The short answer. Sales commissions is one of the best apps for a company to build for itself. Your plan is specific to you and changes every year, the inputs already live in systems you own (CRM deals, HRIS, payroll), and packaged tools charge per payee. The catch is that the app decides what people are paid, so you need an owner, an audit trail, statements reps can check, a way to handle disputes and finance sign-off. Tray Helix gives you the running side. Build it when your plan is your own, and buy it when your plans are genuinely complex.
What a commissions project looks like today
A software company has 140 reps. Quota carriers earn 8% on new business up to quota, 12% above it. Renewals pay 2%. Deals over $250,000 split with a sales engineer 80/20. New hires get a three-month ramp. That is the plan, and it fits on a page.
Now watch what happens when the company buys a commissions product to run it. A consultant spends weeks turning that page into plan templates, crediting rules and data mappings. The CRM and HR data get copied into the tool. Then January arrives, the plan changes, and every change is another ticket to the person who knows the tool.
Meanwhile the reps still do not trust the statement. A rep who sees $8,400 when she expected $11,000 wants to see which deals were counted and why. If the tool shows only a total, she opens her own spreadsheet, and so does her manager. The commission data and payroll problem is not the arithmetic. It is showing the working.
That pattern is why commissions is a strong candidate for building your own. Four things point the same way.
The plan is yours, and it changes every year
Commission plans are about as company-specific as a business document gets. Rates, tiers, accelerators, caps, splits, ramps, clawback windows and the date that makes a deal count are all your choices, and most companies rewrite some of them every year.
Packaged software is a general engine. It carries a large vocabulary of plan constructs, and your job is to translate your page into that vocabulary. Each translation is a configuration project, and each annual change is another one.
Most plans are simple to say in plain language even when they are tedious to configure. “Renewals pay 2%, and deals over $250,000 split 80/20 with the sales engineer” is close to a build instruction for an AI assistant. The value in this app is your own policy applied the same way every time, not deep shared functionality like tax or payroll.
The inputs already live in systems you own
A commission run needs three kinds of data. Closed-won deals, with amounts, dates and splits, sit in your CRM, and most of those numbers were set at the quote: the price, the discount, the term and the split. A wrong quote becomes a wrong statement, which is why teams that build their commissions app often build their CPQ next. Who is on which plan, with start and leave dates and managers, sits in your HRIS. The approved result goes to payroll.
A commissions product owns none of that. It copies the data in, calculates against the copy, and sends results back, so there is a sync to build and a sync to maintain. An app you build reads the same systems directly. Tray connects to 12,000+ apps, databases, APIs, and services, so Salesforce, Workday and Gusto are connections your app uses, not integration projects to scope. The commission data to payroll guide covers that side.
Where the commission is paid on cash collected rather than closure, the same app can check invoices, as in the order-to-cash process.
Tools priced per payee, and months to implement
Commissions tools are licensed by the number of people paid. A few publish per-user prices, and the larger ones are sales led. We cover what that adds up to in the total cost of buying versus building. Two features of the model matter here.
First, the people paid commission are the sales team, and the cost scales with how many sellers you hire. Add implementation, the partner who knows the product, and the administrator who owns the configuration, and the subscription is rarely the whole bill.
Second, implementation takes time. The plan has to be modeled, the data mapped, the statements tested against the old spreadsheet for a cycle or two. Vendors set their own timelines, so ask each one for theirs in writing.
An app you build and run on Tray Helix has no per-user fees. Hiring another twenty reps does not change what it costs to run. You can start on the Tray AI Orchestration Platform for $50 a month with Tray Helix Builder Edition, plus usage.
For a company of about 2,000 people, Tray’s SaaS savings calculator puts the saving from replacing a typical set of SaaS apps at about 75%. Sales commissions is one of the seven default apps in it. The figure covers the apps you choose, not commissions alone, so use the calculator to size your own case.
Reps distrust statements they cannot check
This is the least obvious reason to build, and the most important.
A commission statement is the most-read report in a sales organization. Reps check it line by line. When the number is a total with no derivation, every rep who disagrees starts a ticket, and answering tickets costs more than the calculation.
Building your own lets you decide what a statement shows. Each qualifying deal, the plan version applied, the tier, the rate, the split and the amount. Adjustments with their reasons. A way for a rep to query one line rather than email about the total. You can design for the questions your reps actually ask, because you have heard them.
Many teams are deciding anyway
Salesforce acquired Spiff, a provider of incentive compensation management software. Salesforce’s own announcement says Spiff’s team joins Sales Cloud to strengthen its Sales Performance Management solutions, and that the acquisition completed on February 1, 2024. If you used Spiff, or you are weighing the category, that is context for a decision you were going to make anyway: renew, switch, or move the rules into an app you own. For the category from the start, read what is incentive compensation management.
