Latest news

Developing a custom bot vs. Using a bot creation platform

No comments

So you’ve decided to build a bot.

There are a lot of decisions to make before building one, but probably the question that will make or break your bot is: Should you develop it from scratch or should you use a bot-creation platform? (Or maybe there’s a third option — we discuss this at the end…)

In my mind there are 3 major factors you should have in mind before starting: bot ability, cost, and time. Here’s a comparison of custom vs. platform for each of these factors.

BOT ABILITY

Or, what do you need your bot to be able to do?

Bots seem simple but they’re actually really complex. On the surface it appears as if all you need to do is build something that detects user input and provide an appropriate response.

But in reality, bots require many of the following things:

Triggers: Keywords, NLP Engine, “contains”, etc.

Features: Chat, Products, Subscriptions, etc.

Payments: PayPal, Stripe, Braintree, etc.

Integrations: E-commerce, Booking, CRM

Platforms: Facebook, Slack, Skype, Telegram, WeChat, etc.

Analytics: Botlytics, Dashbot, etc.

(We didn’t go into the more technical requirements of building a bot but those are obviously significant too)

Custom — The best part about building a custom bot is that there really is no limit to what you can make your bot do. With enough time and money, anything can be done. But this is never actually the case. You’ll always have limits. And then you’ll have to choose what you can keep and what you need to forgo.

One possible benefit of going custom is that the act of narrowing down what you can and can’t do will help you build a better bot that is excellent in one thing rather than mediocre in a bunch.

Platform — The most obvious limitation for bots built on platforms is that they are limited to what the platform can do. Need a shopify integration but the platform doesn’t have one? Unless there’s a way to integrate it yourself, you’re out of luck! You either have to move your bot (but you can’t export) to another platform that has the feature you want (but may not have other things), or settle with not getting the feature you want.

Over time this will improve. At the beginning of website builders, there were things you just couldn’t do. But as wix, squarespace, and others matured, more features became available. Suddenly there was very little reason to build a website from scratch. Bot creation platforms will run the same course.

COST

How much are you willing to spend on your bot?

The Initial cost

Custom — Developers cost a lot. And just like the cost of any software project, the cost to build a custom bot varies wildly. It depends on lots of things like:

  1. Quality of the developers
  2. What you need your bot to do
  3. How many platforms the bot will need to use –
  4. Scale — how many people will your bot need to talk to?
  5. Integrations — does your bot need to integrate with other services like an online store or a CRM? Does it need an API so it can “speak” to a website or 3rd party service?

These are some of the types of things a bot developer will ask you before starting to build.

Smart developers know that software projects always take longer than expected so they will never charge a flat-rate (unless they know they can get away with grossly overcharging you with a flat rate). They will always charge an hourly rate. So the initial cost of building a bot is just a simple equation of developer hourly cost X number of hours.

Platform — By building a bot on a platform, the initial cost of developing a bot will probably be zero or very close to it. A platform will probably charge you for continued use of a bot, so on a monthly, per use, or per transaction basis. You could probably build and test the bot for no cost. This is obviously an extreme advantage in favor of a platform. If you’re not 100% certain that a bot will work for your business, using a platform and not paying anything to build a bot is a wise choice.

Continued development/Maintenance

When pricing projects, people often forget to include or underestimate the changes they will need to make and their cost. Think about how often you update your website. A sentence here or a picture there. What about a new page?

Unless your bot is extremely simple, it will require a lot of changes. At the beginning, it won’t understand a lot of the things people will say to it. Or it might answer people incorrectly. You’ll want to make changes more or less immediately if something goes wrong (luckily, unlike an app, bot changes are published immediately).

Custom — When dealing with a bot that is essentially a lot of code, unless you’re very technical, you won’t be able to make the changes yourself. You may be able to make changes in responses (just changing text), but anything having to do with flow is definitely best left for developers. So there’s that cost too. Keep in mind that developers are in high demand today. If you need a change, there’s a good chance they might not be able to get to it immediately.

Don’t forget you may also have to pay server costs, NLP integration, payments, analytics, etc.

Platform — Making changes on platforms is much simpler than in custom bots. Most platforms keep all the code behind the scenes so you’ll probably be able to make changes on your own, instantly.

Costs for launching the bot and keeping it alive vary between services. Bot platforms range from free for very basic bots to about $1000 a year and every price point in between. You may also have to pay per message/query or per transaction if the bot is geared towards sales.

