal should go elsewhere.
Everything above argues that the portal and the money come apart. Here’s when that doesn’t reach you, because it tells you whether any of this is your problem.
If you carry one vendor, the calendar argument doesn’t reach you. There’s no convergence to miss and no second calendar to reconcile against.
The mechanisms still do: a tier label can go stale, an approved rebate can sit unpaid, a gross figure can arrive net, a booked deal can be excluded after the fact. One vendor makes the problem smaller. It doesn’t close the gap.
If your volume sits far below the nearest threshold on every program you’re in, the timing arguments don’t reach you either. Nothing you decide this quarter changes which side of a line you land on.
If none of your programs is scored per portfolio, per sub-category or per product line, your headline status is your actual status.
The second lets you skip the timing sections. None of the three excuses you from checks 1 and 2, because unpaid approvals, gross-versus-net gaps and excluded deals happen at any volume.
On a normal decision, a quarter of delay costs you a quarter. Here it costs a whole measurement period. Where thresholds and tiers are measured across a period, they reset when it closes, so a quarter spent deciding still gets measured and paid without anyone acting inside it. There’s no version of that period you can go back and improve.
This year’s changes show both shapes. From November 1, 2026, HPE and Juniper partners work under one program, and November 1 is also the first day of HPE’s fiscal year, so whatever the new program measures, it measures from that day. Microsoft’s growth margin works deal by deal instead: eligibility is “locked when the transaction finishes.” There’s no period to catch up in, only quotes that were checked before they went out and quotes that weren’t.
That leaves the case that matters most: a period you’re inside right now, because it’s the only one you can still change. From November 1, HPE and Juniper partners are inside the new program’s first period, and on Microsoft’s growth margin every quote that hasn’t gone out yet can still be checked.
You’ve probably seen a figure for how much money the channel leaves behind. We’re not going to give you one.
The figures in circulation that we’ve found all come from companies selling the fix, ours included, and a figure from us would be worth exactly what those are worth.
We went looking for an independent one. We searched accounting and audit literature, the standard-setters, SEC correspondence and filings, oversight material and channel-association research, and we didn’t find a study that measures how much vendor consideration partners earn and fail to collect. That’s a search result, not a claim to universal knowledge.
Regulators do examine this accounting, and companies do disclose when it goes wrong, as Office Depot did. What nobody publishes is a denominator.
So every hard number in this report comes from a public company’s own filing or announcement, from named third-party research, or from arithmetic you can redo from those, and each source is dated and linked below. The exceptions are the bank analogy and the two illustrative examples, whose numbers are made up, and the shipment case among the mechanisms, which comes from our own work. You can check every other number without taking our word for anything.
Two things this report won’t do. It won’t print any vendor’s rebate rates, tiers or thresholds, and it won’t rank one vendor against another.
Each one is something you or someone on your team can do inside your own systems, and none of them requires buying anything. They’re ordered by what they cost you to run, cheapest first.
How: in the vendor portal, look for a report called exclusions, not eligible or clawback, by vendor, over the last four quarters. It isn’t always filed under rebates. If you can’t find it, ask your vendor’s incentives desk for it by name.
What it reveals: whether deals you booked were later pulled out of rebate eligibility, and how much. If several vendors can’t produce one, that’s the finding. You’re accruing against a number with a subtraction in it that nobody on your side can see.
How: approved lives in the vendor portal; received lives in your accounts receivable or general ledger. Lining the two up is the whole difficulty, so match on the vendor’s own claim or credit reference rather than on amount or date.
What it reveals: every line where approved is greater than received. That gap is money that cleared the vendor’s own approval and then stopped somewhere.
How: for your top programs, ask whether the program is activated and generating rebate for you this quarter. Activation in a prior quarter doesn’t always carry forward.
What it reveals: the answer, if you get one. A rep may be bound by policy and unable to put it in writing. Treat that as a gap to note rather than a verdict on the rep: it means the commitment you’re forecasting against isn’t written down anywhere you can point to.
How: a vendor certification can be held by a named individual rather than by your company. For each specialization that gates rebate eligibility, confirm there’s a second certified person behind it.
What it reveals: which parts of your rebate line depend on one person’s continued employment and one certification’s renewal date.
How: one row per vendor program you’re in, with its period end date taken from that vendor’s own filing or program document, not from last year’s calendar, and your current distance to the next threshold in dollars. Where a program is scored per portfolio rather than company-wide, give each portfolio its own row, because a headline tier can sit above a line that a single portfolio sits below. Review it monthly, starting ninety days before your earliest date.
