How much does a chatbot cost?

Vous voulez vous lancer dans un chatbot mais vous souhaitez savoir à quoi vous attendre en termes de tarifs. On a pensé à vous avec l’article le plus complet possible 👌
Twitter
LinkedIn
Email

I’ve been selling chatbots based on our Vizir.co technology for over 4 years now.

I exchanged with many decision-makers and sent many commercial proposals. We have won and we have lost.

Each time the price was not the determining factor.

The question that comes up regularly is rather the time needed to create the chatbot and to manage the chatbot over time.

To calculate your ROI, it is necessary to take into account these two points: the technical cost (in license or purchase) and the human cost of management.

The pricing methods are multiple, and almost infinite. So my goal is above all to give you the main steps to get the best price for your project.

1. Define the use case

As I detail in the Guide to a successful chatbot project , you must first know what the objective of your chatbot is. This is the first step to creating a chatbot.

There are three main types of chatbots.

Business chatbots

These are the chatbots that you will put in place to automate part of your business, whether internal or external.

Internally, the chatbot will be intended for employees. This can range from IT support, to HR process management and the distribution of commercial information.

Externally, the chatbot is intended for customers, users, partners of your company. It will allow for example to simplify the product recommendation, to process after-sales service requests, or to help the onboarding of a user on a new tool.

Examples of business chatbots

Personal Assistants

These are chatbots (or more often voicebots) made available to your customers, or even the general public. They are most often available on Google Assistant or Alexa .

Their job is mainly to understand spoken language and perform simple actions (set a timer, change the music, etc.).

Examples of personal assistants

Media chatbots

These are chatbots whose objective is to distribute content on messaging platforms. They are mainly created by the media themselves, or by certain sports clubs.

Some media create themselves around a chatbot.

Examples of personal assistants

2. Choose the technology to use

These three types of chatbots require different technological bricks.

Business chatbots

Business chatbots should help companies automate recurring tasks: finding information, updating databases, creating tickets, resetting passwords, etc…

They require three advanced technologies.

First a hybrid natural language understanding . The bot must be able to perform actions (ex: “order a new PC”) AND answer a large number of topics in FAQ mode. It therefore requires a hybrid understanding between machine learning and ElasticSearch.

Then they need great flexibility in creating scenarios . Indeed, a business chatbot can be used for thousands of use cases. Response scenarios can often be modeled with FAQ formats, selfcare formats, and action formats. We must also have a mode for creating free scenarios for unique cases.

Finally, and this is almost the most important, business chatbots must be easily interfaced with your business applications. This is what allows your chatbot to exchange data with your apps, and therefore to be smarter and to be able to do more stuff!

Finally, the corollary to these APIs is the ability of the bot to be able to install itself on your own servers. Your CIO will be happy!

Personal Assistants

Personal assistants must understand spoken and written natural language. Their job is to perform actions for you. So understanding is limited to simple but effective machine learning ( intentions and entities ).

Then they too need strong API connections. Indeed, the intelligence of the bot simply consists in transcribing the natural language into computer code and then sending it to an additional system.

Media chatbots

The media type bot will require much simpler technology.

These bots often do not need natural language understanding. They will limit themselves to keyword detection.

On the other hand, you will have to choose a tool that allows you to easily administer the content because it is the heart of the chatbot. It should allow you to add videos, images, etc…

When it comes to personal assistants and media chatbots, the technology used will be free or nearly free.

The intelligence needed is low (simple scripted scenarios, simple API or no API,…). You will have to choose between doing it yourself or using an agency.

As far as business chatbots are concerned, the technology will represent a significant cost because it is more complex and has specific features depending on the use case you will be using. The stakes are also higher, and the ROI is normally higher (automation of human tasks).

3. Choose your mode of operation

A chatbot project is made up of 3 elements:

  1. Who manages the project?
  2. Who configures the chatbot (setup)?
  3. Who manages the chatbot once in production?

For me, the only non-negotiable point is that you have someone internally to manage point 3.

A chatbot only works if you keep it up to date and optimize it regularly. You need to be sure you have the resources to create a chatbot available internally.

If no one in your house knows how to do this, then it’s going to be complicated…

To increase your chances of finding someone in your teams capable of managing the bot, the corollary is that it is better not to have to code to maintain the bot. Your developers are a scarce resource, they have better things to do than spend their days changing content.

So, here is my perfect combo to optimize its mode of operation:

  1. Project manager: someone at home + a service provider => the service will help you avoid the pitfalls of the chatbot and you will be the essential content / business guarantee for the interest of the chatbot
  2. Setup chatbot : you, if you can (skills + time), or the service that will do it quickly and well
  3. Bot manager chatbot : you or someone on your team, without the need for development skills

As you will have understood, in terms of tools, I recommend a SaaS product without code .

4. Choose your service provider

You have all the cards in hand to choose the right service for you.

France Digitale has published a mapping of AI players in France . Unfortunately the AI industry is not easy to map, so this ranking may be open to discussion.

Do you necessarily have to buy chatbot solutions made in France?

Good question. In itself, not necessarily.

The only constraint is a data constraint.

The data collected by the chatbot (there are a lot of them) must be hosted in Europe, with NO possible leak in the USA.

To do this you have two options:

  1. Work with a chatbot publisher whose technology is all hosted in Europe
  2. Work with any chatbot editor, but host its technology on your servers (cloud or on premise).

Installing a chatbot on your servers requires more effort than using a SaaS tool, available after a simple password login.

So in real life it will be easier for you to sell internally a collaboration with a European publisher with data in France, and even easier to work with a French player whose data is in France.