What it takes to run your own
Building a calculator with an AI assistant is the easy part. A first version that applies a plan to a list of deals can come together in days. What separates a demo from something you can pay people with is everything around it. The app decides what people are paid, so it needs these from day one.
- An owner. One named person answers when the plan changes or a number is wrong. When the builder changes jobs, the app must not become nobody’s problem.
- An audit trail. Every plan change is a new version with who made it and when. Every calculation is stored with its working. Six months later, someone will ask why a deal paid what it did, and the answer cannot live in a spreadsheet or a Slack thread.
- Statements reps can check. Each rep sees their own statement and nobody else’s, except their manager.
- A dispute process. A query attaches to a line, routes to the plan administrator, and the resolution is recorded against it. Held lines do not hold the run.
- Finance sign-off. Finance reviews the period before payroll, and the period freezes once approved. Later corrections become adjustments in the next period, never a silent rewrite.
These are what a vendor supplies without anyone noticing, and they are where self-built apps stall. Our view in what it costs to replace a SaaS app is that the running cost, not the build cost, decides whether replacing an app is worth it.
Where Tray Helix and Tray fit
Tray Helix is where an app built with an AI assistant gets deployed, run and governed. Your RevOps lead builds the commissions app in Claude Code, Codex or Cursor. One command deploys it to a managed runtime and returns a URL. Nobody stands up servers.
Every deployed app sits behind sign-in, and access is granted per project. The app does not hold raw credentials: it names the system it needs, such as Salesforce or Workday, and Tray resolves the authentication on the server side. The app gets a built-in database, so calculation lines and plan versions have a place to live. Single sign-on, audit history and governance come with Helix Enterprise Edition, so check that edition fits before you rely on them for a payroll-adjacent app. Helix hosting is metered usage, and there are no per-user fees.
Tray also covers the integration side. The same platform that runs the app connects it to your CRM, HRIS and payroll, so a closed deal reaches a statement and an approved period reaches payroll without re-keying.
Helix does not make the app good. The quality of your commissions app depends on how clearly you write down your plan, and that remains your team’s work.
Add an agent that answers the questions reps ask
The questions that fill a comp admin’s inbox have answers in the app’s own data. Why is my October payout lower? Which deals counted toward quota? When does the accelerator start? Tray Helix lets you build agents where your apps already live, so the commissions app can carry its own agent. It is defined in the app’s code, deploys with the app and works through the app’s own actions, such as reading a rep’s credited deals, plan version and adjustments.
You decide what each action can read, so write them to return only the asking rep’s lines. Conversations can be kept separate for each signed-in user, so one rep’s questions are not visible to another.
Anything that changes money waits for a person. The agent can draft a dispute or propose an adjustment, and you set those actions to wait for the comp admin’s approval before they run. A schedule can run the same agent before each payroll cutoff to flag lines worth a second look, such as a deal credited twice or splits that do not add up to 100%, for a person to review. Every agent run and every action it takes shows in the project’s logs. How agents work in Tray Helix.
When you should buy
Buying is the right answer in several cases, and a fair comparison says so.
- Many plan types and territories. If you run dozens of plans with overlapping crediting, hierarchies and quota cascades, a mature product has already worked through the edge cases.
- Heavy commission accounting. If finance needs formal capitalization and amortization of commissions under ASC 606, check that your build can meet it before you start. A vendor may already do it.
- Very large payee counts. Thousands of payees, several currencies and local pay rules raise the cost of every mistake.
- No one to own it. If nobody will keep the plan current and answer when it breaks, a vendor’s support desk is worth paying for.
- A tool that works. If your current tool runs well and the renewal is fair, moving is a cost with no gain.
If two or three of these describe you, buy. If none do, building deserves a serious look.
How to start
Do not start with the whole quote-to-pay flow. Pick one plan, write it down in plain language, and run the new app beside your current process for one full period. Compare every statement before anyone is paid from it.
The prompts for each part, and what it takes to run the result, are in the Helix guide to how to build a commission calculator. If you are comparing products first, see CaptivateIQ alternatives and Xactly alternatives.
Sources
- Salesforce, “Salesforce Signs Definitive Agreement to Acquire Spiff” (salesforce.com/news/press-releases/2023/12/21/salesforce-signs-definitive-agreement-to-acquire-spiff/), read November 14, 2026. Source for Spiff’s team joining Sales Cloud, Sales Performance Management, and completion on February 1, 2024.
- Tray, SaaS savings calculator, read November 14, 2026.