One metric, several values
Finance, sales and operations use different formulas, periods or levels of detail for a metric with the same name.
The meeting begins with the reconciliation of the numbers, not with an analysis of the reasons for the variance.
Reporting depends on manual reconciliations
Regular reporting requires exports, spreadsheets, adjustments and the knowledge of individual analysts.
BI does not increase trust in data
The number of dashboards grows, while managers still need to confirm calculation methods and data sources.
Changing the source imperceptibly changes the result
Updating the system, status or processing rules affects the indicator without established control and approval.
Why this happens
The same metric may be calculated differently in financial, commercial and production reports. Accounting balances may differ from warehouse records, while the order portfolio may depend on which statuses an analyst selects.
Management meetings often spend much of their time reconciling source data and calculation methods. Managers debate which figure is correct instead of examining the cause of a variance and deciding what to do.
This practice establishes shared metric definitions, data lineage, ownership and quality rules, so consolidated results can be traced back to underlying records and the accountable person.
Three conditions for reliable management metrics
Metric definition
What exactly is being measured?
Each metric has a defined meaning, formula, period, analytical dimensions, level of detail and scope of use.
Name, definition, formula, units of measurement, period and detail.
Data provenance
How is the final figure calculated?
The path from the initial operation through sources, transformations, reconciliations and consolidation rules to the final presentation is fixed.
Source system, transformations, control points, and history of change.
Ownership and quality
Who is responsible for data and action?
For the indicator and primary data, owners, quality rules, correction procedure and action in case of variance are fixed.
The owner of the indicator, the data owner, the quality criteria and the deadline for eliminating the error.
More about the model
A single version of the truth does not necessarily require one database. It requires a shared definition of each metric, a traceable calculation path and clear accountability for source data quality.
The indicator becomes manageable when the manager understands what it means, where it came from, and who should act in case of variance.
How we build management insight
We start with the questions management needs to answer, agree a metric model, establish data lineage and quality controls, and then embed analytics in the regular decision-making cycle.
Define management questions and data requirements
Determine what decisions are made by the owner, board of directors, CEO and functional managers and what data is needed for each level.
More about our work
Starting point
Reporting follows available data and system capabilities rather than specific management questions.
What we do
We record decisions, indicators, the necessary detail, the frequency of updating, the permissible variance and the action of the head.
How to check the results
Each report has a clear management question, user and action to take when a variance arises.
We agree on the common meaning of indicators
One name corresponds to one definition, one formula and established detail.
More about our work
Starting point
Functions use different formulas, periods and analytical dimensions for metrics with identical names.
What we do
We create a dictionary of indicators, fix definitions, formulas, periods, details, owners and relationships between indicators.
How to check the results
Finance, operations and sales obtain the same result from agreed source data.
Record the origin and quality control of data
The path of the indicator is traced from the final presentation to the initial operation.
More about our work
Starting point
The method of reconciliation and transformation is known to individual analysts, and errors are compensated by manual adjustment of the report.
What we do
We document source systems, transformations, reference data, checkpoints and checks for completeness, timeliness and consistency.
How to check the results
For variance or error, the place of origin, the owner of the original data and the deadline for eliminating the root cause are visible.
Incorporate analytics into the management cycle
A dashboard or report follows the sequence: metric — variance — cause — owner — action.
More about our work
Starting point
BI demonstrates available data, but does not always support a specific decision-making cycle.
What we do
We develop analytical products for defined management questions, with detail on causes, owners and scenario views.
How to check the results
Managers can move from a summary metric to its cause, source transaction, owner and subsequent action.
Services within this practice
Engagements and deliverables: from data audits and metric models to BI platforms, data warehouses, quality controls and predictive analytics.
Data and reporting assessment
A map of discrepancies in metrics, sources, reports and manual reconciliations, showing where data conflicts, lacks ownership or prevents management decisions.
Designing a metrics model
A unified system of management metrics: definitions, formulas, sources, owners, frequency, levels of detail and rules for interpreting results.
Implementing BI platforms and management dashboards
Working dashboards and reports for managers and business functions: metrics, filters, access roles, drill-down, data refresh and regular usage scenarios.
