As scary as it may sound at first, migrating from one platform to another doesn’t have to be a daunting experience. The key is to think holistically and take enough time to do it the right way. Before we dive into the do’s and don’ts, we’d like to lay out for you the most common reasons why our clients changed platform provider.
Some of them were open to experimenting and trying new technology solutions, while others really felt their current provider wasn’t the right fit anymore. Here are the five most common reasons why we see our clients migrate from one platform to another.
1. The platform lacks crucial features, or an analytics tool
With the rapid development of conversational AI, it could be that a platform that seemed great and cutting-edge two years ago might seem outdated today. Think of key features like built-in versioning system, precise and quick NLU model training, omni-channel deployment, and seamless 3rd party integrations.
A number of our clients have pointed out that they had problems copying and editing different versions of the chatbot design. As an example, dialog flows or settings sometimes couldn’t be exported and backed-up immediately, which made the maintenance of multiple chatbots quite difficult. On top of that, deployment in more languages often proved challenging as well.
Another key obstacle is having difficulties with training intents. One of our clients needed to make heavy use of terminology to ensure correct intent recognition, however, simultaneously this caused bias and made it more likely to trigger other similar inquiries. Also, having a convenient way of monitoring the performance of the AI Assistant has become a must-have for a lot of clients.
Others found it really important to deploy their AI assistant across specific channels, or integrate it with third-party party systems like Shopify or Zendesk. When evaluating whether it makes sense to develop a tool or a functionality in-house, cost is frequently one of the key factors. There is often a tendency to think that it would be more cost-effective to have the team already working on the platform simply build the capabilities they need on their own. However, an easy third-party integration will not only save resources like time and money but free up time to focus on efficient integration. And some platforms just do that better than others.
2. The platform is not intuitive, or is too complex for creating AI assistants
Clients often complain about the flow logic of some platforms. As an example, in one platform, the client was struggling to chain multiple flows together and make sure that at the end of each flow the correct follow-up action is executed.
Along with the flow logic, clients also struggle with navigating and using a visual interface. When it comes to chatbot platforms, there are typically 3 types of interface for building dialog flows: linear dialog trees, visual flow builders, and code-based interfaces. Most of our clients prefer having a visual flow builder in order to create AI assistants without complex decision trees or coding required. An intuitive drag and drop interface also makes it easy for non-developers to get started.
There is a trade-off, though. More user-friendly, low-code solutions are often less flexible compared to the more developer oriented platforms. So when picking the right platform for you, you need to take into account what it is that you’re building (how complex is it) and who are going to need to build it?
3. The platform does not offer support plans, or the supporting documentation is too poor or even non-existent
It might seem irrelevant at first, or as an unnecessary expense, but having support plans is what really sets great platform providers apart. Whether you’re experiencing technical issues, or there are updated security frameworks or features, a support plan helps you ensure that the platform is kept-up-to-date and running smoothly.
4. The platform was acquired by a different company
Within the last 12 months, there have been many acquisitions within the conversational AI market. Microsoft finalized its acquisition of Nuance Communications, Zoom has acquired Solvvy, Ubisend has been acquired by Soprano Design, Walmart acquired the conversation design tool Botmock…
All of this has caused some clients to feel a bit lost and wanting to look for alternatives. In a quickly changing platform landscape, you want to stay on top of the larger trends and how the landscape is evolving if you want to build a conversational AI operation that is future-proof.
5. The platform has high license and consumption fees
Generally, as the AI Assistant’s functionality grows and and becomes more widely used, the price of the technology also increases. Usually, there’s an annual subscription fee, plus a volume pricing based on the number of messages per month, or a cost per minute for voice interactions. Some providers even charge per functionality added.
That’s why a number of our clients have encountered barriers to scaling. And, yes, lots of platforms have different payment models. But if you want to be prepared, you want to know exactly what scaling your AI Assistant means in terms of costs and benefits.
Steps to a successful platform migration
Migration is a challenging project from the very beginning, as no two platforms are ever the same. At CDI Services, we always guide our clients through a structured process. Before making a decision on migration, we take the time evaluate whether or not the clients has been leveraging the current platform’s full potential.
If after validation, a migration is the best solution, we begin our pre-migration assessment.
To fit the puzzle pieces together, we:
- analyze the current AI assistant (flows, intent structure, entities, languages, integrations, etc.)
- agree on the crucial features and metrics to track the performance
- go through preliminary questions that help us determine a platform we want to migrate to
- compare capabilities between the two platforms
As each migration is unique, we prepare time and cost estimates specific to the project.
After the pre-migration assessment is completed, together with the client, our AI Trainers decide on the changes and necessary improvements to be made during the pre- and post-migration. Following this, our team of AI Trainers and Conversation Designers implement the dialogs, applying the standard procedure of working in batches, and finally testing the dialogs, when implemented.
Recently, we worked with one of the largest European sanitary fittings manufacturers that migrated from e-bot7 to Cognigy. We helped the company with picking the right platform for migration, implementing dialogs, cleaning training data, and optimizing the NLU model. Today, the company has a well-performing chatbot that handles customer service requests and complaints on their website.
If you seem to be no longer happy with your platform provider, or need help with choosing the right one, book a free consultation with our team. We’ll focus on your business goals, the context, and the technology requirements so we can add much value as possible, whilst simultaneously guiding you through a smooth transition from one platform to another.