I recently published Forrester Research's Landscape on Contact Center-as-a-Service Systems report, and one of the open questions we asked vendors was what they see as the top challenges for buyers in this market. Having worked for vendors in this space for more than 30 years, I was pretty sure that I knew what their answers would be: Running the contact center on a reduced budget, keeping customers satisfied, adapting to AI and the changes that this will bring to customer service.
I was wrong. The biggest challenge identified by the vendors was integrations. This was an open-ended survey question of 30 vendors, and integrations came up 17 times as one of the biggest challenges. There was a three-way tie for second place with the following:
- Getting value from AI.
- ROI on tech investments in the contact center.
- Change management.
I would not have been surprised to see any of these second-place answers as number one, but all were far behind integrations. I'm as pleased as I was surprised. Integration is critical in the contact center. Without access to back-end systems, contact center reps can't answer questions, create tickets, or perform any number of tasks on behalf of customers. But how could contact center's number one problem be something that was supposed to have been solved in the 1990s with computer telephony integrations (CTI) and screen pops? Surely integrations have only gotten better since then with RESTful APIs, software development kits, improved connectivity, and so many technological advances.
Integrations can mean many things, so luckily this was an open-ended question. But, vendors are loquacious, so there was plenty of color in the survey. Of the 17 mentions of integrations, 11 specifically mentioned integration to on-premises or legacy systems, four mentioned complex environments, and two talked of siloed systems. So yes, this was about getting to back-end systems with a bit of general communications to any sort of system.
How could it be, then, that an important function that was solved 30 years ago is still on so many contact center buyers' minds? The answer is easy: This is hard, and the answers today are not sufficient.
It's a big, hairy problem, with so many systems and so many challenges. Contact centers need to interact with CRM systems, enterprise resource planning systems, financial services solutions, healthcare packages, home-grown order processing systems, and more. A typical contact system relies on at least five systems of record, and that number can explode to 25 or even 50 different back ends. The sheer volume, magnitude, and complexity of the environments is an issue unto itself.
The next issue is customization. Many back-end systems are home-grown, custom platforms, and in many cases, the people who built and designed these systems have moved on from the company, leaving very little documentation behind. Even a common CRM system such as Salesforce or Microsoft Dynamics undergoes massive changes; I've seen systems with 5,000 custom fields.
Vendor reluctance, security, and connectivity issues can get in the way as well. Particularly in the healthcare space, connecting to Epic or Cerner is very different than connecting to Salesforce or a common CRM system.
When I talk to vendors about the challenges of integrations, they consistently offer pre-built integrations as the answer. This can help but does not solve the problem. A pre-built integration simplifies the up-front connection, helps with reporting, and ensures some best practices are followed. If you didn't customize your CRM system, a pre-built configuration is your answer. If you have 5,000 custom fields, integration gets trickier. A pre-built configuration approach can get you a schema that identifies your 5,000 custom fields, but not much else. Once you have your schema the heavy lifting starts to understand which fields you care about and how the workflows you care about are executed. It's a heavy lift in almost every case.
The best you can hope for today is a good set of APIs, tools, and documentation, combined with a solid understanding of how the system is configured and how it works. The integration work is still significant, but good tools and some knowledge will get the job done.
There are reasons that companies identified integrations as their biggest challenge. There are no perfect solutions to these problems, and work must be done to understand the back-end process flows and data components. The only solace I can provide is the potential that AI has for this problem. Generative AI has the potential to automatically define not only the back-end system's data schema, but to understand which fields are actually used and how they are used.
This will likely end up under the banner of agentic AI, where a large language model calls another process to work out with the various back-end systems how each needed process works, and which data is required. It's a big hairy problem, and this is a big hairy solution to set up, but I'm pretty sure we shall see this before too long.
Max Ball is a principal analyst at Forrester Research.