Every reason for using a chatbot is ultimately economic, either in the short term or long term. Even the desire to be perceived as a “cool company” for using advanced technology, for example, is an attempt to improve the brand which should lead to more sales which leads to more revenue. The reason that it is important to be familiar with new technologies is to ensure that you are optimizing your operational costs and customer satisfaction.
Huge ROIs are available to enterprises that avoid the mistakes that lead to a chatbot failure . We are confident that it can be demonstrated in many cases that the ROI for chatbots is extremely high, often over 1000%.
There is a reason why customer support is currently the most popular use case for chatbots. That is because it is relatively easy to quantify the economic gains in this use case and relatively easy to calculate an ROI.
What we are interested in here is not all the related costs associated with the chatbot, but the change in costs due to the introduction of the bot. The same is true with regards to increased income.
Every use case for a bot is different and every company does things in different ways so the examples here will not fit your exact use case. An internal chatbot for employees and an external bot for customers should have very different metrics for cost savings and revenue gains.
Hopefully, this article gives you a rough method and some good ideas as to how to go about calculating an ROI.
Of course for a project to be approved it needs to be approved by the business managers. Business managers need to be pitched on economics, not on qualitative improvements. While they would be interested in improving customer service in general, they want to understand the cost/benefit of implementing the new technology.
It should be noted that the ROI is directly related to the extent to which the bot is leveraged in the organization. The huge advantage of the bot is that it can be scaled for a low cost, so the more justified use of the bot, the better the ROI.
This means that bots that serve less customers or “replace” less agents will normally have a lower ROI.
Estimating chatbot ROI entails making many assumptions. Of course, it is better to err on the side of caution when making assumptions and make them conservative. This will ensure that your business case remains solid.
When considering the cost savings and revenue gains it is important to consider potential overlaps in the measures. These can be mistakes like first assuming that the number of agents will be reduced but then applying the agent turnover savings to the original number of agents. Or it could be more subtle, for example implicitly assuming that revenue gains from referrals are counted in both the “Revenue from better service” category and the “Revenue from referrals category”.
It is worth stating again that you should only be interested in incremental cost savings and revenue gains that come from implementing the chatbot, not gross cost savings and revenues.
For use cases that are more difficult to quantify it may be better to run a POC to test the main assumptions before going ahead with the project. This is what I would expect you to do in almost every case anyway, but at least have a hypothesis before starting.
To help you work out your ROI we have developed the attached spreadsheet. Feel free to modify this to fit your use case. It is invariably the case that the assumptions and calculations will differ from organization to organization and use case to use case.
The list of cost savings and revenue gains below is not exhaustive but it covers many of the key assumptions and should help you develop your thinking on these issues.
There are many ways that the chatbot can save money and improve ROI for the organisation. Some of the main ones are listed below.
The most readily identifiable place for cost saving is labour costs for call center agents. The need for call center agents is reduced if the bot can answer a material percentage of inbound inquiries. For example if the cost per interaction with a human is $5 and the cost per interaction with the bot is $0.50 or less there is a clear saving.
Even with this cost saving there are a few points to bear in mind. When calculating the cost of an agent it is important to calculate the fully loaded cost of the agent. This should include direct salary costs and indirect costs such as incentive pay, benefits, training, recruitment and technology and communication costs. It should also include the allocated cost of the fully loaded management cost.
Working out on average how many client contacts the agent deals with per day will give you the cost per contact.
Normally these costs would be saved not by firing people but through natural attrition i.e. by not hiring new agents as the business scales and by not replacing agents that choose to leave. If the bot can answer the simple repetitive questions, this leaves more time for the agents to give better service to customers who have more complex questions and need more attention, improving overall employee productivity. This is a great opportunity for the company to radically improve ROI and customer service which leads to other income benefits which will be outlined later on.
Working with AI-powered chatbot is good for agent morale. This is because the agents no longer need to deal with the most repetitive inquiries. Instead they can focus on helping people with more complex questions.
The cost of turnover is not just the low morale caused by people leaving, it is also the loss of company knowhow, which is the cost not just of training the new recruits but higher rates of errors and delays as new recruits learn their job.
The current agent turnover rate should be known. The reduction in the turnover rate once the bot is in place can be estimated along with the cost of turnover. This can help you estimate the cost of turnover.
Reducing mistakes saves money and improves ROI. Low morale and employee turnover is one cause of mistakes. Human error and operational errors also cause mistakes. If implemented properly the bot will reduce mistakes.
The cost of mistakes is difficult to estimate but it could be assumed that mistakes lead to loss of customers either through dissatisfaction or being given the wrong information. The revenue per year for these customers that is not lost can either be seen as a cost saving or a revenue gain.
Mistakes also lead to having to provide monetary compensation to unhappy customers and to having to spend extra time on these customers which in both cases is a cost that needs to be estimated.
Mistakes can also lead to less referrals and lower satisfaction rates on online websites and this can lead to hard to measure losses and are a cost saving if avoided.
When the end user of the bot is internal, the time that the bot saves employees can be a material cost saving. If they can get to answers faster, the price per contact goes down significantly, while resolution rates go up. As usual, estimations need to be made about the fully loaded average salary of the employee and the time saving due to the use of the bot, but the calculation is straightforward.
There are many ways that the chatbot can increase the revenue and ROI for the company. Where the bot is an external bot and the end user is a customer, the main assumption is the bot will improve customer service and this will lead to new customers or existing customers buying more.
