At DTW Ignite 2026 in Copenhagen, Abe Nejad of The Network Media Group hosted Rakuten Symphony SVP Faiq Khan and Telenor SVP, Global Business Security Officer and Head of Network and Cloud Technology Strategy Terje Jensen to discuss KPIs telcos should watch closely in the AI era. This article explores why activity-based AI metrics like token consumption and transaction volume are the wrong measure of success, what enterprises are already demanding as a result and how outcome-based measurement is starting to take shape.
Telecom has simultaneously ushered in and navigated constant evolution in how the world connects and communicates.
Recently, a mature service and saturated user base have resulted in an endless game of telcos trading customers with each other to remain profitable and grow, albeit incrementally.
Along the way, there has been a tendency to continue focusing on metrics that are no longer capable of indicating true success. ARPU and churn are especially prevalent KPIs that have long since provided a helpful business signal.
ARPU tends to smooth over the difference between high-value customers and the general base, ignoring lifetime value. Churn can give the impression everything is fine while masking real business health problems.
The meteoric rise of AI threatens to ensnare telcos in a similar trap. Already, token consumption and AI agent transaction volume are dominating conversations. If both are high? Well, it must be proof of successful adoption.
The other side of the coin is that “tokenmaxxing” and gamified leaderboard strategies within organizations is far from an infallible measure of worker productivity.
Maybe those tokens are being wasted on completely unproductive tasks. Maybe a human could have done a similar task more competently and efficiently. The point is there is so much more to the picture than tokens burned.
Is the usage creating profit? Is it providing a real cost benefit? If operators cannot tell, they must change what they track.
Scrutinizing what gets measured
Telecom has faced similar uncertainty before. The multi-decade transformation from 2G to 3G to 4G was accompanied by a continued inability to point to a clean business case or ROI, even as the industry moved in lockstep to adopt these new mobile network generations.
AI appears to be setting telecom on this trajectory once again. This time, the stakes are even higher given the bets being made across operations, network investment and how people are deployed in a new era of change.
In this moment where there are still more questions than answers about how AI will ultimately impact organizations, it will be valuable to scrutinize metrics like AI-driven revenue growth, cost reduction or network robustness—everything from risk reduction and security improvement to anticipating problems before they impact the bottom line.
Some operators are already thinking along these lines.
Telenor has shifted away from input-driven KPIs like daily AI users or completed trainings toward output and outcome measurement.
An April 2026 STL Partners survey of 138 telecom operators, vendors and analysts pointed to an industry still building the plane even as it sails through the skies. Just 22% reported 10+ percent operational cost savings, which was the strongest-performing domain in the survey. 25-30 percent of respondents said “I don’t know” when asked about AI’s cost impact.
It is clear: investment is running ahead of the industry's ability to measure what it's buying.
Market needs demanding measurement maturity
The challenge presented by the current state of AI investment and measurement is the pressure building outside of telcos’ operations, offices and networks in the form of changing customer requirements.
Historically, enterprise customers may have shopped in the form of GBs and minutes. When there were few levers to pull, imaginative selling was not in high demand.
AI is forcing a shift. Enterprises are increasingly looking for outcomes. Perhaps they want to reduce fleet downtime by 15% or improve fuel efficiency above a certain threshold.
Delivering an outcome-based result means proving it happened. If a tokenmaxxing telco itself can’t accurately attribute its own AI investment to outcomes, how can it do this for a customer?
In Telenor’s experience, enterprises are asking for complete solutions, not isolated small components that someone else puts together. Because connectivity is a business and sometimes mission-critical asset enterprises depend on, meeting new expectations means moving from reactive to proactive. And the move has to be measurable to be trusted as successful.
On this front, telecom is not going it alone. Organizations across industries are facing similar challenges and will ultimately need to learn from one another and adopt best practices. The key is having the willingness to think beyond usual telco metrics.
A framework for moving forward
Best practices on AI outcomes measurement are still being tuned, but leaders are advancing efforts in meaningful ways.
Telenor’s move toward measuring output and outcome is defined by a supporting framework comprising three levels tracked simultaneously:
- Level one: revenue and cost efficiency. This financial indicator is sometimes the slowest to appear, potentially taking months or full quarters after an initiative goes live.
- Level two: customer. Net promoter score and whether the base is growing or shrinking is important here.
- Level three: process and technical. This level considers metrics like mean time to repair, how fast new functionality gets built and deployed, and how well systems recover from an incident. It is the fastest of the three to act on.
No single level tells the whole story. Measuring outcomes rather than activity ultimately means these three levels are tracked together, because a financial number that takes a quarter to materialize can't be the only signal a technical team gets day to day.
Quick wins versus real transformation
At DTW Ignite 2026, there was a visible mix of short-term quick wins and real transformation versus incremental improvement.
The question is whether an organization and its processes can be taken further and made more radical instead of putting patches in strategic places. To this end, quick wins can serve as evidence for building an internal case for the longer-term transformation. In other words, demonstrate value through the smaller cases that ultimately justify the larger structural bet.
External pressures are likely to proliferate along the way, whether regulatory, economics-based or beyond.
But within the next 12-24 months we are likely to see “What is your AI strategy” give way to “How fast are you executing?”
On the road ahead, the differentiator will be speed of execution, not which AI model or technology an operator is standardizing on—so long as an operator can determine whether the execution is producing results. Simply moving fast on the wrong metric won’t put anyone ahead, it will just mean getting to the wrong result faster.
Operators that get both right will be the ones best positioned to compete.
.svg)
