Cost

What a payment bot costs

Adding payment to a bot is four to eight delivery days beyond the same bot without it. Checkout is the small part. The cost is reconciliation, refunds, idempotency when a webhook arrives twice, and the audit trail your accountant needs — plus whichever provider will actually underwrite your category.

What a payment bot costs: effort, cost and what moves it

Fixed price: $4,200 USD
Delivery: Awaiting sign-off

Effort: 10 delivery days
Biggest cost driver: Provider eligibility review — settled before any code
Commercial basis: Fixed price agreed before work starts; scope changes quoted separately
Year one, beyond the build: Support retainer, Managed hosting, plus gateway transaction fees, telegram stars commission, chargeback and dispute handling

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

Why this costs what it costs

Payment is the point at which a bot stops being a convenience and starts being a system of record, and the price reflects that shift rather than the number of screens involved. A checkout button is an afternoon. What follows the button is the work: knowing with certainty whether a given customer paid, being able to prove it three months later, refunding without creating a second entry, and behaving correctly when the same confirmation arrives twice because your first acknowledgement timed out.

That last case is not hypothetical. Telegram retries webhooks it believes were not acknowledged, and payment providers do the same, so any handler that is not idempotent will eventually process one payment as two. The customer sees a double charge, your reconciliation shows a discrepancy, and the fix after the fact costs more than the idempotency keys would have. Every payment build here includes them, and it is not a line a client can remove.

Before any of this, though, comes a question that is commercial rather than technical: who will accept your business. In the UAE this decides more payment projects than architecture does. Card acquirers will not underwrite virtual asset businesses, and that is a category refusal rather than a negotiation. Telegram Stars may only be used for digital goods and services, which excludes anyone shipping something physical no matter how the listing is worded. Several global providers support the UAE while prohibiting categories that are entirely ordinary businesses here.

So the sequencing that saves money is: settle eligibility, then settle settlement currency and payout timing, then build. Reversing that order — building the integration and then discovering the merchant account will not be approved — is the single most expensive mistake available in this category, and it is common enough that we open every payment conversation with it.

The breakdown

Work	Days	Cost	What it covers
Provider eligibility review	1	$420	Which providers will underwrite your entity and category, what they settle in, and how long onboarding takes. Done before any code, because it can change the whole plan.
Checkout flow	2	$840	Product or amount selection, invoice construction, the payment surface itself, and confirmation copy that is accurate about what has and has not happened yet.
Provider integration	2	$840	One gateway or Stars: credentials, invoice creation, webhook or callback handling, and the pre-checkout decision inside the window the provider allows.
Idempotency and retry safety	1	$420	Keys on every write, duplicate webhook tolerance, and a state machine where processing the same confirmation twice is a no-op rather than a second charge.
Reconciliation and ledger	2	$840	A durable record of every intent, success and failure that can be matched against the provider's own statement. This is what turns "the bot said it worked" into evidence.
Refunds and disputes	1	$420	Refund initiation from the admin surface, partial refunds where the provider supports them, and the customer-facing messages for each outcome.
Admin and audit surface	1	$420	What your finance person opens: today's takings, failed payments with reasons, a searchable history, and export in a shape their accounting software accepts.
Total	10	$4,200	Fixed on signature. Scope changes are quoted separately, never absorbed silently.

What moves the number

Telegram Stars may be used only for digital goods and services.

Anything physical is outside the policy regardless of how the item is described, and enforcement lands on your bot rather than on your listing copy. This single rule removes Stars from the shortlist for most retail businesses.

A pre-checkout query must be answered within roughly ten seconds.

Every stock, eligibility or fraud check in that path has a hard latency budget. A synchronous call to a slow inventory system inside that window fails checkouts, so the design has to precompute or degrade deliberately.

Card acquirers in the UAE do not underwrite virtual asset businesses.

A crypto-adjacent product needs a chain rail or Stars, not a better pitch to an acquirer. Discovering this after integration is the most expensive ordering mistake in this category.

Payment webhooks are retried, and retries can arrive out of order.