What it reveals: which vendor is closest to a boundary, and what it would take to cross it. It’s the only one of the five that tells you what to do rather than what happened.
And one question before any of the five. You can almost certainly say how many vendors you manage. Can you say how many separate programs you’re measured on?
The second number can be much larger, because a single vendor can run several sub-programs, each scored on its own indicators. If the first number exists and the second doesn’t, reconcile them first. The programs you aren’t counting are still being measured.
If you’ve run all five, you’re instrumented, and this report was confirmation. If you’ve run two or fewer, it isn’t a judgment about your team. It means this runs on knowledge rather than on a system, which works until the person holding it is on vacation, or leaves, or the program changes underneath them.
Six criteria. They apply to a spreadsheet, an in-house build, a consultant and to us, equally. Use them on us.
A tool that answers for one vendor can’t tell you where your attention is worth most this week.
A spreadsheet holds what somebody knew on the day they built it. It doesn’t know that a weekday-rule year end moved, or that a program was restructured in August.
If the answer lives in one person’s head, you don’t have a system. You have an arrangement.
Reporting the gap after the period closed is history. The value is in the window when the gap was still fixable.
Most tools in this category grew up serving the company paying rebates out, not the company collecting them, and their workflows and data model still show it. If you’ve evaluated tools that felt like a poor fit, that may be why: you were being shown the other side of the table.
Below roughly $50 million in revenue, or with fewer than three vendors that matter, don’t buy anything. At that size a spreadsheet is the right answer, and anyone who tells you otherwise is selling.
If you’ve read this far, you’ll have questions for us. Some have short answers here.
How many partners we’ve done this with, who you can speak to, whether we contact vendors on your behalf and how soon you see the first output are questions we answer in a conversation.
Go to the table in Sources and find the first year end after today’s date. Count the program changes in this report too, such as HPE and Juniper’s on November 1, 2026: they’re separate from year ends, and they move the rules.
That date is your deadline, and check 2 is your first step: approved against received, one vendor, the one closest to its boundary. Start there rather than with check 1 because it uses data you already hold, while check 1 can depend on a vendor producing a report. It takes an afternoon and it needs nobody’s permission.
Everything else in this report can wait until you’ve seen that number.
Every quotation in this report was checked word for word against its source.
The ten fiscal calendars. Nine come from the company’s annual report on Form 10-K. Lenovo is listed in Hong Kong, so its row comes from its own full-year results. The middle column states each company’s rule; the last column is computed from it.
| Vendor | Fiscal year ends | Year ends, 2026 to 2028 |
|---|---|---|
| Cisco | The 52 or 53 weeks ending on the last Saturday in July | July 31, 2027 · July 29, 2028 |
| Nutanix | July 31 | July 31, 2027 · July 31, 2028 |
| HPE | October 31 | October 31, 2026 · October 31, 2027 |
| HP Inc | October 31 | October 31, 2026 · October 31, 2027 |
| Broadcom | A 52- or 53-week year ending on the Sunday closest to October 31 | November 1, 2026 · October 31, 2027 |
| IBM | December 31 | December 31, 2026 · December 31, 2027 |
| Dell Technologies | The 52- or 53-week period ending on the Friday nearest January 31 | January 29, 2027 · January 28, 2028 |
| NVIDIA | A 52- or 53-week year ending on the last Sunday in January | January 31, 2027 · January 30, 2028 |
| Lenovo | March 31 | March 31, 2027 · March 31, 2028 |
| Microsoft | June 30 | June 30, 2027 · June 30, 2028 |
The dates in bold were computed from each company’s rule rather than copied. October 31, 2026 falls on a Saturday, so Broadcom’s “Sunday closest to October 31” is November 1: a year everyone calls October ends in November. January 31, 2027 falls on a Sunday, so NVIDIA’s “last Sunday in January” lands on the 31st, a coincidence of that year rather than a rule to carry forward. Dell’s “Friday nearest January 31” is January 29, 2027, and Cisco’s “last Saturday in July” is July 31, 2027.
Rebates-On (rebates-on.com) watches your vendors’ rebate, incentive, certification and compliance programs across the vendors you sell, on the partner’s side of the table rather than the vendor’s. We wrote this for US IT solution providers at $50 million and up carrying three or more vendors, the same line criterion 6 draws.
This report is our professional opinion, not advice. Each vendor always has the final say on its own programs.