What is the best solution if I have developers available?

If you have developers on hand and want to build your chatbot yourself from A to Z then you will have to turn to natural language understanding solutions.

Indeed, as we have seen above, chatbots are in fact an assembly of natural language understanding and programming (logic, API, etc.).

It will be longer and much less agile, but you can try it!

For the NLP part you can look at the following solutions:

What is the best solution if I don't have developers available?

If you don’t have developers on hand, you should opt for a no-code solution (like Vizir ).

You will often pay a monthly license with optional service for the creation of the chatbot and its improvement over time.

If you go on this option, pay attention to the product. Request a demo. Try manipulating the admin interface.

Your choice should be on the solution that offers the best technologies according to your use case, and that makes them easily accessible in an intuitive way.

Remember: a bot is smart only if it is updated quickly based on collected data.

5. Understand the price

The idea is therefore to break down the benefit and understand its options. I give you three, a bit like the backpacker’s guide.

Tight chatbot budget (< €2,000)

Vous voulez créer un bot mais vous n’avez pas de budget. 

 

La seule excuse valable pour ne pas avoir de budget c’est que vous créez votre entreprise, OK ? 😋

 

Si vous avez du budget, mais que vous hésitez à l’investir dans un chatbot, c’est que votre cas d’usage n’est pas assez bien définit, et autant ne rien faire. Croyez moi !

 

Si vous souhaitez faire un bot média ou un assistant vocal, vous pourrez vous en sortir gratuitement ou quasi gratuitement avec les solutions mentionnés dans le graphique.

 

Si vous souhaitez faire un bot métier, vous devrez forcément passer par la case développeur, et dans ce cas je vous conseille d’utiliser un framework de chatbot (rasa.aidialogflow, etc…).

Average chatbot budget (< €50,000)

You have a small budget to put on the table.

Several reasons for this:

  1. the POC : you want to test the techno and create your first experience on bots. If it works, that the ROI is there, you will be able to convince your colleagues to invest more massively,
  2. the coup de com : you want to release a bot to talk about your brand, publish a press release, a newsletter etc… You are aiming for a reduced scope and you will see later the optimizations to be made

This is a real chatbot project with 4 elements:

  1. The product : the prices will vary between 200€ / month and 2000€ / month depending on your case
  2. The configuration : have your service provider do it. It will go faster and you will limit the risk of error. The price will not be exorbitant: on average 10 days of work (i.e. €6,000 to €600/day, it’s up to you to negotiate)
  3. Specific devs : normally quite weak here, since the use case is a POC or the beginning of a chatbot. I start with 0€.
  4. The run : you will have to train yourself to use the tool. Support will of course be included in the monthly subscription, but it may be worth taking a quick training: let’s start over 3 days at €600 = €1,800

In both cases, I advise you to choose a SaaS tool according to your use case, to take out a paid subscription.

High chatbot budget (< €100,000)

Do you have a tricky topic at hand that a chatbot could solve?

You clearly see the value of a chatbot in this specific case, and you know how to estimate (even with a ladle) that it will be profitable.

And you don’t want to screw up. So you want to take a solid performance, with robust and proven technology.

Here is the price breakdown:

  1. The product : prices will start at €2,000/month and potentially go very high (€10,000 or more) depending on technology, technical infrastructure and conversation volume. Be careful to fully understand the fee schedule. Let’s start with an average of €3,000/month.
  2. The configuration : must be done by the service provider. Your use case is probably too complex to be created by your internal team. The volume of days is, for a classic project of 15 days but it can fly away depending on the use case. Let’s start with the average of 15 days, at €800 / day (service + qualified) or €12,000.
  3. Specific developments : barring a miracle, you won’t be able to cut it. At least to manage API connectors with your internal IS. And in general the service provider is more flexible than your DSI to do this kind of thing. Let’s put 10 days, they will be used. €8,000.
  4. The run : the same as before, you must know how to use the configuration tool. Same format: 3 days at €600 = €1,800

A total of €57,800.

With this kind of budget we can make quite interesting bots.

Space chatbot budget (> €100,000)

You have an idea for a bot, or you have already successfully experimented with the bot on a given use case.

The ROI is there.

But we must go further…

IS integration, increased security, change management, HR impacts, etc…

In this case the overall price can soar quite high.

The product price will be sensitive to variations in volume and technical infrastructure and security.

The configuration part will vary depending on the content to be added. It will be similar to the high budget case.

Specific developments risk salting the note. You will also probably use a service provider other than the chatbot specialist (SSII type Capgemini, Sopra etc…).

Conclusion

Here, I delivered to you with transparency all that I could note after hundreds of proposals sent.

If you are looking to make a chatbot and you think of us to work with you, you will know that we do not take our proposals out of our hat.

They are based on several years of research, tests and pivots following experiences (successful or unsuccessful) with other companies.

You can try to estimate the return on investment of your chatbot project here :

To know all about the  chatbots and AI, subscribe to the Newsletter

🍔 Menu

Need help with your Chatbot project?

News

The essential

Posts not found

OUTIL GRATUIT – Calculateur de retour sur investissement (Excel)

Estimez en toute transparence votre retour sur investissement à partir d’hypothèses simples : le nombre de demandes entrantes, le pourcentage de demandes récurrents, ainsi que le nombre et la rémunération des agents qui gèrent actuellement ces demandes.

Simulateur ROI

You seem to like what you read...

Why not subscribe to our newsletter?

Receive this type of content and many others (tools, news, testimonials, podcasts…) every week directly to the mailbox of your choice. Unsubscribe at any time.