How Walmart's CDO built credibility and structure in the first 90 days

August 21, 2026 · 5 min · Free, with full transcript

When Walmart reorganized its data function in the early 2020s, the incoming data leadership faced a familiar problem: scattered ownership, competing priorities, and a business that wasn't sure what to expect from a CDO. The choices made in those first three months set the terms for everything that followed.

Listen to the podcast

4 min

Read the full transcript

Welcome to the Leaders Insights podcast.

Today's episode, how Walmart's CDO built credibility and structure in the first 90

days.

A new chief data officer walks into Walmart.

Ninety days later, either they've got a mandate or they've got a nice title and no power.

What decides which?

The first thing they don't do, which is build a giant data platform nobody asked

for.

Nobody expects the new data chief to show up with technology.

The smart ones show up with a list of problems the business already knows it has.

Back up.

What even is a chief data officer?

Because half the execs listening think it's just the IT person with a fancier chair.

Fair, because it often is.

A chief data officer, the person accountable for how the company collects, trusts, and

uses its data, sits in a weird spot.

They don't own the systems like the technology chief does, and they don't own the profit

and loss like the business heads do.

They own something nobody else wants, whether the numbers everyone argues over are actually

right.

So they start with no natural power base?

None.

And at Walmart, in the early 2020s, the problem was scattered ownership.

Merchandising had its own view of sales, supply chain had another, e-commerce had a

third.

Three teams, three truths, all confident, all slightly different.

That's not a data problem.

That's a family dinner.

And you don't fix a family dinner by installing a new dishwasher.

That's the trap.

The incoming leadership resisted the urge to announce a grand architecture.

Instead, first 30 days, pure listening.

Sit with the merchants, the finance people, the store operators.

Ask one question.

Where do you not trust the numbers?

Isn't that just polite stalling?

30 days of coffees while the clock runs?

It feels like stalling until you realize you're building the only currency that matters,

which is credibility.

A data chief who prescribes before they diagnose gets ignored the second the first report is

wrong.

So you map who owns what, where the duplicates are, and crucially, which broken number is

costing someone real money right now?

Give me the concrete version.

What does day 31 look like?

You pick one painful, visible problem, and you fix it fast.

At Walmart's scale, think inventory accuracy, whether the system says a product's on the

shelf when it actually is.

If store associates don't trust the count, they over-order, and you've got warehouses

full of stuff nobody needs.

You don't boil the ocean.

You take one category, clean the data pipeline feeding it, and show the merchants a number

they can bet on.

One category?

That sounds almost embarrassingly small for a company that size.

It's supposed to, small enough to finish in 60 days, visible enough that the person who

runs that category tells their peers.

That's the whole game.

You win one internal customer.

They become your salesperson.

Credibility spreads sideways, executive to executive, faster than any org chart announcement.

Now the structure part, what does structure actually mean here in plain English?

It means deciding, on paper, who is accountable for each important data set.

This is data governance.

The boring but essential rule book for who can define, change, and approve a number.

Before, five people could each redefine active customer.

After, one owner per definition.

Everyone else can argue, but only one signs off.

People hate that.

You're taking away everyone's private version of reality.

They hate it right up until the quarterly review, when two teams show up with different

revenue figures, and one of them gets humiliated in front of the boss.

Then suddenly, everyone wants a single owner.

Governance sells itself the first time someone's caught with the wrong number.

So credibility first, structure second.

Why not the reverse?

Impose the rules, then earn trust.

Because rules imposed by someone nobody respects get quietly ignored.

You can't police a room that doesn't think you belong in it.

Fix the painful thing first.

Earn the right to make rules, then the rules stick.

Order matters.

If an exec listening is 90 days into a data roll right now, what's the one thing they

do tomorrow?

Find the single number your business fights about most, and go make it trustworthy.

Not a platform, not a strategy deck.

One number.

Win that, and you've earned the right to everything else.

That's a wrap.

Fresh CDO briefings drop daily at MBA-training.com.

Read and go deeper