One CX Platform or Ten? Why Vendor Count Is the Wrong Metric

Contact centers run an average of 3.9 platforms, and half of European CX professionals say multiple vendors inflate maintenance and support costs — yet two of the most instructive architecture decisions of the past year went in opposite directions. TalkTalk wanted rid of 12 contact center applications. Pets at Home deliberately kept its retail, veterinary, and insurance operations out of one enormous Salesforce instance. Both had defensible reasons, which is exactly the point: vendor count is a poor proxy for whether a CX stack works.

The 3.9-platform figure comes from Puzzel's State of Contact Centres 2026 survey of 1,505 European CX professionals. It is survey data, not audited spending — worth remembering before anyone treats it as a consolidation business case.

Why do fragmented stacks cost more than the subscriptions suggest?

Fragmentation usually starts with reasonable purchases. Marketing buys better feedback tools, the contact center selects a new platform, IT approves another AI application. A year later the bill keeps growing and agents still cannot find basic customer information.

Kevin McNulty, Senior Director of Product and Industry Marketing at Talkdesk, described the mechanism:

"Tech stack complexity is the natural byproduct of years of buying point-solutions through urgency and not strategic thinking."

Subscriptions are only part of the bill. Businesses also pay to maintain APIs, train employees across interfaces, and reconcile conflicting reports, while customers repeat information the business already holds somewhere.

The National Residential Landlords Association shows the upside when consolidation is done with supporting changes. After consolidating its contact channels with Puzzel, average email response times fell from 35 hours to five, and call waits dropped 35%. Those are vendor-reported numbers, and they came alongside routing and workflow changes — so consolidation alone cannot take the credit. The real starting point is identifying which parts of the stack cause the biggest problems.

What are the three architecture options, actually?

• Single platform. CCaaS, CRM, reporting, and workforce tools under shared administration, with fewer contracts. Verify how much is genuinely native: acquired products can keep separate databases, permissions, and consoles. If you still maintain five environments, what have you consolidated?
• Best-of-breed. Specialist applications chosen for their strengths — for example, a WFM tool with forecasting a suite cannot match. You pay to maintain more connections, but replacing valuable functionality with an inferior native module can cost more.
• Composable. APIs, shared services, and orchestration let you replace individual capabilities without tearing everything apart — assuming IT can keep the arrangement working. Best-of-breed concerns which products you use; composable concerns how they connect. They fit together.

Dr. Martens illustrates why hybrid usually wins. Technology chief Graham Calder, in a September 2026 interview, described the prior state bluntly:

"We were as fragmented as you could get."

The company consolidated Salesforce Service Cloud into one global instance but kept two Amazon Connect environments to manage regional voice latency. Identical arrangements make little sense when requirements genuinely differ.

A shared customer-data warehouse can sit under any of the three models, but it does not guarantee comparable reports — check how CCaaS, CRM, and workforce data join up in customer analytics.

Should you consolidate? Six questions first

Counting vendors tells you little. Six applications can work beautifully together; two can make agents want to throw their headsets across the room. Before funding a migration, the people living with the problems should answer:

• Are you paying for tools that do essentially the same job? (Retire duplicates.)
• Do agents re-enter information or switch applications for simple tasks? (Fix workflows.)
• Do systems disagree about customer identities, records, or permissions? (Address governance.)
• Are custom integrations expensive or constantly breaking? (Reconsider architecture.)
• Could the existing suite replace a specialist tool without losing useful features? (Test selective consolidation.)
• Is nobody clearly responsible for cross-system journeys? (Sort ownership first.)

For contact center and CRM consolidation, the awkward questions are about ownership: which system controls customer identities and case histories, how voice and digital channels keep records current, and who steps in when information conflicts. Buying everything from one supplier does not resolve those issues.

When does consolidation make things worse?

Prashanth Krishnaswami, Head of CX Strategy at Zoho, put it directly: "simply moving platforms is often only a half-measure." Carry the same poor processes into another suite and you have paid heavily to preserve them.

Pets at Home is the cautionary case. Simon Ellis, Head of AI Transformation and Enterprise Architecture, explained the decision:

"We've not gone for a big single instance… because the scenarios are very different… insurance is regulated."

Veterinary operates differently from retail, and insurance carries its own restrictions on customer data; identical processes would create needless headaches. Be suspicious of any proposal that sacrifices specialist features or deepens single-vendor dependence without showing what customers and employees gain.

On the other side of the ledger, bundled native tools can be the pragmatic choice. In June 2026, 8x8 reported more than 170% growth in existing customers activating its bundled workforce management tools. That does not prove those tools beat specialist alternatives — but a smaller contact center still scheduling with spreadsheets may find the bundled option entirely adequate.

Test the awkward workflows, check your exit options, and get the real operating costs before signing. Architecture is a capability question, not a procurement saving.