What is a System of Record?
A system of record (SOR) is an authoritative source of business data. Systems of Record can contain authoritative data on customers, employees, products, suppliers or other assets and entities related to everyday business processes.
The first use of the term appears to be by the US Social Security Administration in the 1930’s describing manual payroll records but the concept of keeping source records goes back thousands of years. With the advent of enterprise computing in the 1970s and 80’s, the term was often used by companies like IBM, Oracle, SAP and Microsoft to describe source records from which business processes would be generated.
You may have heard of different ways of describing a System of Source. Some popular synonyms include: “Single source of truth”, Authoritative data source”, Source system of record” or, “Database of record”.
Why is a System of Record Important?
Systems of Record are critical to many business processes. If you don’t have a place to record whether or not you came to work or what you did with your time we lose the ability to look back and see if you should be paid at all. Some of that could be recorded automatically. We might have an access door code of you entering and leaving the building. But if no one signs off on that being accurate, then the record becomes less valuable. And we can do some analysis of what time you spent at your computer or at a manufacturing station but the auditable record of that time will ultimately be what we use to measure progress of both your workday and your productivity.
Auditors will often start an audit by an end result. They will look at a list of accounts receivable records for example. Then they will start to trace that record back. “Where did this get generated from?” and for each element they are given, they will trace that back, “and where did this invoice get generated from?” until they can see the source record. When an auditor can be returned to an original source record and verify that it matches what was in the amalgamated, synthesized final product, the audit is complete. If they are unable to get back to a source record, the audit becomes problematic.
System of Record Examples
You are surrounded by examples of Systems of Record that you don’t typically think much about but which may be crucial to you personally or to your business. CRM systems note interactions with clients, Medical records include not just the treatment you received but the notes by both you and your doctor over diagnosis and treatment. HR keeps a record of your application for employment, your signed agreement to employment agreements and your accrued vacation and vacation taken. Payroll keeps a record of what you are due for pay and what you have been paid in the past along with the timesheet entries that validate that pay if that’s relevant to your remuneration.
Characteristics of a System of Record
Single Source of Truth:
A System of Record must be an authoritative system where original data is created.
Audit Trails & Access Control:
A System of Records must structure data with strict permissions and ability to trace any changes once data has been accepted into the system.
Data Integrity:
A System of Record must capture and reduce errors that occur when data might be entered incorrectly.
TimeControl is a System of Record
We knew from the beginning of making timesheet systems that what we were creating would have to be a System of Record. That predates even TimeControl being released as a commercial product. Perhaps our perspective was affected by our first couple of customized timesheet development projects where the timesheet would be used for both project management progress reporting and for payroll. The project managers were less concerned over the auditability of the timesheet records. The Payroll managers were not.
So as we created TimeControl as a commercial of the shelf product, we knew that it would be called upon to produce auditable records that could be reviewed for Payroll, Accounting, Tax purposes and more. We needed records that could not be locked and not altered casually after the fact. We needed to ensure that security was managed in a way that data was accessible but only to those people who would need it. We needed a way for the people entering the data to be able to sign off on it electronically with a user name and password or some sign-in mechanism and we needed to track every change in the final records by source. After all what’s a source record if you don’t know the source of it?
This is what made TimeControl such a powerhouse of an enterprise product. No one thinks of whether data is auditable or not until an auditor shows up and says “Prove it.” For anyone who has ever done an audit whether that was just of a payroll cycle, an R&D tax credit, a Sarbanes Oxley report or a contract compliance report, you know what we’re talking about.
What about AI then?
We’re often asked these days “Why can’t AI just fill in the timesheet for us?” Of course it could, but then the AI algorithm would be the record of source. “Sounds great,” says the end user who never liked doing their timesheet anyway but the result might not make them happy. Perhaps the AI algorithm has determined that by its measure you haven’t put in enough time on valuable tasks to warrant the pay you’re expecting so it’s decided to reduce your pay. Perhaps AI has determined that what your timesheet should have had in it was more aligned with what the project plan was, so that’s what it entered on your behalf. The pay works but now the amount of effort on each task is an AI estimate, not an input from the person doing the work.
No, there is always going to be some need for a human to make an entry of some kind. Not always and not in every situation but timesheets is going to need human entry for the foreseeable future.
What Artificial Intelligence can bring to systems absolutely depends on it having access to some source data. After all, if AI is just feeding on analytically created data and then creating more analytically created data as a result, the end product becomes a vicious cycle of analysis looking at analysis. The system of record grounds that analysis to say “Start with data that we can count on because we entered it and checked it and tested it.
The place where AI can likely make a difference in the timesheet business is once that data has been collected, approved and stored. Now, AI algorithms can reach into large volumes of data and look for patterns or trends in a way that we could only dream of in early data-mining exercises. Indeed, the structure of timesheet data is ideal for AI trend analysis. It is highly structured data with dates, volumes, people, meta data and more. At HMS we see a place for AI to deliver value in numerous ways including reporting and in more analytical exercises such as project management within our TimeControl Project functionality.
There will always be a need for Systems of Record
But of course there will. Core data, Source data, the “Single Source of the Truth” data is always going to be needed and that means that systems that can receive and store such data along with its own audit of who entered it and how it might have been changed over time are likely to be around for as long as we can see forward.
