A Change Scales Only as Far as Its Least-Moved Facet

Most organizations scale a change with the technology at one end of the line from adding tools to reinvention and the metrics, roles and decision rights still at the other. A facet alignment check before each move closes that gap.

The challenge is that people moved and the organization did not

The data shows the gap clearly. BCG’s 2026 survey of close to 12,000 employees, managers and leaders finds that 74% of frontline employees now use AI regularly [2]. Yet only 42% of organizations use it to reshape workflows end to end or invent new business models, and 66% of frontline employees get limited or no guidance on what to do with the time they save [2]. McKinsey puts a number on what that costs. Organizational readiness explains 48% of the difference between leaders who capture value from AI and those who do not; personal readiness explains 25% [1].

Telecom shows the same pattern. EY’s 2026 study of nearly 100 telecom C-suite leaders finds that 97% expect AI to deliver major productivity gains within five years. EY ties those gains to “the organization’s ability to redesign operating models and embed AI into core business processes” [4].

AI is the occasion in these surveys, not the subject. The same gap opens with a new pricing model, a platform or a merger: people pick up the change, and the arrangement around them stays as it was.

The continuum: from adding tools to reinvention

Seen as a continuum, a change passes through four stages (Exhibit 1). Three of them follow McKinsey’s horizons [1]. In the first, adding tools, people get tools that assist with parts of their existing jobs. In the second, automating today’s work, existing workflows run faster at scale. The third, redesigning the work, is the stage I add between McKinsey’s horizons, where the flow, the decisions and the roles are rebuilt around what the change makes possible. In reinvention, roles, workflows and operating models are designed again from scratch [1].

Value follows the line. Of leaders whose organizations are at enablement, 13% report realizing enterprise value from AI; at automation, 24%; at reinvention, 48% [1]. Few organizations are far along. Only 11% of leaders place their organizations at reinvention [1]. These are self-reported survey results, and they show association rather than cause; stronger firms may simply be the ones that redesign. They are still the best measure available of where value sits along the line.

The useful move is to read the continuum facet by facet. An organization does not sit at one stage. Its technology can be at automation while its metrics are still counting usage and its decision rights have not moved at all.

Why the facets drift apart

Technology moves first because it can be bought, deployed and counted. The other facets have to be negotiated, and the evidence shows them lagging.

• Metrics. HCLTech’s 2026 survey of more than 200 senior telecom leaders finds that 60% “still measure AI success through cost efficiency alone” [11]. A cost-per-head metric keeps a change at automation, because it rewards doing today’s work more cheaply.
• Talent. In the BCG survey, 72% say the skills expected of them have shifted, and only 36% feel they have had adequate upskilling [2]. In telecom, 43% of EY’s leaders cite insufficient skills as a barrier [4].
• Culture. Only 28% of BCG’s respondents see “a strong connection between what leaders say and what the organization actually does” [2]. McKinsey finds trust in the organization to be the critical readiness factor at every horizon [1].
• Structure. Structure is moving, but mostly through cuts. Coinbase flattened to “five layers max below CEO/COO” and cut 14% of staff [10]. T-Mobile US cut nearly 4,700 jobs in the first half of 2026 to streamline operations by “centralizing leaders and teams, reducing organizational layers, and eliminating duplicative roles” [9].

Delayering moves one facet. It does not, on its own, redesign the flow of work, the metrics or the roles. When structure moves alone, the people left carry the gap. In Korn Ferry’s 2025 survey, 44% of U.S. employees reported fewer management levels, 72% of U.S. executives said they are “stretched beyond their capabilities,” and 40% of employees said they lack direction at work [7]. Gallup finds that “manager engagement declines with larger spans of control” [8].

The practice: a facet alignment check at every move

The practice is a check held at each gate: between pilot and scale, and before a change moves one stage along the continuum. For each facet the Transformation Leader and the CIO ask one question and look for evidence, not intent (Exhibit 2). The rule that follows is my reading of the evidence above. Scale the change only as far as its least-moved facet, then move that facet first.

The check needs both owners. The CIO controls the facet that runs ahead, the technology, and usually the budget that lets it run. The Transformation Leader holds the facets that lag, the process, the roles and the measures. Run by either one alone, the check tilts toward that owner’s facets.

Where it worked, and where it did not

Industrial (Accenture client). Accenture reports that a global industrial solutions company “redefined how work moves across nine mega-processes, including lead-to-cash,” and reached “~70% touchless processing” with lag days cut by up to three in the areas already in production [3]. Process, structure and metrics moved together. Accenture’s operating-model shifts name the same facets: organize around end-to-end value streams, give every AI agent a named human owner, and measure cost per outcome instead of cost per head [3]. This case is reported by Accenture, the firm that ran the work.

