Why you don’t trust the data in your company
Why You Don’t Trust the Data in Your Company
Data is supposed to make business easier.
It should help leaders make decisions with greater confidence, help teams identify problems before they become expensive, and give everyone a shared understanding of what is actually happening inside the organization. Yet in many companies, something rather peculiar happens: there is an abundance of data, but very little faith in it.
Executives ask for reports and then question the numbers. Managers open dashboards but still request spreadsheets. Employees spend hours checking figures against other figures because nobody is entirely certain which version is correct.
The result is a strange paradox. The company has become data-rich but certainty-poor.
If people do not trust the data, even the most sophisticated analytics platform can become little more than an expensive ornament. The issue is rarely that employees dislike data. More often, they have learned through experience that the data can be incomplete, inconsistent, outdated, inaccessible, or simply wrong.
So why does this happen?
The Real Problem Is Not Always the Data
When companies say they have a “data problem,” they often assume the solution is technical. Buy better software. Build a new dashboard. Introduce artificial intelligence. Hire data analysts. Migrate everything to the cloud.
Sometimes those investments are necessary.
But technology cannot magically transform unreliable information into reliable information. If poor data enters a system, an impressive visualization may simply make the poor data look more convincing.
Trust is built through repeated experiences. If a sales manager discovers that the monthly revenue dashboard differs from the finance report three months in a row, skepticism becomes rational. If a marketing team notices that campaign conversions are routinely adjusted weeks after reports are published, people stop treating the numbers as authoritative.
Eventually, employees develop their own informal systems.
They keep private spreadsheets. They download reports. They maintain personal databases. They ask colleagues for “the real numbers.”
That is when an organization has crossed an important threshold: people are no longer merely experiencing data-quality issues. They are experiencing a crisis of data credibility.
1. Your Data Comes From Too Many Places
Modern organizations rarely have a single source of information.
Customer details may live in a CRM. Financial records may exist in an accounting platform. Website behavior comes from analytics software. Marketing data resides in advertising platforms. Operational information may sit inside an ERP system, while employees maintain additional spreadsheets for everything the official systems cannot conveniently handle.
Each system may work perfectly well on its own.
The trouble begins when the company tries to combine them.
Imagine that the CRM says a company has 8,450 customers, while the billing system says there are 8,217 active accounts. The marketing platform reports 9,030 contacts. Meanwhile, a spreadsheet maintained by the sales department lists 8,611 prospects and customers.
Which number is correct?
There may be a perfectly reasonable explanation for every discrepancy. Different systems may define “customer” differently. Some records may be inactive. Some contacts may represent multiple accounts.
But if nobody understands those distinctions, the numbers appear contradictory.
Over time, employees stop asking which number is technically correct and start asking which number is politically acceptable.
That is a dangerous development.
2. Definitions Are Inconsistent
One of the most underestimated causes of data distrust is semantic ambiguity.
A company may use the word “customer” as though everyone understands what it means. But does a customer mean someone who has purchased something? Someone with an active subscription? Someone who purchased within the last twelve months? A business account containing several individual users?
The same problem occurs with terms such as revenue, lead, conversion, churn, active user, qualified opportunity, retention, and profitability.
Two departments can use identical terminology while referring to entirely different measurements.
Marketing might define a lead as anyone who completes a form. Sales might consider a lead meaningful only after qualification. Finance may count revenue according to accounting principles that differ from the timing used by the commercial team.
None of these definitions is necessarily wrong.
The problem is that the organization has no common lexicon.
Without standardized definitions, dashboards become semantic labyrinths. People spend more time debating what a metric means than discussing what the metric implies.
3. Data Is Often Incomplete
Incomplete data is particularly corrosive because it can look legitimate.
A report containing thousands of rows feels substantial. A dashboard filled with charts looks authoritative. Yet missing information can quietly distort the conclusions drawn from it.
Suppose a company wants to understand customer churn but only has cancellation information for customers who explicitly contact support. Customers who simply stop renewing may not be captured in the same dataset.
The report still produces a number.
That number may even look precise.
Precision, however, is not the same thing as accuracy.
A figure such as 7.42% can create an illusion of scientific rigor while concealing substantial gaps in the underlying information. The decimal places may be impeccable. The methodology may be questionable.
This is why data completeness deserves as much attention as data accuracy.
4. Nobody Knows Where the Numbers Came From
Another major problem is lineage.
When someone asks, “Where did this number come from?” there should be a straightforward answer.
Instead, employees sometimes hear:
“It comes from the dashboard.”
That is not an explanation.
A trustworthy data environment allows people to trace information backward. They should be able to understand the original source, transformations, calculations, filters, and assumptions involved in producing a particular figure.
Without this lineage, data becomes opaque.
