Opening context
Cerence (NASDAQ: CRNC) provides voice and conversational AI technology to the world’s leading automotive manufacturers. With 2,000 employees across North America, EMEA, and APAC — spanning engineering, product, sales, and professional services — and technology embedded in more than 400 million vehicles, Cerence operates as a genuinely distributed global enterprise.
That global workforce was built quickly, and under pressure. When Cerence separated from its parent company, it had to construct its entire IT infrastructure from the ground up — every system, every vendor relationship, every support model — while simultaneously rebranding and launching as a NASDAQ-listed business.
By the time the infrastructure was in place, the question that remained unanswered was the one that mattered most: how any of it was actually landing, across 2,000 people in a dozen countries, doing very different jobs.
The situation
Bridget Collins had built the function. She had the service catalogue, the vendors, the team, and the processes. What she did not have was any reliable, independent account of how the function was performing across the people who depended on it.
The information available to her was anecdotal — sentiments from executive conversations, escalations that surfaced at the help desk, the occasional signal from a specific function or geography. None of it was representative. None of it was comparable across roles or regions. And none of it pointed clearly to where she should direct her attention first.
That gap had a specific consequence. More than half of Cerence’s workforce sat in Research & Development and Professional Services — the two groups most dependent on enterprise IT performing correctly, and the two groups whose relationship with IT would most directly affect the company’s ability to deliver.
Without a clear, independent account of how IT was landing in those functions, and how that account compared across the rest of the organization, there was no independent basis for deciding where to focus.
“I can honestly say that without that data, I may have actually prioritized the wrong thing.”
— Bridget Collins, CIO, Cerence
That risk was not abstract. It was a specific decision not yet made, a service not yet evaluated, a vendor relationship that had not been independently tested.
The absence of the record was not a gap in data — it was a gap in the foundation that consequential decisions would have to rest on.
What the independent record showed
More than half of Cerence’s global workforce participated in the first round. The volume of verbatim input — thousands of comments, organized by service, role, geography, and function — gave Collins and her team a view of the IT landscape that had not previously been available in any form.
Two findings had immediate consequence.
- The functional gap.
IT’s own team scored highest of any group in the company. Finance scored close behind. Research & Development and Professional Services — together, more than half of the Cerence workforce — showed the widest gap from IT’s self-assessment: over 40 percentage points. For a technology company in which the majority of employees sat in those two functions, the record showed clearly that the people most dependent on IT performing well were the ones furthest from satisfied with how it was. The place to focus was not where the loudest signals were coming from. It was where the independent picture showed the largest distance. - The service desk.
The outsourced IT Service Desk function was already on Collins’ radar. The team was already building the business case to bring it in-house. What the independent record did was confirm the scale of the problem — not just its direction. The data named specific failures, organized by service and geography, in enough volume and specificity to carry weight in a business case that would ultimately result in paying early-termination penalties to exit the contract. The record did not create the decision. It made the decision unambiguous.
Regional differences also surfaced clearly — VPN issues appearing across multiple service categories in specific geographies, making visible patterns that anecdotal feedback had not been able to draw at scale.
What changed
The decision to exit the outsourced service desk contract — including paying the cost of early termination — was validated by the independent record. The function was brought in-house. When the second round of measurement was established, IT Support had improved by 67%.
Within six months of the first round, the IT team used the independent record as part of a renegotiation with its core sales automation vendor. User comment data had identified consistent problems with the tool — including specific usability failures and workarounds that users had built around its limitations. That data, cross-referenced with direct input from the sales and commercial operations teams who relied on the tool most, formed part of the evidence base for a complete rebuild of the vendor contract. The contract was restructured, with a 60% reduction in contract value.
Beyond the vendor conversations, the record changed how individual service owners inside IT approached their work. The comments — specific, organized by service, and searchable — gave each service owner something they had not previously had: an independent account of how users were experiencing their area, distinct from their own view of it. Collins observed the effect directly: one service owner began raising user data findings in conversations with business stakeholders, proactively, without being prompted. That the data existed — and that the team was visibly using it — was consequential in itself.
The IT function set an OKR for the following year: improve overall results by 10 points. The second round showed a 20-point improvement. The goal was exceeded by 100%. In 2021, Collins was named BostonCIO of the Year in the Corporate category — selected by her CIO peers.
“I set the team a goal of 10 points, and they doubled it. That’s the kind of result that changes how the whole organization sees IT, and what we are capable of.”
— Bridget Collins, CIO, Cerence
The outcome
The IT environment at Cerence was not broken. It had been built deliberately, under significant constraint, and largely correctly. What had never existed was an independent account of how it was landing — across roles, functions, and geographies — that the decisions which followed could rest on.
Once that record existed, two things became clear: what the team was getting right, and where the attention needed to go next. Both mattered equally. The first gave the team confidence. The second gave it direction that no internal signal had been able to supply.
For a CIO building a function from scratch, the question is not whether to establish the independent record. It is what it costs to make consequential decisions before it exists.