Telecom (Deutsche Telekom, in progress). Deutsche Telekom started by giving employees ChatGPT Enterprise and encouraging experimentation, then began redesigning customer-facing workflows, starting with customer care [5]. Its Chief Product & Digital Officer, Jonathan Abrahamson, frames the aim in these words. “Becoming AI-native is not about adding AI to the way we work today. It is about redesigning the work itself” [5]. The lessons it lists include making “leaders accountable for driving process change, not just tool adoption” [5]. The results reported so far measure adoption (50,000+ monthly users, a 546% rise in AI tool usage since early 2026), not business outcomes [5]. The case is published by OpenAI, Deutsche Telekom’s vendor.

Fintech (Klarna, a reversal). By February 2024 Klarna’s AI assistant handled 75% of customer chats, and the company had cut about 700 employees. In May 2025 it began bringing human agents back, with its CEO stressing the “quality of human support” [6]. Read along the continuum, the technology reached automation while the measure of service quality and the human roles in the flow had not moved with it. That reading is mine; Klarna has not described it that way.

How to adopt it: the Transformation Leader and the CIO

Now (next 90 days)

• Transformation Leader: pick one change that is about to leave pilot. Score each facet on Exhibit 1 with the team that runs it, and name the lowest one.
• CIO: list every automated step or agent in that flow and assign a named human owner to each, as Accenture recommends [3].
• Joint: replace the pilot’s usage metric with one cost-and-quality-per-outcome measure, agreed with finance before the scale decision.
• You can tell it’s working when the scale decision is taken on the facet scores, and a change is held back at least once because one facet lags.

Next 12 months

• Transformation Leader: rewrite the affected roles before the change scales, including what freed time is for. Two-thirds of frontline employees get limited or no such guidance today [2].
• CIO: fund technology against redesigned flows rather than tool licences, so that technology spend follows the process facet instead of leading it.
• Joint: run the check at every gate in the portfolio and publish the facet scores to the executive team.
• You can tell it’s working when outcome measures replace usage counts in portfolio reviews, and redesigned flows show removed steps rather than faster ones.

Longer term

• Joint: move decision rights to the teams that own end-to-end flows, and only after those flows run on outcome metrics. Removing layers before the flow is redesigned leaves the remaining managers carrying the gap [7] [8].

Pitfalls

• Treating delayering as redesign. Structure moves, and process, metrics and roles stay where they were.
• Scoring intent instead of evidence. A roadmap slide is not a moved facet.
• Letting the CIO own the check alone. Most of the facets are outside technology, so the check needs the Transformation Leader as an equal owner.
• Counting adoption as success. Deutsche Telekom’s published results are adoption numbers [5]; outcome evidence comes later and has to be asked for.

How to tell it is working

The facet scores of a change converge before it scales. The pilot’s metric changes from usage to outcome. Roles are rewritten before the rollout, not after the complaints. And value per change rises as more changes reach redesigning the work, which is the pattern McKinsey finds along the horizons [1].

Which facet of your largest change would you score lowest today, and who owns moving it?

Sources

1. McKinsey & Company, “From adoption to impact: Three horizons of AI transformation” (July 2026; global survey of 750 employees and leaders, of whom 608 leaders rated organizational readiness). https://www.mckinsey.com/capabilities/people-and-organization/our-insights/from-adoption-to-impact-three-horizons-of-ai-transformation
2. BCG, “AI at Work: Why Strategy Matters More Than Tools” (3 June 2026; close to 12,000 frontline employees, managers and leaders). https://www.bcg.com/publications/2026/ai-at-work-why-strategy-matters-more-than-tools
3. Accenture, “Operating model in the age of AI: Five shifts” (6 October 2026). https://www.accenture.com/en/insights/consulting/operating-model-age-of-ai
4. EY, press release on the EY-Parthenon Telco of Tomorrow study (16 September 2026). https://www.ey.com/en_gl/newsroom/2026/09/telcos-expect-major-ai-driven-productivity-gains-but-talent-and-operating-model-gaps-threaten-delivery
5. OpenAI, “How Deutsche Telekom is rewiring telecommunications with AI” (customer story, 2026). https://openai.com/index/deutsche-telekom/
6. Nicole Mousicos, Tech.co, “Klarna Reverses AI Customer Service Overhaul” (20 May 2025). https://tech.co/news/klarna-reverses-ai-overhaul
7. Korn Ferry, “Korn Ferry Reveals Workforce 2025 Research” (17 April 2025; about 15,000 workers). https://www.kornferry.com/about-us/press/korn-ferry-reveals-workforce-2025-research
8. Gallup, State of the Global Workplace (2026 report). https://www.gallup.com/workplace/349484/state-of-the-global-workplace.aspx
9. Iain Morris, Light Reading, “T-Mobile US has just cut more than 4,500 jobs in latest shakeup” (10 August 2026). https://lightreading.com/business-transformation/t-mobile-us-has-just-cut-more-than-4-500-jobs-in-latest-shakeup
10. LeadDev, “Coinbase flattens management and trims workforce in AI-driven restructure” (6 May 2026). https://leaddev.com/ai/coinbase-flattens-management-and-trims-workforce-in-ai-driven-restructure
11. HCLTech, Telecom Industry Survey Report 2026 (more than 200 senior telecom leaders). https://www.hcltech.com/telecom-pulse-survey-report