The Things That Break While Nobody Is Looking At It
Nothing in your app changed. The operating system did, the certificate expired, a supplier retired an API and the store tightened its rules. A retainer puts someone responsible for all of that — with the limits written down rather than implied.
Software Fails Quietly Long Before It Fails Loudly
The expensive part of an unmaintained app is not the outage. It is the weeks between something breaking and somebody noticing. A payment webhook stops arriving and the orders keep looking fine on screen. A push certificate lapses and notifications simply stop, with no error anywhere. An Android release changes a permission and one in five customers quietly cannot upload a photograph any more.
Picture a construction firm whose site-inspection app quietly stops syncing, and which finds out when a client asks for eleven weeks of records that no longer exist. Nothing crashes. Nobody is alerted. That is the failure mode maintenance exists to prevent, and it is why monitoring rather than availability is the first thing we set up.
What Arrives Whether You Budgeted for It or Not
A rough but realistic year for a live product with a mobile app and a web back office. None of it is caused by your code. All of it lands on your code.
Month one — a dependency you have never heard of gets a CVE
A library four levels down your dependency tree is found to have a vulnerability. Nothing breaks, and if you sell to enterprises or the public sector, their scanning tools will find it before you do and the question will arrive by email from a procurement officer. Patching it is twenty minutes. Explaining why it sat there for five months is considerably longer.
Month three — a payment provider deprecates an API version
Stripe, GoCardless and the Open Banking providers all move forward, and they give notice. The notice goes to the email address on the account, which is frequently a developer who left. Miss it and the failure lands on the day the old version is switched off, which is invariably a day you are busy.
Month five — the annual iOS and Android releases
New OS versions arrive every autumn and your users update within weeks. Layouts shift, permission prompts change wording, background behaviour tightens. Most apps need a modest compatibility pass. Apps that skipped the previous year need a considerably less modest one.
Month seven — a certificate or key expires
Push notification certificates, signing keys, API credentials and TLS certificates all have expiry dates, and they are the purest form of avoidable outage: a diary entry nobody made. We keep the renewal dates on a list with owners against them, which is unglamorous and prevents more incidents than anything clever.
Month nine — a store policy change forces a resubmission
Apple and Google regularly tighten requirements around privacy labels, data declarations, account deletion, target SDK levels and age ratings. A submission that sailed through last year gets rejected this year. The work is small; the surprise is the problem, because it usually collides with a release you had already promised somebody.
Month twelve — a component stops being maintained
An open-source package is archived, or a vendor closes a product line. There is no outage and no error. There is simply no future, and the longer you leave it the more the replacement costs. Spotting these early is the least visible and most valuable part of a retainer.
Which Bucket Any Given Request Falls Into
Most arguments about maintenance contracts are really arguments about categories. Here are ours, before you sign anything.
Covered as Standard
Uptime and error monitoring with alerting. Security patching of dependencies and framework versions. A tracked list of every certificate, key and credential with its expiry date. Triage and diagnosis of any defect you report. Compatibility work for new iOS and Android releases. Store resubmission when a policy change requires one. Backup verification, meaning an actual restore rather than a green tick. A written update at the close of every working day on which something happened, and a monthly summary of work done and work coming.
Sized by the Retainer
The number of development hours available for fixes and small features. How quickly we start on a serious issue. How often we release — monthly on the smallest band, fortnightly on the largest. Whether a staging environment is maintained alongside production. How much performance and cost-optimisation work we do proactively. How many scheduled calls sit in the month.
Costs That Are Not Ours to Mark Up
Cloud hosting, on an account in your company name. The annual Apple Developer Programme fee and the Google Play registration. Paid SDKs, mapping, SMS and email providers. Error-tracking subscriptions. Penetration testing by a specialist firm. We will set all of it up and manage it, and we would rather you saw the real bill than a re-invoiced one.
Not Maintenance at All
A new module, a second platform, a redesign, a migration off a legacy database, replacing your payment provider, or building the reporting suite finance keeps asking for. These are projects. They get a written scope and a fixed price in sterling excluding VAT, and they run alongside the retainer rather than eating it. We also do not provide tier-one support to your own customers — we support your team, not your end users.
