The builder-versus-custom question is almost always framed wrongly, as cheap against expensive. It is really speed now against control later, and the line sits in a specific, nameable place you can identify before any work starts. Below: what no-code builders genuinely handle, where the ceiling arrives, how subscriptions compare with a one-off build over three years, and what to check so you do not end up locked into a platform.

What a builder genuinely covers

Start with the honest part: for most small businesses a builder is enough, and paying for development there is simply spending money. Platforms like SendPulse, Chatfuel and Salebot give you a visual flow editor in which an owner assembles something functional in an evening, without a developer.

The real advantage is not price, it is edit speed. The owner fixes the copy, adds a button and restarts a broadcast without writing a ticket, waiting for a slot and paying for a developer hour. If your script changes weekly, that outweighs any saving — which is why we regularly tell clients to start in a builder rather than commissioning a bot build from scratch.

Where the ceiling starts

Builder limitations rarely announce themselves as a missing feature. They show up as workarounds: one more integration through Make, one more spreadsheet, one more webhook — until a chain of ten services is more fragile than code. These are the points where we consistently see teams move to development.

1. 1You develop your own data model. A contact with tags lives happily in a builder. The moment you need orders with line items, stock and statuses, you start stuffing JSON into a contact text field — and that is the end of it.
2. 2You need payments with receipts. Taking money is often supported; receipts, refunds and idempotency almost never are. We covered why payments in a bot are a state machine, not a button.
3. 3The logic stops being linear. Wholesale versus retail pricing, stock checks against an inventory system, round-robin manager assignment — a dozen branches nobody can modify six months later.
4. 4Your base reaches the thousands. Pricing tracks base size: what cost $25 becomes $150–300 a month at 20,000 contacts, whether or not those people are active.
5. 5You need Bot API features the platform never exposed: inline mode, editing sent messages, Mini Apps, queues and retries on failure.
6. 6Data requirements appear. Customer data sits on the vendor’s servers, and the question of exactly where often has no answer. For healthcare and finance that is a blocker.

// A typical requirement no builder expresses without workarounds:
// price depends on customer type, stock and quantity, and the
// source of truth is your inventory system, not the platform DB.
async function quote(userId: string, sku: string, qty: number) {
  const client = await crm.findClient(userId);
  const stock = await erp.stock(sku);

  if (stock < qty) return { ok: false, reason: 'out_of_stock' };

  const tier = client?.type === 'wholesale' && qty >= 10 ? 'b2b' : 'retail';
  const price = await erp.price(sku, tier);

  return { ok: true, total: price * qty, tier };
}

The arithmetic: subscription versus build

Comparing a one-off build with a monthly fee is meaningless unless you total the cost of ownership over a realistic horizon. Three years is a fair lifespan for a bot — after that either the business or the platform has changed.

Break-even usually lands at 15–20 months for a base of a few thousand contacts. On its own that number decides nothing: if you plan to rework the script three times over the next year, the saving evaporates into rebuild costs. Count the hours you or your manager spend holding a five-service chain together, not only the platform invoice.

Platform lock-in: what to check before you start

Lock-in is not a theoretical risk. Platforms change pricing, drop product lines and occasionally disappear, while your funnel lives inside their visual editor and exports in no usable form. The question is never whether that happens, only what the move will cost when it does.

• Who owns the bot token. It must sit in your own BotFather account, not the agency’s. Basic hygiene, violated more often than you would expect.
• What actually exports. Nearly every platform gives you contacts as CSV. Flows, conversation history and tags — almost none, and those are the asset.
• Where personal data is stored. Server jurisdiction, retention period, deletion on request. Sooner or later a B2B customer asks this in writing.
• How the base is billed. Whether inactive contacts count, whether you can prune them, and what happens at the limit — a block or an automatic upgrade to a pricier tier.
• Whether raw API access exists. The ability to fire an arbitrary Bot API call is the best single indicator that you will not hit the ceiling on your first non-standard requirement.

How to decide in five minutes

The decision reduces to three questions: does money change hands inside the bot, do you own data more complex than a contact, and how often will the script change? From there the table leads almost automatically.

One more piece of sequencing advice: do not start by choosing a tool. Describe the process you are automating first — where it breaks today and how many hours it eats. Our guide to business process automation shows how to cost that out in half a day. Often it turns out no bot is needed at all, just a properly configured field in the CRM.

Frequently asked questions

In short

• A builder covers roughly 80% of small-business scenarios — menus, leads, broadcasts, quizzes. For those jobs, custom development is wasted money.
• The ceiling arrives with your own data model, payments with receipts, non-linear logic, or requirements about where personal data is stored.
• Break-even on a custom build is around 15–20 months, and it moves further out the more often you rewrite the script.
• Before you start, check who owns the bot token, what genuinely exports, and how the contact base is billed. That is the price of your future migration.
• The hybrid — builder for communication, your own service for money and data — beats either extreme in practice.