What You Are Paying For
Capacity and continuity, rather than a defined list of deliverables.
That is the honest description and it is the source of every argument retainers produce. Some months you use all of it and it feels excellent value. Other months you use little and wonder what you paid for. Both are the arrangement working as designed, which is easier to accept when it was described that way at the start.
Three things sit underneath it. Capacity: time reserved, so work starts in days rather than joining a queue. Continuity: the people working on your system already know it, which on anything of moderate complexity is the largest hidden value, because the alternative is paying somebody to relearn it every time. And the maintenance nobody asks for: dependency updates, security patches, certificate renewals, monitoring, backup verification.
The Work You Never See
This is most of it, and it is the part that gets cut first because it produces nothing visible.
Software does not sit still when nobody changes it. Vulnerabilities are published in the components your system is built from, and automated scanners begin looking for them within days. A system left alone for two years has not stayed the same; the world around it moved.
Ongoing patching is what defeats the overwhelming majority of realistic attacks, and it is invisible right up until the moment it has not been done.
It is also why a website that comes to us from another provider is not treated as a clean slate. A switch to us moves the existing site onto our hosting and fixes the one problem the owner names; the patching that was skipped for years before it arrived is retainer work, not part of the move.
What Should Be In It Explicitly
Ask for these to be named rather than implied:
- Dependency and security updates, and how often
- Monitoring, and where alerts actually go
- Backup verification, meaning a restore has been tested rather than a job reporting success
- Certificate and domain renewals
- A response commitment, with the distinction between response time and fix time
- What counts as included work and what needs quoting separately
The Question That Decides Its Value
What happens to unused time?
Three common answers, each describing a different product. Time that rolls over for a defined period turns it into a flexible bank of hours, which suits uneven demand. Time that expires monthly is a pure availability arrangement, fine if availability is the point and poor value if you rarely call on it. Time that rolls indefinitely usually means the supplier expects you not to use it.
None is wrong. Knowing which you bought prevents the resentment that builds when it was never stated.
When A Retainer Is The Wrong Shape
It suits you badly if the system is genuinely finished and stable, if changes arrive in occasional large batches rather than a steady trickle, or if you have no realistic way of judging whether the monthly work happened.
In those cases project work with a separate arrangement for security updates is the better commercial fit, and a supplier worth using will say so rather than push a monthly fee you do not need.
What Moves The Price
Response commitment. Same-day is a different cost base from same-week, because it constrains how the supplier schedules everything else.
Out of hours. Whether anybody answers at the weekend, and what that covers.
How many systems. Each one carries its own dependencies, its own monitoring and its own update cadence.
How much discretionary time is included, as distinct from the maintenance floor.
What Goes Wrong
It becomes invisible. No reporting, so nobody can name what it did last quarter, and it gets cancelled at renewal regardless of value. This is the most common way a good arrangement ends.
The maintenance quietly stops. Discretionary work absorbs the whole retainer, patching slips, and nobody notices until something is exploited.
Scope drifts upward. Small requests accumulate until the retainer is permanently overspent and the relationship sours.
Questions To Ask Before Signing
- What is included that we would otherwise be billed for?
- What happens to unused time?
- What are we sent each month? A retainer with no reporting becomes invisible, and invisible costs get cancelled.
- When did you last verify a restore for a client? Not whether backups run.
- What sits outside this, and how is it priced?
- What happens if we need you at 6pm on a Friday?
A Brief You Can Send Anyone
What we run: [systems, roughly how big, what they do]
Who uses them: [internal, customers, or both, and how many]
What breaking costs us: [per hour, honestly]
What we need help with now: [the actual current pain]
How much change we expect: [steady trickle, occasional batches, or none]
Who currently maintains it: [including “nobody”]
Hours we need cover for: [business hours, extended, or genuinely 24/7]
What Good Looks Like
The strongest sign a retainer is working is that you can name what it did last quarter without asking. If nobody in the business can, it has become a subscription rather than a service, and that is worth fixing before the renewal rather than at it.
Ours are structured with the maintenance floor and the discretionary time reported separately, so both are visible. Get in touch if that is the shape you need, and see the wider services if it is not.
If you are weighing a retainer against employing someone, hiring a developer vs an agency retainer and outsourcing vs hiring your first developer compare the options. If the system was built with AI and a developer has just reviewed and signed off the current version, a retainer is how we stay accountable for it from then on, because the sign-off itself covers only the version we reviewed.