Three Sizes, Fixed Monthly in Sterling
Indicative monthly figures, excluding VAT, invoiced monthly in advance. The right band depends on how much your product is still changing, not on how large your company is.
Caretaker
£850 / month ex VAT
Around 8 development hours. For a stable product with steady usage that mainly needs watching and keeping current.
- Monitoring, alerting and error tracking
- Security patching and dependency review
- Certificate and key expiry tracking
- Defect fixes within the hours allocation
- Monthly release window
- One scheduled call each month
Active
£2,200 / month ex VAT
Around 24 development hours. For a product with real customers where small improvements need to keep shipping.
- Everything in Caretaker
- Small feature work inside the allocation
- Fortnightly releases
- Staging environment kept in step with production
- Quarterly review of cost, performance and risk
- Two scheduled calls each month
Embedded
£4,400 / month ex VAT
Around 50 development hours. For a product that is central to how the business runs and still changing every month.
- Everything in Active
- A named engineer who knows your codebase properly
- Releases on demand rather than to a window
- Proactive performance and hosting cost work
- Evidence packs for client security questionnaires
- Weekly call and a roadmap kept up to date
Figures are indicative starting points for a product of ordinary size, not a price list. A three-month minimum applies, then 30 days notice either way. Unused hours do not roll forward — you are paying for availability and vigilance, not buying a block of time. Full detail on the pricing page.
Targets We Can Meet Rather Than Targets That Sound Good
These sit inside UK working hours. Read the row about weekends before you compare us with a supplier promising an hour at any time of day.
| Severity | What it looks like | Acknowledged | Work starts | Realistic resolution |
|---|---|---|---|---|
| Critical | The service is down, payments are failing, or personal data is exposed | Within 2 working hours | Same working day | A mitigation the same day where one exists; a full fix often next day |
| High | A core journey is broken for many users but the service is up | Within 4 working hours | Within 1 working day | Usually 1 to 3 working days |
| Medium | A feature misbehaves, or a workaround exists | Within 1 working day | Into the next release | The next scheduled release |
| Low | Cosmetic problems, small annoyances, wording | Within 1 working day | When it fits the allocation | Batched, typically within a month |
How the working day actually lines up
The delivery team works to Indian Standard Time, which overlaps the UK working day from roughly 9am to early afternoon. Anything raised in that window gets attention the same day. Something reported at 4pm UK time is picked up the following morning — that is the honest shape of it, and it is why the targets above are stated in working hours rather than clock hours.
What we will not claim
There is no staffed 24/7 on-call rotation. Nobody is paid to be awake at 3am waiting for your alert, and any supplier of our size telling you otherwise is describing a phone on a bedside table rather than a rota. Monitoring runs around the clock and a lot of failures recover automatically; a human response is a working-hours commitment.
The Unexciting Work That Prevents the Exciting Kind
Monitoring and error tracking
Uptime checks from outside your network, error tracking that groups incidents rather than emailing one line per failure, and alerts routed to a person rather than an inbox nobody owns. Crucially we also watch for silence — a job that stops running produces no error at all, which is why absence of activity is itself an alert.
Security patching
Dependency updates reviewed and applied on a schedule rather than in a panic, with anything urgent pulled forward. Framework versions kept inside their support window so you are never one release from an unsupported platform. Where you sell into regulated sectors, we keep the evidence a buyer will eventually ask for.
Backups you have actually restored
A backup nobody has restored is a belief, not a backup. We restore into a scratch environment on a schedule and record how long it took, because the number you need during an incident is recovery time, not storage size. Retention is set against what UK GDPR allows you to keep rather than what the provider defaults to.
Data protection does not stop at launch either. Retention rules, the record of processing and the route for servicing a subject access or deletion request all drift as a product changes, so they get reviewed as part of the retainer rather than the week before someone asks.
Where the Line Sits Between a Tweak and a Project
You can change things inside a retainer. The distinction is not how important the change is to you, it is how much of the system it touches.
Comfortably inside the retainer
- A new field on an existing form, and the report column that goes with it.
- Changing VAT handling on invoices, or adding a purchase order field for corporate customers.
- An extra filter on an admin list, or a saved view for a team that keeps asking.
