Insights

Productivity won't earn you trust

10 min

re.set - Training

AI is making every organization faster and cheaper. The ones that turn that into real value for their clients will be the ones that stay ahead.

A few years ago I watched the agency I was working for lose one of its best accounts in a single week.

The client was an investing brand we had worked with for two years. The work was good, every interaction was positive, and the last quarterly client survey gave the agency a Net Promoter Score (NPS) of 72. The team was looking forward to the next brief.

It never came. Within a week, the next campaign went to another agency. The reason was almost embarrassing in its simplicity: the other agency had shared an idea nobody had asked for, the client liked it, and the work was theirs. No pitch.

That week changed how I think about trust. The NPS had measured how well the agency delivered. Nothing measured whether the client saw it as ahead of them.

We all say we want to be a “trusted partner.” I’ve said it more times than I can count, and the word has been used so much that it barely means anything anymore.

Trust rests on two things. The first is excellent execution: great work, on time, the way you promised. The second is being ahead: seeing what’s coming and bringing it to the client before they ask. That’s the part the agency was missing.

This is where AI changes things. Properly set up, with the right context, instructions and checks, AI is very consistent at delivery, so execution gets better and cheaper for everyone. What AI should allow organizations to do is be ahead. I believe that is what clients and consumers will value most, in B2C and B2B alike: someone who brings them valuable things they hadn’t thought about.

So the question for any organization becomes what to do with the efficiency AI gives it, and how to turn that efficiency into opportunity. This article is my answer so far, and a reflection on how our own way of working at re.set is evolving.

Why AI makes being ahead harder


In my conversations with CMOs over the past year, the complaint I hear most is that they feel well supported day to day and alone when things get uncertain.

They want someone who sees what’s coming before they do. Your clients now have the same tools you do. A CMO can get a category analysis before breakfast, so any insight that comes from asking a model the right question has a short life.

Being ahead used to last years, thanks to a proprietary method, a research panel or specialists nobody else had. Today it might last a quarter. And brilliant people still fall behind if the organization around them needs a year to change how the work gets done.

Staying ahead is an organizational problem more than a talent one.

Agility helped with part of this, because agile teams react quickly to what is happening. But many foundations stayed rigid: budgets, roles, who does which work, what stays in-house and what goes to partners. The organizations that stay ahead will be designed for elasticity, able to move work between people, AI and partners and turn an efficiency gain into an opportunity almost immediately. That is what wins the next project without a pitch.

Two kinds of AI, two kinds of OKRs

In our workshops, this is the distinction that changes the conversation fastest. Most organizations use AI for one purpose and call it a strategy.

We work with OKRs, Objectives and Key Results: a few goals per quarter, each with two or three measurable results. When AI comes in, we ask teams to separate two kinds.

Efficiency AI does what you already do with less effort. We measure it in margin: lower cost to serve, faster turnaround, fewer errors. For our clients, that means agents across the workflow, from research and synthetic audiences to project management, plus new AI design tools so production stops being the bottleneck.

Opportunity AI does things that weren’t possible, or affordable, before. We measure it in growth. Among our clients, that includes GEO audits that show a brand how AI assistants describe and recommend it; brand narratives combined with real-time social listening, so a client can react to a trend before its competitors; and business cases built in days, so a new idea is stress-tested before anyone commits budget.


Each kind gets its own OKRs and its own key results:


One rule makes it work: the capacity your Efficiency OKRs free up goes to a specific person, working on a specific Opportunity OKR. Skip that step and the organization quietly absorbs the time, and a year later you have a cheaper version of the same company.

The one rule: the time efficiency frees up goes to a specific person and a specific opportunity. Otherwise the organization absorbs it.

Your competitors are buying the same efficiency from the same vendors. Efficiency on its own won’t make your clients trust you more.

One workflow, both kinds of value

The example I use most in workshops is small on purpose.

For years I wanted a weekly trends report: what moved in my clients’ categories, ready to read on Monday morning. Sometimes an intern built one, and when the intern left, it stopped. At my current client, an intern started doing it again, spending most of her Friday on it. That is when I proposed automating it.

Today Claude Sonnet 5.5 handles the research, curation, design and distribution to the business leads, every single week.

That’s the Efficiency OKR: a full working day back, and a report that no longer depends on someone having time.

So what does she do on Fridays now?

While AI builds the report, she analyzes social listening for the same clients, finds the conversations that are gaining momentum, and turns them into a short brief for the creative team. The day we got back went to one person and one piece of work that gives clients something new.

That is the Opportunity OKR, and it matters more. On Monday, value arrives on two tracks. The business leads get the trends report and a reason to call a client. The creative team gets a brief and can take ideas to clients before anyone asks. In the first two months, three of those Monday calls became briefs.

An idea nobody asked for can just as easily annoy a client. These work because they start from what is moving in the client’s own category.

One workflow, redesigned once, now produces both kinds of value. We look for that pattern in every workflow we touch.

Running both kinds of OKRs takes people with new combinations of skills. We see three roles emerging, Growth Architects, Innovation Architects and AI Systems Architects, and they are the subject of the next article.

From agility to elasticity


Turning efficiency into opportunity, week after week, depends on how the organization is built.

Agility gave teams the rhythm. Among the roughly 200 squads in ten countries we have worked with, those that kept the method for six quarterly cycles or more improved their productivity against their own OKRs by 22% on average, and their satisfaction by 25%. But agility didn’t change the foundations, and that is where efficiency gains get stuck.

Elasticity is the next step, and the Monday report shows it at a small scale. Work moved from a person to AI in a few weeks, and she moved to work that gives clients something new. Nobody had to reorganize anything. That speed, from an efficiency gain to a new opportunity, is what we now design organizations for.

In a rigid organization, the same workflow ends differently. The report gets automated, and within a month her Friday fills up with other tasks, because her hours were budgeted for production. Nobody has the authority to move her to social listening, and the creative team has no slot for a brief nobody ordered.

The efficiency is real, and the client never sees it.


We don’t have data on elasticity yet. It’s a model built on what agility taught us and on the first workflows we’ve redesigned. I’m fairly sure of the direction; the numbers will take a few more quarters.

Five questions to ask about your AI transformation

Lower costs are easy to see. Whether AI is also earning you more of your clients’trust is harder to see, and a satisfaction survey won’t tell you; that NPS of 72 measured delivery. These are the five questions I would ask.

The agency that lost the investing brand would have scored well on everything it measured. It had never asked itself any of these questions.

Subscribe to our Newsletter

Subscribe to our Newsletter

Subscribe to receive the latest news from our blog

Start working better