Off-the-Shelf Chatbot Platform or Custom-Built Bot?
An off-the-shelf chatbot platform gives you a fast start for a monthly or per-case price, while a custom-built bot is an investment that gives full control over data, tone, and integrations. Rule of thumb: standard needs and moderate volume favor a platform – high volumes, sensitive data, or deep integrations favor building custom. Always evaluate escalation, language support, and case-system integration.
The customer service lead evaluating an AI chatbot faces two completely different deals: subscribe to an off-the-shelf platform or invest in a custom-built bot. Both can be right – but they differ in economics, control, and what the bot can actually do for your customers. Here’s the comparison to help you choose a track, and the criteria that apply whichever track you take.
Two pricing models, two logics
The platform is priced as an operating cost: typically SEK 5,000–20,000 a month, or with some vendors, per resolved case. The barrier to entry is low, the start is fast – often 2–6 weeks – and you pay in step with usage. The downside is that the cost grows with volume, and you work within the platform’s boundaries: its flows, its range of integrations, its tone.
The custom build is an investment: typically SEK 400,000–1,500,000 plus operations, and 3–6 months to launch. What you’re buying is freedom – the bot can look up the customer’s actual order, follow your business rules, and sound exactly like you. The cost per case also falls as volume grows, since the investment has already been made.
Volume is the simplest compass: at a few hundred cases a month, the platform’s low barrier to entry is hard to beat, while tens of thousands of cases a month make the custom build’s fixed investment cheap per case. Run the numbers on your own figures – cost per case today against expected resolution rate – before gut feeling gets to decide.
The case for building custom: data control and brand voice
Two arguments keep coming up among those who choose to build custom, and both are hard to buy your way into after the fact.
Data control. On a platform, customer conversations flow through the vendor’s infrastructure on the vendor’s terms. For many organizations that’s acceptable – but if the bot handles customer data, orders, or sensitive cases, your compliance function will want to answer where data is processed, how long it’s stored, and who can read it. In a custom build, you decide: model choice, EU processing, retention, logging.
Brand voice. A bot that answers generically erodes a brand that’s otherwise careful with its language. Platforms offer surface-level customization – greetings, colors, some tone – but the bot’s deeper behavior is shared across all customers. A custom build can be tuned until it sounds like your best employee: the same phrasing, the same judgment about when to apologize and when to hold firm.
The criteria that apply either way
Most of what decides whether the bot turns out good is shared between both tracks. Apply the same requirements in both evaluations:
- Escalation. How does the bot hand off to a human – with the full case context, or does the customer have to start over? The share of failed handoffs is the fastest measure of a bad bot.
- Language support. Does the bot handle your primary language with quality – and the other languages your customers actually write in? Test with real customer phrasing, not demo sentences.
- Integration with the case system. Are conversations logged as cases? Can the bot read case history? Without that connection, the bot becomes an island that creates duplicate work.
- Measurability. Resolution rate, customer satisfaction per conversation, escalation rate – can you track these weekly without manual work?
- Content ownership. Can you export the knowledge base and conversation data the day you switch solutions?
| Decision factor | Platform vs. custom build |
|---|---|
| Pricing model | SEK 5,000–20,000/month or per case – vs. SEK 400,000–1,500,000 plus operations |
| Time to launch | 2–6 weeks – vs. 3–6 months |
| Data control | Vendor's terms and infrastructure – vs. your own choices all the way through |
| Tone and behavior | Customization within the platform's limits – vs. full control |
| Integrations | What the platform offers – vs. whatever you need to build |
A scenario: the online retailer that switched tracks in the right order
An online retailer with growing volume starts with a platform bot against its FAQ. Six months in, the data shows: the bot resolves a good share of the general questions, but half of all conversations are about orders and returns – and there the platform can only answer generically, since it can’t reach the order system. With that knowledge, a custom bot gets built with order integration, escalation flows, and its own tone; the platform-era conversation data becomes the requirements spec. Neither investment was wasted – the platform bought insight, the custom build bought capability.
At Weapp we build custom chatbots and AI solutions for organizations that have outgrown the platform stage – and we’ll tell you honestly when a platform is still the right answer. Get in touch with your case volumes and we’ll run the numbers together.
Frequently asked questions
What does an off-the-shelf chatbot platform cost?
Typically SEK 5,000–20,000 a month depending on volume and features, and some platforms price per resolved case instead. Low barrier to entry – but the cost grows with volume, and the content work of keeping the knowledge base current is always on top, regardless of pricing model.
What does a custom-built chatbot cost?
Typically SEK 400,000–1,500,000 to build, plus ongoing operations. What drives the cost is the integrations with your systems, the requirements around permissions and escalation, and the testing. The language model itself is a smaller part of the total – the architecture and the quality work are the big line items.
Can we start with a platform and build custom later?
Yes, that's a common and often wise path: the platform gives you fast insight into which cases customers actually raise and where the bot needs to improve. Just make sure you own your content and can export conversation data – otherwise the custom build starts from zero instead of from what you've learned.
How important is the bot's tone for the brand?
More important than most people budget for. The bot is often the customer's first contact with you – a generic platform tone can feel wrong for a brand that's otherwise careful about its language. Platforms offer some customization; full control over tone, phrasing, and behavior in practice requires building custom.
How quickly can we be up and running?
A platform bot can be live in 2–6 weeks, with the content work as the biggest item. A custom-built bot with integrations usually takes 3–6 months. The time difference is real – but it should be weighed against what the bot actually needs to handle once it's live.