- Copy, pricing, email templates and notification wording.
- A new export format because a client or an auditor asked for one.
- Sorting out a slow page that has been irritating everybody for a month.
Needs its own scope and price
- A customer-facing mobile app where there was only a web product.
- Multi-branch or multi-site support in something built for one location.
- Swapping payment provider, or adding Direct Debit alongside cards.
- A reporting and analytics suite rather than another report.
- Single sign-on, role hierarchies or an audit trail for a corporate buyer.
- Anything an investor, a regulator or a procurement process has made a condition.
When something falls on the wrong side of the line we say so the same day and quote it separately rather than quietly consuming three months of your allocation. See how fixed-price work is scoped.
When a Retainer Is the Wrong Purchase
You need continuous delivery, not cover
If you have six months of roadmap and want work shipping every fortnight, a retainer is an expensive way to buy capacity. A dedicated team costs less per hour and gives you people who are genuinely yours.
The product is fundamentally unwell
Maintaining something that fails constantly is paying us to stand next to a problem. Where an audit says the foundations are the issue, fix that first. That is a rescue job, not a retainer.
You already have engineers
If you employ developers who know the product, you do not need us watching it. You might need occasional senior oversight instead, which is what fractional CTO work is for.
Retainer Questions We Get Asked
Are those monthly figures inclusive of VAT?
No. Every figure on this page is quoted excluding VAT, and VAT at the prevailing rate is added on the invoice. Retainers are invoiced monthly in advance in sterling, with a purchase order number on the invoice if your finance team needs one. Hosting and third-party licences are billed to you directly by those providers rather than passing through us with a margin on top.
What actually happens if the app goes down at 7pm on a Friday?
Our monitoring raises the alert immediately and, if the service can recover itself, it usually does so without anyone being woken. What does not happen is a person picking up a phone at 7pm, because we do not staff an overnight rota and we are not going to pretend otherwise. A weekend outage is normally picked up at the start of the next working morning UK time. If your business genuinely cannot absorb a weekend outage, the right answer is not a bigger retainer from us. It is designing the system so that a failure degrades rather than stops, and telling you that honestly is cheaper for you than selling you cover we cannot staff.
Who pays for hosting, the Apple Developer Programme and third-party licences?
You do, directly to each provider, on accounts in your own company name. That includes cloud hosting, the annual Apple Developer Programme fee, the one-off Google Play registration, error tracking and any paid SDKs or APIs. We would rather you saw the real numbers and kept the accounts than have them routed through us. The retainer covers our time managing those services, not the cost of the services themselves.
We supply the NHS and have to keep the Data Security and Protection Toolkit current. Does the retainer help with that?
It helps with the technical half. We can supply the evidence a submission asks for on our side of the line: patching records, access control arrangements, backup and restore testing, encryption in transit and at rest, and a current record of where personal data sits and how long it is kept. What we do not do is complete the submission for you or act as your compliance adviser, because the toolkit covers organisational matters such as staff training and policy ownership that belong inside your own organisation. The same applies to a Cyber Essentials assessment or a client security questionnaire.
Can we pause the retainer during a quiet period?
Pausing is possible but it is usually a false economy, and we will say so before agreeing to it. The work that continues regardless of how quiet your quarter is happens to be the work that prevents emergencies: security patches, certificate renewals, monitoring, and compatibility with whatever the mobile platforms released last month. A paused retainer means those stop while the clock keeps running. Where budget genuinely has to come down, moving to the smallest band is a better answer than a gap, because restarting after a gap usually costs more than the months you saved.
Do you need access to our live database and real customer data?
As little as the work allows. Most diagnosis is done from logs, error reports and a copy of the database with personal data removed or replaced. Where a live investigation genuinely requires production access, it is granted to named individuals, it is logged, and it is time limited. We sign a data processing agreement setting out what we may do with personal data, we keep a record of processing, and we work in UK or EU regions where you require data residency. If a breach ever occurred, you are the controller and the reporting clock is yours, so we would tell you immediately rather than investigating quietly first.
Put Someone Sensible Behind Your Live Product
Tell us what is running, who built it and what worries you about it. We will suggest a band, or tell you that you do not need one yet.