The improvement in service can be the speed of response but also in removing some of the pain points that make many customer service experiences so bad. Just as a reminder these pain points include waiting in a queue to be connected, having to use an IVR especially when the desired option isn’t covered or the options are ambiguous, being passed to the wrong agent, being transferred between agents which causes delays and having to explain the problem again, having the call dropped and having to phone back and start again, and having to follow up multiple times and having to start again each time.
A chatbot-based approach eliminates many of these issues. Of course there will be times when the customer needs to speak to someone but if the conversation is recorded for both sides on a chat platform and communication that is best suited to asynchronous messaging is done through the chatbot platform, the customer experience can be much improved.
The irony is that not only does the bot increase the speed of response and the need for humans to be involved in many issues, but simply switching to a chat based approach allows the agents to be able to deal with more conversations at the same time versus support delivered by phone.
Some of the main reasons why a chatbot might increase revenue and ROI are listed below.
It’s intuitive that better service can lead to customers staying loyal to the business and also to prospects deciding to buy. Bots can be used in lead generation and sales conversions.
Estimating the impact of better service means estimating the value of a customer per year and then the percentage gain in customers as a result of the bot.
Online ratings influence sales although it can be hard to estimate by how much. If services improve, online ratings will improve especially in terms of recent reviews which is important.
Again it’s easiest to estimate this benefit in terms of the percentage of new customers that the company acquired as a result of the better service.
Like online ratings, referrals are vitally important to businesses and a great customer service experience is likely to get people talking, even more so if it was a great experience with a bot.
This can be estimated in a similar way to favourable online ratings, in terms of incremental gain in customers acquired as a result of the bot.
Customer service agents are often providing information on products before customers buy, and can upsell other relevant products and services. The bot can, however, make connections between the current conversation and relevant products and services in ways that agents can’t. These suggestions can either be given to the customer (if there is a high level of confidence that the suggestion is highly relevant) or can be suggested to the agent on their side to raise with the customer if relevant. This can lead to new customers or an increase in revenue and ROI from existing customers.
Having chatbots available on-demand, 247 is a great service to customers (assuming the bots are implemented effectively). Live chat and bots can help companies both attract and qualify leads. In addition, providing timely and relevant information can help companies convert leads into customers or at the very least book more appointments with customers. The fact that bots can be deployed on so many channels including Facebook Messenger, Web Chat and Wechat means that there are a lot of channels that can be exploited besides the fairly new channels.
Unlike many of the calculations above, the impact of the bot in this domain is highly measurable.
Before the bot is built it is possible to estimate what the impact will be by looking at similar implementations. It is also possible to replicate what the bot will do in some cases with humans to manually test assumptions.
In the case of marketing and lead generation it is also possible to create prototypes on drag and drop platforms that are specifically created for creating marketing bots. In some cases these platforms may be sufficient for your use case, especially if you are a small business. For a larger business there are normally more considerations, including chatbot security and extensibility. A large business would normally want to add increasingly sophisticated functionality to the bot in future and link it to the organizations CMS and skills already developed for other bots.
For an internal bot, besides the time savings for employees associated with receiving faster / better service, there is a strong argument that this can lead to more sales and opportunities. Getting the right information in a timely manner can help salespeople in particular close more sales and make general employees more productive. Again this is fairly hard to estimate but it is important not to neglect this benefit.
One way to estimate this number is to try to estimate how many additional customers will be acquired due to better service to employees.
Of course to calculate the return on investment we need to calculate the investment side of the equation. The two key factors here are the development costs and licensing costs. How to reduce these costs and other considerations and addressed extensively in our “Successful chatbot projects avoid these mistakes” blog so there is no need to address that here. We will only address a few issues related to this calculation.
There are many factors that influence the implementation cost although often many of these factors are not known upfront. Cutting corners by hiring cheap but less professional developers may lead to a false conclusion about the costs of development. In this case the true cost will only be revealed once the damage is done and the cost of addressing the damage is known.
It should also be mentioned that a good bot platform will make the cost of developing bots incrementally cheaper because the later bots can reuse skills and functionality developed for previous bots.
Ironically, if the platform is not open and there is vendor lock-in the cost of developing new costs or extending the bots you have will increase with further development for obvious reasons.
It is, of course, important that the cost of implementation is amortized over the period over which the bot is reasonably going to be used to determine the annual cost. For example, if the implementation cost is $40,000 for example and the bot is anticipated to be used for 4 years, the annual cost of the implementation is $10,000 per year.
Determining the licensing cost is fairly straightforward as it is normally a quoted price. Depending on the situation it is normally possible for enterprises that are looking to enter into a large enterprise license to waive the licensing free for the POC.
Determining the expected ROI is a critical step in any project, not just a software or chatbot project. If it is not possible to expect a high ROI there is usually little justification for going ahead with the project. At the very least the exercise of calculating the ROI can challenge assumptions about the project and lead to a more targeted use case and implementation.
Smart algorithm-powered interfaces are multiplying and becoming more widespread across various sectors such as healthcare. Read our top chatbots for 2021
People have discovered that they can multitask, specifically, they can listen to something...
Building a bot together from scratch and have it integrate most of the basic functionality of Botpress. The series will cover installation, knowledge-based questions and answers, simple conversation flows, custom code, user input validation, and displaying media.