Cost

What it costs to keep a bot alive

Maintenance is a monthly retainer rather than a project, and the work is mostly other people's changes: Telegram deprecations, third-party API versions, expiring credentials and security updates. An unmaintained bot does not degrade gradually. It works perfectly until a token expires, then stops entirely.

What it costs to keep a bot alive: effort, cost and what moves it

Fixed price: $1,680 USD
Delivery: Awaiting sign-off

Effort: 4 delivery days
Biggest cost driver: Bot API deprecation tracking
Commercial basis: Fixed price agreed before work starts; scope changes quoted separately
Year one, beyond the build: Support retainer, Managed hosting, plus third-party subscriptions, incident work beyond the retainer

As of 2025-10-01, Telegram Bot API 13.4

Why this costs what it costs

The maintenance conversation almost never happens at the right time, which is before launch. A bot goes live, it works, and maintenance looks like paying for nothing. Six months later a refresh token expires over a weekend, the bot stops writing to the calendar, nobody notices until Tuesday, and by then forty bookings have been silently lost. That specific sequence is the single most common way a working Telegram bot dies, and it costs nothing to prevent and a great deal to discover.

The important property of bot failure is that it is binary and quiet. A website that breaks shows an error to the person looking at it, and they tell someone. A bot that breaks stops replying, and the user assumes the business is slow rather than broken. There is no error page, no support ticket, and no traffic graph that obviously dips. Monitoring is not a premium add-on here; it is the only thing that turns a silent failure into a known one.

The second property is that almost all the work originates outside your project. Telegram ships Bot API changes several times a year, some of which deprecate behaviour bots relied on. Google, Microsoft, HubSpot and every payment provider version their APIs and retire old ones on their own schedules with their own notice periods. Your dependencies ship security patches. None of this is triggered by anything you do, which is exactly why it gets forgotten — there is no event in your calendar that corresponds to it.

What a retainer actually buys, then, is somebody whose job it is to read those announcements and act on them before they become your incident, plus the monitoring that catches the thing nobody announced. The small change allowance in the same retainer is genuinely secondary, though it is the part clients value day to day, because a bot that cannot be adjusted stops fitting the business within a year and gets abandoned for reasons that have nothing to do with reliability.

The breakdown

Work	Days	Cost	What it covers
Monitoring setup	1	$420	Uptime checks, webhook delivery failure alerts, error-rate thresholds and a dead-letter queue that pages a human. One-off, at handover, and the single highest-value day in this table.
Runbook and access handover	1	$420	Written procedures for the failures that actually happen — token expiry, webhook deregistration, provider outage — and credentials held somewhere that survives one person leaving.
Bot API deprecation tracking	0.5	$210	Ongoing, per month. Reading Telegram release notes, identifying what touches your bot, and scheduling the change before the deprecation date rather than after it.
Dependency and security updates	0.5	$210	Ongoing, per month. Patch review, upgrade, regression run, deploy. Unglamorous and the reason a two-year-old bot is still safe to run.
Credential rotation	0.5	$210	Ongoing, quarterly in practice. Refresh tokens, service account keys and gateway credentials rotated before they expire, which is the failure this whole page exists to prevent.
Small changes	0.5	$210	Ongoing, per month. Copy edits, a new menu item, an extra field, a changed business rule. Included up to the retainer allowance; larger changes are quoted as projects.
Total	4	$1,680	Fixed on signature. Scope changes are quoted separately, never absorbed silently.

What moves the number

Telegram deletes messages older than 48 hours only within that window; after it, cleanup is impossible.

Bots that post into groups need their retention behaviour correct from the start. Discovering it late is not a fix-it-later item, because the messages that needed removing can no longer be removed.

OAuth refresh tokens for Google and Microsoft can be revoked by an administrator at any time.

A workspace policy change on the client side takes the bot down with no notice and no error visible to users. Retainers exist largely to make that a twenty-minute fix rather than a week of confusion.

