Most of us push for a data- and evidence-informed culture for good reasons. We want to know we’re heading in the right direction, to have chances to adjust course and to celebrate the wins. But in doing so, we often ask people to use a new dashboard, collect more data, go to training, cover another item in team meetings, record their decisions in more detail and make time to reflect on what the evidence is saying. All of that quietly assumes the rest of their workload stays exactly as it was.
In reality, the old reports are still due, the regular meetings still go ahead, the legacy spreadsheets keep getting passed around, and the usual processes carry on, simply because no one has said they can end. We expect people to work differently without reshaping their work to make that possible. And then we’re puzzled when data feels like an add-on or when it doesn’t get done.
This is a familiar reason change efforts fail. In How Leaders Get in the Way of Organizational Change, Harvard Business Review notes that leaders tend to underestimate how much work is involved, overestimate how much capacity the organisation has, and overlook that the people delivering the change still have their day jobs to do. Real progress only came when leaders put some initiatives on hold and identified parts of people’s existing work that could be dropped to free up space for the change.
I think this matters even more when it comes to data. Working in a genuinely data-informed way asks for something extra: time to look at the evidence, think critically, challenge assumptions, weigh different interpretations and decide what happens next. You can’t squeeze that into the last five minutes of a meeting. It takes real attention, and attention is a limited resource in any organisation. If we want people to use data well, something has to give.
Why deimplementation deserves more attention
Deimplementation is a clumsy word for a very practical idea… It means deliberately cutting back, replacing or ending the things that no longer give back enough value for the effort they take. Most of us are far better at implementation. We enjoy launching initiatives, adding measures, bringing in new systems, setting up working groups, scheduling meetings and building processes. Each one usually made sense at the time, but the work keeps piling up, and deciding to stop, simplify or remove something is much harder.
Forbes contributor Cheryl Robinson makes a related point in High-Impact Leadership Starts With a Stop-Doing List, Not a To-Do List. Clearing away low-value work frees the energy and focus needed for what matters more. I’d go further when it comes to building a data- and evidence-informed culture, because this can’t just be about personal productivity. Telling people who are already stretched to manage their time better isn’t the answer… We need to make deliberate choices about which work and tasks we no longer need.
In practice, that could be retiring a monthly report that rarely shapes a decision, getting rid of duplicate data entry, cutting the number of measures teams have to track, shortening a regular meeting, or asking whether information really needs to be collected as often and in as much detail as it is now. Ideally, it means actually ending something… even if a few loyal fans of the old way object!
Let go of what no longer earns its place
I’m a big advocate of rethinking how we work, andI believe every new task involving data should come with a conversation about what can be removed, lightened, automated or built into existing work.
It isn’t only about efficiency or saving time, though those are welcome benefits. For me, it’s about protecting the quality of people’s thinking. When people are surrounded by noise and competing demands, they don’t become more data-informed; they just get more selective about what they pay attention to. Then the work turns into box-ticking rather than curiosity, and there’s no headspace left for the deep thinking that good data-informed practice needs.
A handful of meaningful measures, tied to real decisions and everyday workflow, will usually give you far more than a mountain of activity that no one has time to make sense of.
Lighten the work that still matters
Of course, this isn’t about removing everything. It’s an opportunity to ask what could happen less often, with less detail, or in less time. Could a weekly report go monthly? Could a 40-field form shrink to 10 fields? Could teams look into exceptions instead of manually checking every result? Could one piece of evidence cover several reporting requirements, rather than being repackaged three different ways?
These questions are worth asking because we tend to treat the way a process currently works as if it were the same thing as its purpose… and it isn’t. If the purpose still matters, we can protect it while cutting the effort it takes to get there.
Automate, but use judgement
There’s plenty of repetitive data work that can sensibly be automated: moving information between systems, refreshing reports, cleaning standard fields, running routine calculations, and flagging when a result passes an agreed threshold.
But automation should come after process thinking, not instead of it. In How One Health System Got Rid of Bureaucratic Busywork, Gary Hamel and Michele Zanini point out that digitising a pointless or badly designed process can make work even harder. More recently, HBR’s Stop Automating Old Processes. Design New Ones Instead drew the same line: automating separate tasks isn’t the same as redesigning the workflow that creates the value.
The real question isn’t whether technology can do the job (these days, with generative AI, it often can…). It’s whether the work still needs doing at all, and if it does, where we still need people involved. Done well, automation takes away repetitive effort so people have more time for interpretation, conversation and judgement, which are exactly the parts of data-informed work we should be protecting time for.
Build data into the work itself
Finally, using data shouldn’t need its own meeting, template or initiative. The more we weave evidence into the structures we already have, the less likely it is to become another layer of activity. A project review can include what has changed and what the evidence suggests. A planning conversation can bring assumptions into the open. A team meeting can use a few well-chosen measures to decide where attention is needed.
When data is built in like this, it stops being something people do on top of their work and becomes part of how the work gets done. To me, this is one of the clearest tests of whether data is truly a priority. Leaders often show priorities by talking about them, funding them or writing them into strategic plans. But employees experience priorities in their day-to-day life: what gets protected, what gets resourced and, just as importantly, what they’re given permission to stop doing.
Where to start
If we ask people to work in a more data-informed way but keep every existing expectation in place, we haven’t really made data a priority; we’ve just added it to the pile. So my challenge to you is this: the next time you introduce a new data request, think about what you can remove or redesign at the same time so the work becomes more efficient and more meaningful, rather than simply bigger.
FAQ
1. What is a data-informed culture?
A data-informed culture is one where people use evidence to guide decisions, question assumptions and adjust course, while still applying professional judgement. Data supports the thinking. It doesn’t replace it.
2. How do you build a data-informed culture?
Give people protected time to examine evidence, challenge assumptions and decide what happens next. For every new data request, also decide what can be removed, lightened, automated or built into existing work. Without that trade-off, data becomes an add-on.
3. Why don’t people use data at work?
Data is often added to a workload that otherwise stays the same. Old reports, meetings and processes continue, so data feels like an extra and there is no headspace left to think about what it shows.
4. What is deimplementation?
Deimplementation means deliberately cutting back, replacing or ending work that no longer gives enough value for the effort it takes. In a data context, that might mean retiring a report that rarely informs a decision or removing duplicate data entry.
5. How do you reduce workload when introducing new data processes?
Retire reports that rarely inform decisions, remove duplicate data entry, track fewer measures, shorten meetings, or collect information less often and in less detail. Make these choices when the new process is introduced, not afterwards.