Consider a financial KPI displayed on an executive dashboard. If nobody can explain which database supplied the underlying records, which formulas were applied, when the information was refreshed, and which exclusions were used, executives are justified in treating the KPI cautiously.
Trust requires provenance.
People are much more willing to believe information when they can understand its journey.
5. The Data Changes Without Warning
Data is not static.
Customer records change. Transactions are corrected. Historical figures are reclassified. Marketing campaigns are renamed. Products are discontinued. Employees modify records.
Changes are normal.
Unexpected changes are not.
Imagine presenting a quarterly report to senior leadership on Monday and discovering on Thursday that the historical figures have changed because a data pipeline was modified. Even if the revised numbers are more accurate, confidence in the reporting process will suffer.
People begin wondering whether yesterday’s figures were ever reliable.
This is where versioning and governance become important. Organizations need mechanisms for recording significant changes, communicating them, and understanding their impact.
Otherwise, every revised report becomes a small earthquake.
6. Spreadsheets Have Become Shadow Systems
Spreadsheets are extraordinarily useful.
They are also remarkably good at becoming unofficial infrastructure.
An employee may export data from the CRM, clean it manually, combine it with information from another system, add formulas, and create a report that everyone eventually depends upon.
The spreadsheet may contain critical business logic.
Nobody else may understand it.
If that employee leaves the company, the organization can suddenly discover that an apparently mundane Excel file was actually holding together an important operational process.
These shadow systems emerge because official systems often fail to accommodate real-world workflows. Employees build what they need.
The lesson is not that spreadsheets are bad. It is that widespread spreadsheet dependence can signal deeper shortcomings in data architecture, integration, or usability.
When people repeatedly create their own version of the truth, the official version may not be serving them.
7. Data Quality Is Nobody’s Job
Perhaps the most fundamental issue is ownership.
When data is everyone’s responsibility, it can easily become nobody’s responsibility.
Who decides whether a customer record is complete?
Who determines the correct format for industry classifications?
Who investigates duplicate accounts?
Who approves changes to important metrics?
Who is responsible when two systems disagree?
If the answer to all of these questions is vague, data quality will inevitably deteriorate.
Good data governance does not mean creating an enormous bureaucracy around every field and spreadsheet. It means assigning clear accountability.
Specific people or teams should understand which datasets they own, what standards apply, how quality is measured, and what happens when those standards are breached.
Accountability gives data stewardship a tangible shape.
8. People Remember Bad Data
There is also a psychological dimension.
Once someone has been burned by unreliable data, they remember it.
A manager who made a poor decision because an operational dashboard was wrong may become skeptical of future dashboards. An analyst who spent two days reconciling conflicting figures may become reluctant to use automated reports. An executive who discovers that a supposedly critical KPI was based on outdated information may begin demanding manual verification.
This creates institutional memory.
And institutional memory can outlive the original problem.
Even after the data infrastructure has improved, people may continue distrusting it because their mental model has not caught up with the technical changes.
That means rebuilding confidence requires more than fixing databases.
It requires demonstrating reliability repeatedly.
9. Too Much Data Can Reduce Trust
More data does not automatically produce more certainty.
In fact, an excess of poorly organized information can create what might be called informational sediment: layer upon layer of metrics, reports, dashboards, extracts, and indicators until the organization can no longer distinguish signal from noise.
A leadership team might have hundreds of KPIs available.
How many actually matter?
When every metric receives equal visual prominence, employees struggle to identify what deserves attention. Contradictory indicators become especially troublesome.
A successful data strategy therefore requires curation.
The goal is not to measure everything.
The goal is to measure what matters reliably.
10. The Company Measures Performance Differently Across Teams
Trust can also deteriorate when departments are rewarded according to conflicting metrics.
Sales may focus on new contracts. Customer success may prioritize retention. Marketing may optimize lead volume. Finance may emphasize profitability.
Each team can appear successful according to its own dashboard while the company as a whole moves in the wrong direction.
This is not necessarily a data-quality problem in the narrow technical sense. It is a measurement architecture problem.
Data reflects organizational priorities.
If those priorities conflict, the numbers will inevitably tell competing stories.
A coherent measurement framework should therefore connect departmental metrics to broader business objectives. Otherwise, teams may optimize their local performance while unintentionally undermining the wider organization.
How to Start Trusting Your Data Again
Restoring confidence does not require rebuilding the entire technology stack overnight.
Start with the most important business questions.
Identify the handful of metrics executives and operational teams genuinely depend upon. Then examine those metrics from source to dashboard.
Where does the information originate?
Who owns it?
How frequently is it updated?
What transformations occur?
What definitions are being used?
What known limitations exist?
This exercise often reveals that the most important problems are surprisingly mundane.
