“We're really fast to say 20%, 100% because it's what management wants to hear. But basically, it's just a guess, isn't it?"
A data lead told me a few weeks ago that the value numbers reaching his board are guessed anyway. That’s an honest thing to admit. The number was set months earlier, because someone needed it to get the initiative approved.
Even if you're a few weeks into a new data seat, you probably already know exactly which number I mean. The one someone needed at intake so the initiative could get prioritized, and that nobody has traced back ever since.
At Mindfuel, we hear a version of that sentence often. You can almost say that business cases are inflated in order to get them prioritized. Or, in other words, whoever screams the loudest wins. We hear the same pattern from the people sitting at the table where those decisions are made, no matter the industry or the size of the team or organization.
In this article, I want to talk about how to make your first days count and actually earn credibility to hold up under the hard questions.
Why the first 90 days count for more than they should
In interviews around this year's 20th CDOIQ Symposium, veteran data leader Peter Aiken put the average CDO tenure at about 18 months. If that’s the case, your first 90 days are a sixth of it. On the same podcast, Mark Ramsey went further: half of CDOs never make it past 3 years.
It’s easy to read those words as "the job is brutal" and move on. But the more I listen, the less it sounds like the job simply just being hard, and the more it sounds like something specific.
The role tends to come with a broad mandate. Expectations are high and a little vague at the same time. And when the budget conversation gets tight, the question is always the same one: what did all of this actually return?
“If you are not getting measurable business value from your data and AI investments, or a clear path to it, go back to the office this afternoon and shut them down” - Randy Bean, Founder & CEO of the Data and AI Leadership Exchange
So the real focus needs to go to the business value.
A leader who spends the first 90 days accumulating visible activity like dashboards shipped, meetings held, and pilots launched can still run out the clock with nothing anyone will vouch for when the budget conversation gets tough.
Getting a seat at the table isn't really the problem anymore. The seat is already there. Deloitte's 2025 Chief Data Officer Survey found that 87% of CDOs now report directly into the C-suite. The problem is 54% still feel less influential than their peers.
What's missing is the thing that turns a seat into influence: being able to show, in terms the business already trusts, that the work paid off.
The easy part and the hard part have swapped places
For years, building was the constraint. You needed specialists, infrastructure, budget, and time to ship anything real, and the data team was the bottleneck. That’s no longer true.
Almost any team can now put something live in days: a dashboard, a model, an agent, or a quick pilot. In a Bain & Company survey, 81% of respondents are already implementing AI beyond pilots. New use cases are popping up everywhere in the company.
Getting things built stopped being the bottleneck. Picking the right things, and showing what they were worth, became the new one. In the same survey, 64% named demonstrating clear ROI as a barrier.
You could fairly ask whether any of this holds when everyone can build. If a business unit can spin up its own agent in an afternoon, who needs a central team fussing over use cases?
Easier building actually makes that job bigger. When building was scarce, you could track everything in a spreadsheet. When every team is building its own agents and automations, the organization loses sight of what exists, what it costs, where it overlaps, and whether any of it works. The question shifts from "can we build it" to "should we, does it already exist somewhere else, and is it actually generating value?"
For a new data leader, that's what's waiting on day one: a backlog you didn't choose, full of requests that may already be half-built somewhere else.
The value was never in the dashboard
This is the trap even very advanced teams fall into. Under pressure to show something, they reach for what is visible. A dashboard with strong adoption rates and a pilot that shipped on time.
Our own CEO and Co-Founder Nadiem says he’s seen plenty of dashboards with heavy daily use and zero, sometimes negative, business value sitting behind them. A tool people open every day proves the tool gets used. What it doesn’t prove is if anything got better. Usage and value are two different things, and the gap between them is where credibility can be lost.
Also watch: Don't Panic! It's Just Data Podcast: Why No Use Case Means No Value in Data and AI Investments
The value of a data or AI initiative comes from the decision underneath it: which problem you chose to solve, why that one over the louder one beside it, and what it was meant to be worth.
Gartner's survey of 782 infrastructure and operations leaders found that only 28% of AI use cases fully succeed and meet ROI expectations. Among the leaders whose projects stumbled, 57% said they had expected too much, too fast.
That choice is usually the least rigorous moment in the whole process. The value number gets written down fast, in the language leadership wants to hear, then disappears into a ranking nobody revisits. A perfect example is the quote at the opening of this article.
Without a shared way to define, score, and track a use case, value management turns into anecdote: teams claim impact with no agreement on what was promised, what was delivered, or what can actually be attributed. Prioritization becomes a negotiation, re-run every planning cycle and won by whoever is most senior rather than whoever is right.
What to actually do with your first 90 days
By this point the pattern probably looks pretty familiar. Almost anyone can build something now, requests are coming from everywhere, and a lot of the numbers attached to them were written to get something approved. In your first quarter, the job is to become the person who can tell which of those numbers hold up.
Start by agreeing on what success means before anyone starts judging results. CDAO Wendy Batchelder puts it well:
“Without that alignment, even strong execution can appear unsuccessful because different stakeholders may be evaluating outcomes through different lenses."
That applies to every initiative you inherit, too.
Then take a proper look at the backlog you walked into. You didn't choose most of it. Some of it is probably already half-built in another team, because AI made that easy. Before you add anything new, run what's already there through the same questions you'd ask about a fresh request.
For every item, new or inherited, answer a few honest questions:
- Was a business problem ever written down behind it?
- Does it name a KPI, a baseline, and a target?
- Is the value estimate built on anything? How confident are we in the number, said plainly, high, medium or low?
- Do you know what it costs, and what could stop people from using it?
- Does something like it already exist somewhere else?
- Who owns it on the business side, and who checks the value by when?
Don't expect clean answers. On a lot of inherited items, the answer to the confidence question will be "low" or "nobody knows," and that's useful in itself. It tells you which numbers you can stand behind and which ones you should stop repeating.
None of this means skipping the quick win. Most first-90-days advice still recommends one, and a new leader does need something to point to. A quick win only counts if the people judging it agreed beforehand on what it was supposed to prove. So pick one that holds up against the questions above, rather than the one that's easiest to demo.
The harder part is saying no. You have capacity for a fraction of what's asked, and what gets questioned eventually is the call: why this, why now, was it worth it?
The people whose ask didn't make it will want to know why theirs didn't. If you can walk them through your answers, they may still disagree, but they can see how you got there. That's what executive trust comes down to: people can follow your reasoning without re-checking it themselves.
A good test by day 90: if someone asked you today, could you defend every item in your current portfolio? If not, you know where the next quarter starts.
Where this leaves you
If you’re early in a data or AI leadership seat right now, looking at a backlog you did not fully choose, this is the advantage nobody hands you on day one. You don’t earn trust by shipping the most or staying busy.
You earn it by being the person who can explain why this, why not the other one, why now, and what it will be worth. That holds up long after the 5x prioritization meetings are done.
We've spent the last three years inside more than 50+ data organizations watching this play out, which is the only reason I can say the pattern is this consistent. The teams that last are the ones who treat the decision, not the delivery, as the thing worth getting right.
If you are in the middle of that, I would like to hear how you are approaching it. Book a meeting with me.
