Meter data management (MDM) basics for utilities
MDM sits between smart meters and billing, turning raw interval reads into validated data. Here is what an MDMS does, how VEE works, and why data quality matters.
Advanced meters produce an enormous amount of data, and almost none of it is useful to a utility until it has been validated. A meter data management system, or MDMS, is the software layer that does that work. It collects the reads coming off the metering network, checks them, fills gaps in a controlled way, and passes only trustworthy data to the systems that depend on it. As one plain definition puts it, an MDMS collects, validates, and routes meter reads to billing.
This is the explainer. If you are evaluating a system, the companion piece on meter data management system benefits and requirements covers the business case and the requirements checklist.
The scale is easy to underestimate. A single 25,000-meter utility reading on 15-minute intervals generates roughly 876 million reads per year. Across the U.S., there were about 119 million AMI installations as of 2022, around 72 percent of all electric meters. No billing system was designed to reconcile that volume directly, which is exactly why the MDM layer exists.
Where MDM sits in the stack
An MDMS sits in the middle of the metering data flow. Below it is the metering infrastructure and the head-end system that collects the raw reads. Above it are the systems that consume clean data: the MDMS sits between meters and downstream systems that depend on validated consumption data: billing, CIS, customer portal, analytics, and compliance reporting. Get the MDM layer right and everything above it inherits good data. Skip it, and anomalies flow straight to the bill.
VEE: the core of meter data management
The heart of an MDMS is a three-step rule engine called VEE, for Validation, Estimation and Editing. Each step has a distinct job.
Validation
Validation checks every read against configurable rules and flags anomalies: consumption that is far above or below the account's own history, zero reads, missing reads, and out-of-sequence reads. The point is to separate reads that can be trusted from reads that need attention, automatically, before any of them reach billing.
Estimation
When a read is genuinely missing, the MDMS fills the gap using historical data and statistical models, and records the estimate and the method used so it can be audited and later replaced by an actual read. Estimation is not a shortcut. It is a controlled, documented way to keep billing running when data is temporarily unavailable.
Editing
Editing is the controlled workflow that lets staff review and correct what validation flagged, with every change logged. That audit trail matters for regulatory compliance and for resolving customer disputes, because it shows exactly what was changed, when, and why.
A well-tuned MDMS resolves 90 to 94 percent of reads automatically on the first pass through VEE, so staff time goes to the small share of genuine exceptions rather than the whole stream.
Interval data and why granularity matters
Interval data is consumption recorded at fixed time steps rather than a single monthly total. Standardized energy data is commonly exchanged in 15-minute, hourly, daily, or monthly intervals. Granularity is what makes time-of-use rates, demand charges, load analysis and demand response possible, because they all depend on knowing when energy was used, not just how much. It is also what makes the data volume large, which is why validation has to be automated.
MDM feeds everything downstream
| Consumer of MDM data | What it needs | What breaks without validation |
|---|---|---|
| Billing / CIS | Validated usage per account and period | Estimated or anomalous reads become billing errors |
| Customer portal | Accurate, timely usage | Customers see wrong data and lose trust |
| Analytics | Clean, complete interval series | Load and program analysis is built on noise |
| Compliance reporting | Auditable, documented data | Reports cannot be defended to a regulator |
Choosing an MDM approach
Most organizations end up in one of three places, and the right one depends less on features than on who is going to own the validation rules once the project team disbands.
| Approach | Fits | What you take on |
|---|---|---|
| Dedicated MDMS platform | Distributors running AMI at scale, with regulated reporting and several downstream consumers | Licence cost, plus an integration project for each system that consumes the data |
| MDM module inside the CIS or head-end | Smaller distributors already committed to one vendor stack | Validation depth and analytics are limited to what that module chooses to expose |
| Data platform with a validation layer on top | Retailers, aggregators and large customers holding data from several distributors | You own the rule set and the monitoring, so it needs a named owner internally |
The third pattern is the one most large energy buyers arrive at without planning it, because they hold data from utilities that each deliver it differently. That is a different problem from running an MDMS: the reads are already validated by someone else, and the work is reconciling them into one series you can defend. Data quality across energy programs covers what that reconciliation involves.
What to ask before you buy
Feature lists rarely separate these products. The answers to these questions do.
- Which validation rules ship by default, and can we see the rule set itself rather than a summary of it?
- How is an estimated read produced, and is the method recorded on the record so it can be explained later?
- What does the exception queue look like for the person clearing two hundred flagged reads on a Monday morning?
- Is every edit logged with who changed what, when, and on what grounds?
- What interval granularity is supported, and how long is history kept at that granularity rather than summarized?
- How does validated data reach billing, the customer portal and analytics, and what happens when one of them is unavailable?
- Can we export the full validated series in a standard format if we move to something else?
- What pass rate is achieved on data like ours, rather than on the demonstration set?
The last question is the one worth pressing. A pass rate quoted without a matching data profile tells you nothing: a portfolio of commercial accounts with frequent meter changes will not behave like a residential AMI deployment.
How MartinAI complements meter data management
MDM validates reads as they arrive. MartinAI adds a second, independent layer of validation across the data that ends up on the bill. It structures bills, meter data and standardized data-sharing files into consistent records, then checks usage against reads, reads against history, and charges against the tariff. Where an MDMS confirms a read is plausible, MartinAI confirms the whole bill hangs together, and surfaces the exceptions with the supporting evidence.
- Turn interval and bill data into consistent, analysis-ready records
- Cross-check usage, reads and charges against each other and the tariff
- Flag anomalies against each account's own history
- Keep an auditable trail of what was checked and what was flagged
Frequently asked questions
What is a meter data management system (MDMS)?
An MDMS is software that collects, validates, stores and routes meter reads from smart meters or AMI infrastructure to downstream systems such as billing, the CIS, customer portals, analytics and compliance reporting. It is the data-quality layer between the metering network and everything that depends on validated consumption data.
What does VEE stand for in meter data management?
VEE stands for Validation, Estimation and Editing. Validation checks each read against rules and flags anomalies, estimation fills genuinely missing reads with documented calculated values, and editing is the controlled, logged workflow staff use to correct flagged exceptions. A well-tuned MDMS clears 90 to 94 percent of reads automatically.
What is interval data?
Interval data is energy consumption recorded at fixed time steps, commonly 15-minute, hourly, daily or monthly, rather than a single monthly total. Granular interval data is what makes time-of-use rates, demand charges, load analysis and demand response possible.
Do you still need MDM if you have a CIS?
Yes. The CIS bills and serves customers, but it relies on validated usage to do so correctly. The MDM layer is where raw AMI reads are checked and estimated before billing. Without it, anomalous or missing reads flow straight into the CIS and become billing errors.
Meter-to-cash: reducing utility billing errors at the source
Most utility billing errors start upstream, in the reads and data feeding the bill. Here is how the meter-to-cash cycle works and where to catch problems early.
Modernizing a utility Customer Information System (CIS)
A utility CIS runs meter-to-cash, but many were built before smart meters existed. Here is what modernization involves, what drives it, and how to de-risk the move.
Offering Green Button Connect My Data as a utility
Green Button Connect My Data lets customers authorize secure, automated sharing of their usage data. Here is what the standard requires and how to implement it well.
Utility data collection
Customers and sites onboard themselves. Green Button authorizations, interval feeds and bill uploads are normalized into one program-ready dataset.
Data & platform
All articles in this category.
Consultants & auditors
How MartinAI applies to this sector.
Utility bill overpayment calculator
Estimate the billing errors a portfolio is likely carrying.
Energy data by location
Grid, utility and reporting facts for your province, state or city.