A field may not be consistently populated. Two systems may use different customer identifiers. A calculation may have been copied from an old spreadsheet. A dashboard may refresh once a day while users assume it is real-time.
Small discrepancies can have enormous consequences when they sit inside important decisions.
Establish a Single Source of Truth
A single source of truth does not necessarily mean having one database for everything.
It means having an agreed-upon authoritative source for specific categories of information.
Finance should know which system defines official financial figures. Sales should know which system defines pipeline data. Marketing should understand which dataset governs campaign performance.
Clarity matters more than architectural purity.
Create a Data Dictionary
A data dictionary can eliminate enormous amounts of confusion.
For every important metric, document its definition, calculation, owner, source, update frequency, and limitations.
If “customer churn” has a precise organizational definition, everyone should be able to find it.
This transforms tribal knowledge into institutional knowledge.
Measure Data Quality
What gets measured tends to receive attention.
Organizations can monitor dimensions such as:
- Accuracy
- Completeness
- Consistency
- Timeliness
- Uniqueness
- Validity
The objective is not to achieve perfect data. That is often unrealistic.
The objective is to make quality visible.
Explain Changes
When a number changes, explain why.
A short annotation such as “historical revenue figures updated following accounting-system reconciliation” can prevent unnecessary confusion.
Transparency is an underrated component of trust.
Make Good Data Easier to Use
If the official data is cumbersome to access, people will create alternatives.
That is not necessarily user laziness. It is often a rational response to friction.
Employees need intuitive tools, reliable dashboards, sensible permissions, and straightforward workflows. When the trustworthy path is also the easiest path, shadow systems become less attractive.
The Difference Between Data Accuracy and Data Trust
Ultimately, trust the data is not merely a technical objective.
It is an organizational one.
A dataset can be statistically accurate and still fail to inspire confidence if nobody understands its origin. Conversely, a dataset with known imperfections can remain useful when those limitations are clearly documented and consistently understood.
Trust emerges from several ingredients working together: quality, transparency, consistency, ownership, governance, accessibility, and experience.
That last element is crucial.
People trust systems that repeatedly behave as expected.
The goal, therefore, should not be to convince employees that every number is infallible. No serious data environment can make that promise. Instead, organizations should create an environment where people know which numbers are authoritative, understand how those numbers were produced, recognize their limitations, and have a clear mechanism for challenging questionable information.
That is a much more durable form of confidence.
Conclusion
The reason people do not trust the data in their company is rarely because they are inherently skeptical of technology.
They have usually been given reasons to be skeptical.
Conflicting systems, ambiguous definitions, incomplete records, mysterious calculations, unstable reports, spreadsheet dependencies, unclear ownership, and poor communication can gradually erode confidence until employees instinctively verify everything themselves.
And once that happens, the cost extends far beyond data management.
Decisions become slower. Meetings become longer. Analysts waste time reconciling reports. Managers rely on intuition because the numbers seem questionable. Executives receive competing versions of reality.
The irony is striking: companies invest heavily in collecting information, only to discover that the real scarcity is not data.
It is credibility.
Building that credibility requires discipline. Define important metrics. Establish ownership. Document lineage. Improve quality. Reduce unnecessary complexity. Communicate changes. Make reliable information easy to access.
Most importantly, create a culture in which data can be questioned without being dismissed and trusted without being worshipped.
Because the objective of a mature data organization is not simply to produce more numbers.
It is to produce numbers that people can confidently use to make better decisions.
Why you don’t trust the data in your company
Everyone wants to use the data. Big Data is in fashion, the Internet of Things has multiplied the opportunities to obtain statistics, the blockchain promises to guarantee the integrity of the data, and the achievements of companies that have already started using it are difficult to contest. However, there still seems to be a brake on use: we don’t quite trust the data our companies obtain.
This is demonstrated by a recent study by the consultancy KPMG in which they have interviewed almost 1,300 CEOs from countries such as the United Kingdom, Spain, the United States and Japan and belonging to a wide variety of sectors: automotive, banking, infrastructure, insurance, manufacturing, retail, etc.
From the outset, these CEOs are aware of the usefulness of data and its analysis. Those surveyed recognize that the use of numbers will be essential for the evolution of companies. To the point that 48% of them believe that there will be profound changes in their sector over the next three years due to technological innovation. But there is still a leap when it comes to moving from this recognition to the use and application of data, and not everyone dares to take it.
This is especially relevant in our world, that of logistics, where the integration of data throughout the supply chain is increasingly important and in which a multitude of numbers are produced in order to analyze its performance.
Why we don’t trust the data
The efficient use of data increases sales and profitability of companies. It allows them to get to know themselves, their clients, what they do well and what they do badly, what their clients stand out from them, etc. Ultimately, it gives them a framework of real numbers from which to make better decisions. Not trusting data analysis means turning your back on reality, in addition to investing money in obtaining and analyzing it and then not taking advantage of it.