A bot taking payment carries an obligation that a bot sending notifications does not.

Payment paths need reconciliation checks and faster incident response, which is why anything handling money should be on a retainer and anything broadcasting reminders can reasonably not be.

Every integrated third party adds its own deprecation schedule.

Maintenance load tracks integration count more closely than user count. A bot with four integrations and a hundred users is more work than a bot with one integration and fifty thousand.

Unmonitored webhook failures are invisible to everyone including the users.

Without delivery alerting there is no moment at which anybody learns the bot has stopped. The cost of monitoring is trivial next to the cost of the silence it prevents.

What year one actually costs

Running cost	Year one	Detail
Support retainer	$3,480/yr	The subject of this page: deprecation tracking, updates, rotation, incident response and a monthly allowance for small changes.
Managed hosting	$660/yr	Operational ownership of the runtime and the data, including backups and tested restores. Frequently bought together with the retainer because the two responsibilities are hard to split cleanly.
Third-party subscriptions	billed by the provider	Unchanged by maintenance and billed directly to you. Worth reviewing annually anyway, since seat counts on a CRM tend to grow and never shrink.
Incident work beyond the retainer	billed by the provider	Charged by the day at the same rate as project work, and rare. Most incidents are a credential or a deprecation and fit inside the allowance.

Buying the same thing elsewhere

Where you buy it	What the money buys	What you carry
No maintenance at all	Zero monthly cost, and for a simple broadcast bot with no integrations this is a defensible choice. Nothing external can expire if the bot depends on nothing external.	The moment an integration exists, this becomes a bet that no credential will expire and no API will change. That bet is lost within about eighteen months in every case we have seen.
The original freelancer, ad hoc	Pay only when something breaks, and the person who wrote the code is the person who fixes it. Cheap while it works.	Availability is unknown at the moment you need it, and there is no monitoring, so the clock starts when a customer complains rather than when the failure happens. Response time is the product here, and ad hoc has none.
Your internal IT team	People already on payroll, in your timezone, with your context. Works well when the bot is simple and somebody genuinely owns it.	Bot API deprecations are not on any general IT team's reading list, and the failure mode is that the bot is nobody's job until it is an emergency. Assigning a named owner matters more than the team's size.

Questions about the money

Is a retainer actually necessary, or is it just recurring revenue?

It is necessary in proportion to how much the bot depends on other people's systems. A bot with no integrations that posts scheduled messages genuinely does not need one. A bot writing to Google Calendar and taking card payments does, and the honest reason is that both of those will break on somebody else's schedule.

What is actually included in the monthly allowance?

Monitoring and alerting, Bot API and dependency tracking, security updates, credential rotation, incident response, and up to four hours of small changes. Anything larger is scoped and quoted as a project rather than quietly consuming the allowance.

Can we start without a retainer and add one later?

Yes, and it is common. Monitoring and the runbook are delivered at handover regardless, so the bot is observable from day one. Adding the retainer later costs the same as starting with it — there is no re-onboarding fee.

What happens if the bot breaks and we are not on a retainer?

It is charged by the day, and it goes into the queue behind committed work rather than in front of it. We will not leave a bot broken, but a same-day response is what the retainer buys and it would be dishonest to imply otherwise.

How often does the Telegram Bot API actually break things?

Genuine breaking changes are rare — Telegram is conservative about backward compatibility. Deprecations with long notice periods are frequent, and the risk is not the change itself but that nobody is reading the announcements. That reading is most of what you are paying for.

Does maintenance include hosting costs?

They are separate lines. Managed hosting is operational responsibility for the runtime and the data; the retainer is responsibility for the code and its dependencies. Most clients take both, but a team that runs its own infrastructure can take the retainer alone.

What if we want to bring maintenance in-house later?

The runbook, the monitoring configuration and the deployment are yours already, so the handover is a walkthrough rather than a migration. We would rather that than have a client paying for a retainer they have outgrown.