What a stack looks like once the numbers sit together
Most teams know each tool’s price and almost none know the shape of the whole. That shape is where the decisions are. A stack is rarely too expensive across the board; it is usually one line carrying a third of the total, a middle group that is roughly fair, and a long tail of small subscriptions nobody has thought about since the day they were signed. Those three groups need completely different treatment, and you cannot tell them apart one invoice at a time.
The largest line deserves the least suspicion
The instinct is to attack the biggest number first. Usually it is the one most worth its price: it is large because it does something central, and the value scales with what you are paying for. The real question for the top line is not “can this be cheaper” but “are we on the right shape of plan” — per seat when we should be on usage, or the reverse. Deel prices the same worker at $14 a month for HR tooling and $599 for employer-of-record; the person has not changed, only which liability the vendor carries. Being on the wrong model costs far more than being on the wrong tier.
The long tail is where the free money is
Small subscriptions are the ones that survive scrutiny because each is individually defensible. Six tools at $29 a month is $2,088 a year, which would never be approved as a single purchase. The tail is worth reviewing not because any one item is wasteful but because it is the only part of the stack where cancelling costs almost nothing: little data to migrate, few integrations, no retraining. Start there when you need a result this month.
Per seat, per usage, per outcome
Three pricing models behave very differently as you grow. Per-seat cost rises with headcount whether or not usage rises with it, and it drifts upward on its own because licences outlive the people who had them. Usage pricing tracks activity, which is fairer and much harder to budget — Ava AI sells $29, $99 and $499 tiers that are really buckets of call minutes, with overage at $0.07 to $0.10 a minute beyond them. Outcome pricing, still rare, charges for results. Knowing which model each line uses tells you what your bill will look like at twice your current size, which is the only forecast worth making.
Annual billing hides in every total
If some of your numbers came off pricing pages rather than invoices, part of your total is probably wrong in a predictable direction. Pricing pages open on annual billing with the discount applied, so the displayed figure is the yearly-commitment rate. And the discount is not uniform: reading Aircall from a US connection on 29 August 2026, Essentials went $40 monthly to $30 annually — exactly the advertised 25% — while Professional went $70 to $50, nearer 29%. Use what you actually pay, which is why this tool asks for your figures instead of guessing them.
The tools that refuse to be counted
Some vendors publish no price at all. Rippling’s pricing page is a quote request; ZoomInfo is sales-gated and will not serve plain requests. In a diagnosis these lines matter twice: once for what they cost, and once because they are the hardest to benchmark or replace. A stack heavy in quote-only vendors is a stack with limited negotiating information, and that is worth knowing even when every individual deal is fine.
What the diagnosis cannot tell you
It works from what you type and what our research knows about each tool. It cannot see usage, so a tool nobody opens looks identical to one running your whole operation. It cannot see contracts, so an annual commitment with eight months left looks as cancellable as a monthly plan. And it cannot see the political cost of removing something people like. Read the output as a map of where to look, then check the two or three lines it highlights against reality.
Turning the diagnosis into a decision
Three follow-ups cover most of it. For lines that look duplicated, the stack auditor works through overlap specifically and which of two tools should survive. For anything you are considering adding rather than cutting, the ROI calculator estimates what it needs to return before it pays for itself. And for the tools you decide to keep, put the renewal dates somewhere you will see them — an annual plan gets audited when it renews, which is often eleven months after the seat count stopped being right.
Frequently asked questions
Do my numbers leave the browser?
No. Everything you enter stays in the page: we never receive it and never store it. That is also why the tool cannot remember your stack between visits.
Why don’t you fill in the prices for me?
Because a published price we hold is a price that goes stale, and a stale figure presented as current is worse than an empty field. Vendors change tiers, run promotions, and price differently by country — we have found tools charging the same number in euros and dollars depending on where the page is read from. Your invoice is the only figure guaranteed to be true for you.
How is this different from the cost audit?
The cost audit answers “what am I really paying, and is there a cheaper equivalent for each line”. This page answers “what does my stack look like as a whole, and which part of it should I be looking at”. Same inputs, different question.
How often is this worth doing?
Twice a year, plus whenever headcount moves sharply. The failure mode is not doing it too rarely — it is doing it once, finding nothing dramatic, and concluding it is not worth repeating. Overhead creep is gradual by definition.