The KPMG study shows some startling statistics that add to the problem of not trusting our own numbers:
Only 19% say they have no reservations about the validity of the data with which they have to make decisions.
36% say they can’t make data-driven decisions until they make significant investments in it.
45% consider that the depth of their knowledge regarding their clients is limited by the lack of quality of their figures.
Among the aspects that foster this distrust is the lack of familiarity with numbers. When the people who have to use them are suspicious of the algorithms that generate this data, and see it as something obscure, do not understand how they were obtained or do not handle the software used, the chances of these figures being left aside increase. Managers – and not just CEOs – must learn to trust the numbers and insights provided by their data analysis departments.
“Trust in data analysis should not be negotiable”
Have you ever been in a meeting where the question was asked: “And where does this number come from?” or a similar phrase? If so, it’s a good clue that there’s a lot of work to be done regarding the culture of number use in the company. As KPMG’s own report states: “Trust in data analytics should be non-negotiable.”
How to trust our data
KPMG proposes four areas for improvement to increase confidence in the data and its analysis.
Quality
The first of all is to be able to ensure that our data is reliable. Being able to be sure that when we account for a variable or a process we are obtaining the real data and that when we create an algorithm we are capable of taking into account all the variables it needs to be a reflection of reality.
Effectiveness
Are you doing something with your data? Simply recording them, or even analyzing them without using the findings to make decisions, is a waste of time. What’s more, it will make the company’s personnel wonder why the numbers are collected and interpreted if they are not used later.
Integrity
Data protection is a matter of global importance. Businesses need to ensure that the collection and use of data is done in accordance with the law and the privacy of their customers. Only once these bases have been secured can data analysis be made one of the pillars in business management.
Resilience
Is your data analysis prepared for the medium and long term? Your strategy for its use should be designed for tomorrow and be easy to adapt to changes that may arise. Whether they are changes regarding the way of collecting the data or about the needs that our data have to respond to.
It is very common that when using data to make decisions, companies believe that they are much further from being able to do so than they really are. To complete this leap, it is essential to introduce the culture of numbers into the company. Explain why they should be a basic starting point for management and why leaving it out is competing at an inferiority.
It is also vital to carry out training programs. This will help workers to live naturally with the numbers and to be more reliable when entering data manually into the system (human error continues to be a very important factor in this type of failure and, therefore, in why do we doubt the data).
Another essential step is to clearly establish the criteria for our statistics. For example, in the case of logistics: what do we consider a correct shipment? According to the companies (or in the different delegations of the same company) the definition may vary. Some will equal it to delivery on time, others will add the fact that no claims are made, others will include that all the products in the order were available at the time they were requested, etc. Without clearly unifying these criteria, your figures will never be valid. An idea that also serves to integrate the various computer systems, which often represents another major obstacle.
Looking to the future, data analysis and Big Data do not stop gaining importance in companies. In many cases, logistics will be in the eye of the hurricane of this change and you will have problems if this process catches you on the wrong foot.
Time
The more urgent the delivery service, the more expensive it will cost us to send a pallet. The two most common services for sending a pallet in Spain are the economic (48 hours) and the express (24 hours). Networks specialized in pallet distribution are constantly trying to reduce delivery times, as in the case of the a.m. of Palibex, with which we offer deliveries before 12 in the morning of the following day. Times that bring us closer to urgent parcel services.
Being proactive is one of the easiest ways to save costs on your shipments. More often than not, lack of foresight is among the reasons behind an urgent shipment. So pallets that could have gone cheap end up paying more than they should.
Additional concepts
Fate, size, time, and weight are the first concepts that would come to mind for all of us, but the list doesn’t always end there. There is a wide variety of additional items that can add to how much it costs to ship a pallet. Do you need your pallet to be depalletized during delivery? Is it a delivery at street level, in a basement or do you have to go upstairs? Do you need a hatch or other additional means of unloading?
Another expense that usually catches many people off guard is taxes and customs procedures for shipments to the Canary Islands, Ceuta and Melilla. Although the recipient is often in charge of paying the taxes, it will always be better to clarify it with him before shipping, or we run the risk that the merchandise will be retained at customs and we will have one arranged between both parties.
There are even more concepts that may appear reflected on your invoice -such as gasoline, although it is more common to see it in international transport-, so a good rule of thumb is to consult your logistics operator for its possible existence.
Once you have these basic principles clear, it will be much easier for you to calculate how much it costs to send a pallet. As we mentioned at the beginning, by playing with weights and sizes you can start saving on your logistics, studying whether you can wait a little longer to gather more merchandise on your pallet, whether you can distribute the weight more appropriately according to your rates or how to assemble the merchandise on your pallets and up to what height.
