Skip to content
Our world is data

Every company already has the data. Few use it to decide.

Seven readings on the distance between the data an organisation generates and the decision it makes on Monday morning. Every figure comes from a public study; the source sits with each chart.

FormatVisual essay · 3 acts
Length7 readings · 12 min
ClosingFive-question self-assessment
I
Act oneThe diagnosis

How far apart are what an organisation says about its data and what it actually does with it.

01The gap

Investment is not the problem.

Nine out of ten boards place data and analytics among their investment priorities. Asked whether their organisation decides differently as a result, the share drops to little more than a third. The distance between those two answers is what this page is about.

The same survey identifies where the obstacle sits, and it is not where it is usually looked for.

Reports data and analytics as a top investment priority90.5%
Recognises itself as an organisation that decides with data37.3%
Where the barrier sits
92% people and organisation8% technology

Data & AI Leadership Exchange (Randy Bean, with DataIQ), 2025 AI & Data Leadership Executive Benchmark Survey. Data and AI executives at large corporations.

02The scale

Two out of every three data points never reach a decision.

The data available inside an average company far exceeds what makes it into a report. Not for lack of volume: for lack of a path between the system that produces it and the table where the decision is made.

The five stated barriers
01Making collected data usable
02Managing its storage
03Ensuring the needed data is collected
04Securing it
05Bringing existing silos together
32%is put to work
68%goes unleveraged
Annual growth of enterprise data (2020-2022 projection)+42.2%

Seagate and IDC, “Rethink Data: Put More of Your Business Data to Work—From Edge to Cloud” (2020). Survey of 1,500 enterprise leaders.

03The clock

Almost half the day goes before the analysis begins.

Whoever prepares the numbers spends most of their time getting them, reconciling them and cleaning them. Analysis —the part that produces the decision— gets what is left. Two years on the reported share is lower, though the categories are not strictly comparable across editions; either way it still caps what a team can answer in a week.

45% preparing, loading and cleaning data
21% visualising and presenting
34% selecting, training and deploying models
TrendData preparation: 45% in 2020 → 38% in 2022, with categories that change across editions

Anaconda, “State of Data Science”, 2020 and 2022 editions. The 2022 edition adds a reporting category, so the two series are not directly comparable.

II
Act twoThe cost

Data nobody trusts does not stop being used: it gets replaced by judgement. That has a price too.

04The price of error

Poor data quality never shows up as a budget line.

No budget contains a line reading “wrong data”. The cost spreads across reconciliations, reworked reports, mis-sized orders and decisions held back waiting for a figure someone trusts.

It remains an unsolved problem: only a little over a third of the organisations surveyed say they have managed to improve their data quality.

Reports having improved its data quality37%
of the organisations surveyed. The other two in three still decide on figures that do not add up.

Wavestone, 2024 Data & AI Leadership Executive Survey.

05Two ways to decide

The same decision, two sets of inputs.

Reviewing the price of one product line. The decision is identical in both columns; what changes is what goes on the table beforehand, and whether it can be revisited a quarter later.

Judgement
◆What came up in the last meeting
◆The customer who complained most
◆The historical reference margin
◆A spreadsheet from five months ago
Evidence
◆Observed elasticity of the product line
◆Actual margin per order, not the average
◆Cost to serve broken down by customer
◆The outcome of the three previous reviews

Produces a decision that holds up in the room and cannot be audited three months later.

Produces a decision with an explicit assumption, one that can be revisited and corrected.

III
Act threeThe path

Data maturity cannot be bought: it is climbed one step at a time, and no step is skipped.

06The ladder

Five levels of data maturity.

The right jump is always the next one. Attempting level five from level one is the most common way to spend an analytics budget without changing a single decision.

01ScatteredEvery department keeps its own spreadsheet and its own version of the figure.An inventory of sources.
02ReportedA monthly report exists, but it arrives late and gets argued over in the meeting.One definition per metric.
03TrustedOne figure per concept. Nobody questions it in the room.Automating the path the data takes.
04OperationalData enters the process, not just the presentation.Models on clean history.
05AnticipatoryThe decision is made before the problem shows up.Continuous improvement of the model itself.
07The compound effect

The return is not in one decision, but in its repetition.

A management team makes hundreds of repeatable decisions a year: pricing, purchasing, shifts, maintenance priorities. Improving each one by a point changes no quarter; it changes the fifth year.

Which is why internal analytics is poorly justified by an isolated case and well justified by the volume of decisions it touches.

One point of improvement, compounded
Year 12345

Conceptual illustration of the compound effect. It does not represent market data or a guaranteed outcome.

Self-assessment

Five questions to find your step.

Tick the ones you can honestly answer yes to. There is no good or bad result: there is a step, and a next move.

0 of 5 answered yes
01Scattered

Each department holds its own version of the figure. Meetings begin by arguing about the data, not the decision.

Next step

An inventory of sources and an agreed definition of the five metrics that are actually used.

Discuss it with us →
Methodological note

The figures quoted come from public third-party studies and are reproduced with their source and year. They are not results from Sertoria or its clients. The conceptual frameworks —the maturity ladder and the compound effect— are our own and are marked as such.

01Data & AI Leadership Exchange (Randy Bean, with DataIQ), 2025 AI & Data Leadership Executive Benchmark Survey.
02Wavestone, 2024 Data & AI Leadership Executive Survey.
03Seagate and IDC, Rethink Data: Put More of Your Business Data to Work—From Edge to Cloud (2020).
04Anaconda, State of Data Science, 2020 and 2022 editions.