Do the arithmetic on last year
Before you read further, find last year's invoices and the supplier's usage reports and build four columns: hours purchased, hours used, hours lost, and the cost of the hours lost at the rate you paid. Twelve rows, one per month. If unused hours roll within a quarter, count the loss at the quarter end. Then add up the last column.
You have probably never seen that total. You know usage runs at "about half" because someone said so in a review meeting, and you know the package was discounted, so the gap feels paid for. The number on the page is usually larger than the discount, and it is the number to take into renewal. The sheet that goes with this post is laid out for it.
Lumpy demand, flat capacity
The problem is not the size of the retainer. It is the shape. Demand for platform work is lumpy: say a release needs 60 hours one month, a campaign page needs 15 the next, then nothing for six weeks, then an integration partner changes an interface and you need 40 hours inside a fortnight. Monthly capacity is flat. A retainer sells the same block every month because that is how a supplier schedules people, and the block is sized for the average rather than for any month you will actually have.
Buying less next year does not change the shape. It moves the failure from unused hours in quiet months to a queue in busy ones, with overflow bought ad hoc or work that simply waits. Buying per ticket is the other escape. It works at small volumes and breaks the moment a ticket needs design, because nobody can price design before it has been thought through.
Expiry versus discount
Be fair to the supplier first. A team kept available has to be paid whether or not you call, and a flat monthly block lets the supplier staff to a flat line. That is a legitimate interest. Expiry moves the cost of that convenience to you; the package discount is what you get in exchange.
The discount applies to every hour you purchase. Expiry applies only to the hours you did not use. If the sheet shows you lost a quarter of what you bought, a discount of a tenth did not cover it.
This is why we price without package discounts. Unused capacity is banked rather than forfeited, so you keep the value instead of trading it for a lower rate. Non expiring hours are not a premium product: what you give up is the discount, and for most buyers that is the better trade. Banking runs for a defined period, not for ever: the supplier still has a horizon to plan against, and you have a finite balance to review at renewal.
The all purpose pool
The second leak is quieter. Every retainer contract has an exclusions section: support is covered, changes are not, infrastructure is a separate line, a new integration is a new project. Each boundary is a place to argue, and the argument is conducted in billable hours. A fix that needs a small change to stay fixed becomes a ticket about whether it counts. A security update that touches hosting becomes a question for the account manager. The hours that do get used are partly spent deciding what they may be used on.
Our pools are all purpose by design, with no exclusions list. One prepaid pool, one rate, spendable on fixes, infrastructure work, new features or a further integration. For a European pharmaceutical group we deliberately removed the exclusions section from the offer, and the support versus change argument went with it.
Response as a separate promise
Urgency and capacity are different products, and a retainer sells them as one. A response commitment is a promise about how quickly a named person picks up an incident. A prepaid pool is a promise about how much work you can draw. Sold together, every hour in the block carries the cost of standby.
We structure response commitments separately from the capacity itself, so urgency is not paid for twice. The pool funds the work; the commitment governs how fast it starts. Ask for yours the same way: separate, in writing, with the people it binds named. If your IT lead runs an incident process, that is the document it depends on, not the pool size.
Questions for your current supplier
None of this requires changing supplier. Take the sheet and these questions into renewal:
- What happened to the hours we did not use last year, and where does that appear in the price?
- If we drop the package discount, will you bank unused hours instead, and for how long?
- What may the retainer not be spent on, and which boundary caused an argument last year?
- Is the response commitment written separately, with named people, so IT can test it against the incident process?
- What happens to banked hours if the contract ends?
A supplier with a clear answer to all five is worth keeping. A supplier with none is telling you where the margin comes from.