Another thing to consider with bots is that the work will never end. While websites and apps can be mostly if not completely done, bots will always encounter new scenarios. It could be new words, new languages, unexpected requests…you’ll always be playing catch up to reduce misunderstood inputs.

Marketing Costs

Something that isn’t spoken about so much in the bot community is the difficulty of actually distributing your bot. In the beginning of apps, there were app stores and the goal was to try be listed near the top using app store optimization or get featured.

For now, bot discovery by and large doesn’t exist. There are sites like Product Hunt and Botlist that let you get your bot in front of thousands of people, but other than that, there aren’t many ways of getting your bot out there (it’s also important to ask yourself if this should really be your focus but that’s for another post). Therefore, expect to pay a good chunk of money on marketing.

Every project has a budget and you’ll need to make a tradeoff. If you decide to build a custom bot, you’ll probably spend most of your budget on the development which will leave a minority for marketing/distribution. By building on a platform, you will pay much less to build the bot, leaving you with much more money to distribute the bot.

TIME

How much time until your bot needs to be ready?

Every bot, whether it is built custom or on a platform, will follow more or less the same schedule. First, you’ll need to spec it and map it. Next, you’ll probably want to create a simple bot and deploy it. Then you’ll test it, build more, test it, build more, etc…

Planning

The planning phase of a bot is where you’ll spec it and map it out. Specing stands for specification. If you’ve never written a functional product spec before, I would suggest you read Joel Spolsky’s guide here.

I also find that using flowchart apps are really helpful in figuring out the flow of a bot.

Custom — When planning a custom bot, you’ll need to take into account a lot more things which will take more time. You’ll have to plan the incorporation of

  1. NLP
  2. Analytics
  3. Payments
  4. Subscriptions
  5. Integrations
  6. Platforms
  7. And more!

Each of these parts works together in some form or another and therefore increases the complexity and overall planning time.

Platform — Planning a bot on a platform is easier and therefore quicker than on a custom basis. Some platforms offer a flowchart-like interface that allows you to combine the planning and building steps into one. Most others offer a drag and drop interface that makes building really easy.

Initial Build Time

Custom — Without knowing exactly what you need your bot to do and how many developers you have, it’s impossible to give an estimate of how long building a bot will take. However, expect this to take much longer than on a platform.

Platform — Perhaps the greatest thing about a building a bot on a platform is that getting a first version out there is extremely quick. You can literally get a bot working in a chat app in less than 30 seconds, especially if the platform in question offers templates. You can simply apply the template, insert your own text or products and press launch. At this point it may not be exactly what you want, but getting something online quickly is important because the next phase is the most important.

Testing

Then comes the test phase. This is probably where you’ll spend most of your time. Once you have deployed to a chat app, you’ll need to speak to your bot and make sure it works. This part sounds short but really isn’t. You’ll have to flip back and forth between editing and testing dozens if not hundreds of times.

Custom — Testing a custom bot won’t be fun. You’ll find lots of mistakes, and every change you will want to make will have to go through a developer. They’ll make the change and tell you to test it again. Then you’ll find mistakes…see where this is going? This might be the single best reason NOT to build a custom bot. The back and forth will be a headache.

Platform — Testing a bot on a platform isn’t a walk in the park but it is made a lot easier by a) making changes really easy to make and b) offering a test/development chat within the platform. The nice thing about bots (versus apps) is that changes are applied instantly. As soon as you make a change in the platform, it will be applied to the bot.

At Personify, we’re building a very advanced but simple-to-use bot creation platform. We try to offer a bot that is as close to custom as it gets while keeping the platform clean and easy to use.

We also believe that there’s a middle ground in this custom vs. platform decision. If you want an advanced bot built for you, but don’t want to pay a developer or use a platform yourself, we offer a service where we build it for you. Because we know our platform best, we can offer the very best bots.

WIN A BOT!

We’ll be creating a few custom bots completely free of charge for a few entrants. So just fill out this form and you’ll be eligible to win a bot! If you heart and share this article as well, you’ll improve your chances of winning.

Enjoyed the article? Click the ❤ below to recommend it to other interested readers!

magnoliaDeveloping a custom bot vs. Using a bot creation platform

Related Posts

Leave a Reply

Your email address will not be published. Required fields are marked *