Idempotency keys and an explicit state machine are mandatory, not defensive engineering. Without them a retry becomes a duplicate charge and a manual reconciliation exercise every month.

Multi-currency settlement multiplies reconciliation work rather than adding to it.

Selling in AED and USD with settlement in one of them means every record carries a rate and a timestamp, and the accounting export has to explain the difference. Decide the settlement currency before building.

Subscriptions are a different product from one-off payments.

Recurring billing brings renewal, dunning, failed-payment retries, upgrade proration and cancellation states. It roughly doubles this table and should be scoped as its own project.

What year one actually costs

Running cost	Year one	Detail
Support retainer	$3,480/yr	Strongly recommended on anything taking money. Payment providers change APIs and requirements on their own schedule, and a payment path that fails silently is a revenue incident rather than an inconvenience.
Managed hosting	$660/yr	Includes the ledger and its backups, with restores tested rather than assumed. The record of who paid is the single most valuable data a payment bot holds.
Gateway transaction fees	billed by the provider	Charged per transaction by your provider under your own negotiated contract. Local acquirers and global platforms differ materially, and volume changes the answer.
Telegram Stars commission	billed by the provider	Deducted by Telegram at the platform's published rate, with its own withdrawal mechanics and timing. Model it against your price points rather than against a headline percentage.
Chargeback and dispute handling	billed by the provider	Priced by the provider per case and driven by your category. Chain settlement avoids this entirely, which is one of the few genuine arguments for it.

Buying the same thing elsewhere

Where you buy it	What the money buys	What you carry
A payment link in the chat	Effectively free. Paste a hosted checkout link and let the provider handle everything. For low volumes with a human checking the dashboard, this genuinely works.	Nothing connects the payment back to the conversation, so fulfilment is manual and reconciliation is a person comparing two screens. It stops scaling at the point that person is busy.
A plugin on an existing store	Your storefront keeps the catalogue, stock and payment logic, and the bot becomes a notification and support surface on top. Considerably cheaper than rebuilding checkout.	The customer still leaves Telegram to pay, which is the transition that costs completed orders. You are trading conversion for cost, and it is a reasonable trade at low margins.
A no-code bot builder with payments	A monthly subscription and a working checkout in a day. Fine for a fixed catalogue with simple pricing.	The ledger is theirs, the refund tooling is whatever they built, and reconciliation exports are whatever the dashboard offers. Adequate until an accountant asks a question the dashboard cannot answer.

Questions about the money

Stars or a card gateway?

Stars if you sell digital goods at low price points and want one-tap checkout with no merchant account. A card gateway if you sell anything physical, need settlement into your own bank, or your amounts are large enough that the platform commission outweighs the conversion gain. The dedicated comparison works through the crossover.

Can we take crypto instead?

Yes, and for a crypto-native audience it is often the only rail available, since card acquirers here will not underwrite virtual asset businesses. It settles irreversibly with no chargebacks, and it excludes everyone who does not already hold a wallet.

What if a payment succeeds but the bot fails afterwards?

That is precisely what the ledger and idempotency work exist for. The payment is recorded before any fulfilment is attempted, fulfilment retries independently, and an unfulfilled paid order raises an alert rather than being lost. Money is never the thing that gets dropped.

Do we need a UAE merchant account?

For card payments settling in AED, yes, and onboarding time is usually the longest item in the whole project plan. Stars needs none. That difference in lead time is often what decides the launch date rather than the engineering.

How does a refund work once the bot has confirmed an order?

From the admin surface, through the provider's refund API, recorded against the original transaction so the ledger stays reconcilable. Partial refunds where the provider supports them. Chain payments are the exception — those are irreversible, and the flow makes that explicit before the customer pays.

Is our bot handling card numbers?

No, and that is deliberate. The card details are entered on the provider's surface and your infrastructure sees a token and a result. Keeping card data out of your systems removes an entire compliance burden that is not worth carrying.

Can we run Stars and cards side by side?

Yes, and for mixed catalogues it is often right — Stars for the digital items, a gateway for everything else. It adds roughly two days for the second provider and a shared ledger that treats both as one history rather than two.