Building data warehouses, data marts and pipelines
Implemented data layers for management analytics: sources, transformations, data marts, change history, quality controls and consistent calculation logic.
Establishing data governance and quality controls
Rules for data ownership, quality control, reference data management, error resolution and accountability for metrics used in management.
Implementing predictive and scenario analytics
Analytical models for forecasting demand, capacity utilisation, liquidity, variances, risks and delivery dates, with clear assumptions and limits of use.
What the management team receives
Management question map
Decisions, metrics, detailing and frequency of data for the owner, board of directors and function managers.
Shared metric model
A glossary of definitions, formulas, periods, analysis dimensions, relationships and owners.
Data lineage map
The path from the primary operation through systems, transformations and reconciliations to the final indicator.
Master data model
Uniform rules for customers, counterparties, materials, facilities, projects and other key entities.
Data quality rules
Checks of completeness, timeliness and consistency, owners of errors and deadlines for elimination.
Analytical products
Views following the sequence: metric — variance — cause — owner — action, including predictive and scenario models.
What changes in data and reporting
Results are visible in the ability to reproduce a metric, trace its origins and move from a variance to its cause and a corrective action. The number of reports is not the measure of success.
One metric, one meaningFinance, operations and sales use a shared formula and level of detail.
Starting point
One indicator is calculated using several methods, and the result depends on the unit or report.
What's changing
For the indicator, a single definition, formula, period, detail and owner are fixed.
How to check the results
The same set of inputs reproduces the same value in all validated representations.
Traceability to source transactionsThe final figure is traced to the initial operation and the place of occurrence of the variance.
Starting point
The final indicator is available, but the path of its formation and transformation is not obvious.
What's changing
The source, transformations, reconciliations and the owner of each stage of the indicator formation are known.
How to check the results
From the management report, you can move to the original transaction, document or event.
Reporting without manual reconciliationRegular reporting is reproduced according to the established methodology.
Starting point
Each reporting period requires manual collection, reconciliation and adjustment by employees.
What's changing
Discrepancies are identified according to the rules, and the root cause is eliminated in the source system.
How to check the results
The preparation of regular reports does not depend on the informal actions of individual analysts.
Comparability of companiesLegal entities, sites and business units are compared using a consistent methodology.
Starting point
Group companies use different reference data, statuses and metric interpretations.
What's changing
Shared definitions, reference data and consolidation rules are used.
How to check the results
The results of the companies and sites are comparable without additional manual methodological adjustment.
A reliable basis for forecasting and AIPredictive models and AI use controlled and explainable data.
Starting point
New analytical models reproduce errors and discrepancies in the initial reporting.
What's changing
For the data used, the meaning, origin, quality, limitations and owner are known.
How to check the results
The premises of the model are available to the user, and the variance of the result can be associated with the original data.
When this practice is especially useful
This practice is needed when data and reporting volumes grow faster than the company can agree metric definitions, sources and quality rules.
Different answers to one question
Managers receive different values of revenue, margin, inventory or order portfolio
More details
- Units use different formulas
- The period and detail of the indicator are different
- The meeting begins with the reconciliation of numbers.
- The source of the final value is not obvious.
- No methodology owner is assigned
Manual compilation of reporting
Regular reporting depends on individual analysts and informal reconciliations.
More details
- Data is collected from several exports
- Adjustments are made in the tables
- The rules are known to a narrow circle of employees
- Reporting takes up a significant part of the period
- Changing team composition disrupts reproducibility
BI without trust
There are more analytical dashboards, but trust in the metrics has not increased
More details
- Different dashboards show different values
- Visualization does not reveal the origin of the figure
- Indicators are created for individual requests
- lacks a single dictionary
- Errors are compensated for by adjustments to data marts
Group, forecasting and AI
The company scales analytics, but the initial definitions and data quality are not fixed
More details
- Companies use different reference data
- site results are incomparable
- Changes to the source system affect reporting without control
- Forecasts use undocumented assumptions
- AI scenarios run without quality assessment of source data
This practice in projects
