How to Spot a Fake Data Model

Why is the Data Modeler and your Data Model More Important than the CEO, all C-Level Staff, and the Board of Directors?

Data models can only be created correctly by actual real data modelers with real data modeling budgets. Without a real, correctly made data model, there is no way to utilize your data in real-time or at all in most cases. That means you cannot utilize proactive management anywhere in your entire company or government organization. In that case, every decision becomes reactive. That means no one has true complete control if any control at all. The entire management staff may be useless because they have no reliable data upon which to make proactive decisions.

Without real-time data, reports, and metrics, management cannot optimize business processes. All of the reports will have fake or outdated data that management won't see until 3 months to a year after something has happened. Management cannot detect fraud. Management cannot conduct financial audits with real-time software. Management will not even know if the data is secure, who is securing it, where they came from, nor if they are qualified. Reactive management will always run every company or organization into bankruptcy. No company can survive without its data.

To survive, make sure you have a real data modeler, a real data model, with verification that every penny allocated to data modeling actually goes into the process of data modeling; remote hardware, remote facility, remote contractor, remote data software independent of anyone within the company likely to commit fraud or breach the data. Nearly all data breaches start from within and with internal employees who are unsatisfied with their pay, lack of promotion, etc. Again, the data modeler is more important than any other resource.

FAQ: What the Heck is a Data Model?

What is a data model? A data model is a software representation of the business requirements and security requirements of an organization. Each data model is unique to the organization for which it was made because it is always based on their specific business rules and requirements.

Data models are used to store and secure corporate, private, personal, and government data. The data model structure enables activities such as reporting, KPI metrics generation, business intelligence, proactive management using real-time data, automation, cost reduction, etc. Essentially, nothing works correctly company-wide if there is any issue with the data model. If your KPI metrics and reports are not available in real-time, you cannot change date ranges, you cannot slice and dice the data any way you need, or the reports don't work at all or produce incorrect results, and/or your BI software does not work, your data model is likely fake.

Especially if your BI software does not work, your data model is likely fake, because modern BI software only works out of the box with data models that follow the known published strategy star schema or extended star schema. Why is that? All modern software systems are written using object-oriented code and programming languages; Java, C++, Objective-C. They have to have a known compatible structure they can automatically understand to communicate with data systems. If the structure is legacy or custom, they cannot figure out what it means.

Modern real-time data secured data models must be able to match OOP structure exactly, object-oriented programming, which includes encryption and cyber security, and therefore modern data models must always be atomic, multidimensional, 2NF, extended star schema data models. There is no other structure that works with object code. Anyone who says there is will likely be a con artist running a scam with a fake data model and insecure legacy software. All other non-star schema structures are legacy compatibility structures, have poor performance, usually support one concurrent user, and cannot be secured using modern encryption and security software, AES encryption, etc.

How to tell if you have a real data modeler instead of a fake one? The data modeler must always be able to provide these four minimum items of knowledge.

1. Ask, "What is data modeling?" If they say modeling data, they are fake and have no idea what it is.
2. Ask to see the DDL and UML of a fully atomic star schema data model they have made. If they don't have any data models, that person is not a data modeler.
3. Ask for the business requirements used to create the data model even if they came from a third party, such as the public FHIR requirements.
4. Ask, "Who or what role is usually the most important counterpart on a data modeling project for a data modeler?" If they do not answer "Business Analyst," that person is not a real data modeler. The data modeler and business analyst are like the baseball catcher and the pitcher respectively. The pitchers get all the glory, but the strategy and home plate defense come from the catcher. Pitching is like throwing the business requirements to the data modeler. You can't win a game if either the pitcher or the catcher is missing or not dedicated to each other. Similarly, a Senor business analyst is like a quarterback. An auxiliary, assistant or junior business analyst is like a running back who helps move the ball across the field even if it's just one to ten yards at a time. The wide receiver is like a data modeler who scores the long-range touchdowns. The data modeler knows how to catch the business requirement, run and dodge all of the obstacles like fraud, legal issues, etc., and finally get the ball into the in zone.

Improper Cost Cutting & Embezzlement of Wages and IT Budgets

What causes fake data models and fake data modelers? The short answer is MONEY and GREED! All companies want to cut costs and all managers want to look good by cutting costs, especially if they have no or limited technical skills and experience. However, the biggest problem is that most people are not experienced enough to create a valid cost-cutting plan. Most people in the current generation do not know that IT itself is a cost-cutting strategy.

When one cuts costs from IT or embezzles money from the cost-cutting strategy, costs go up exponentially. Corrupted/fraudulent IT increases costs and reduces profit. It may also cause the entire organization to go bankrupt, as over 700 US banks have in less than 25 years after implementing the failed H1B foreign labor strategy from the George HW Bush administration to cut wages in the middle class inside the US.

1. https://www.fdic.gov/resources/resolutions/bank-failures/failed-bank-list/
2. https://www.forbes.com/sites/stuartanderson/2020/10/22/economic-research-exposes-significant-flaws-in-dol-h-1b-visa-rule/?sh=a0002c661468

Staffing agencies have also been caught putting foreign workers into data security and accounting positions to compromise US security. The agencies were aware that the workers were not qualified, faked certifications, and then profited from the fraud. However, the foreign workers committed even more fraud on their own and got caught. https://www.sec.gov/news/press-release/2022-114 The problem is so large that almost all staffing agencies have criminal records for fraud and job discrimination against US Citizens with real experience. They can't steal money if real, ethnically diverse professionals are on the job watching.

The biggest target for improper cost-cutting and embezzlement is data modeling. The minimum corporate budget accounting line-item cost for data modeling is always $1,000,000, the rate from 1993, which was the year the original standard was released to the public, after having been classified for decades. When middle managers see $1,000,000, they want to steal it.

Wage Theft Examples

1. https://www.epi.org/publication/new-evidence-widespread-wage-theft-in-the-h-1b-program/
2. Fireing of disabled people or minorities in order to steal their wages by leaving wages in the budget, but not paying them to people. It looks legitimate on accounting reports. https://www.eeoc.gov/newsroom/eeoc-sues-state-contractor-and-staffing-firm-disability-discrimination

Wages, education funding, and retirement funding are easy targets for fraud in states built with slave labor that still have an intact slavery mentality, racism, discrimination, and weak or unenforced labor and citizen protection laws. Data modelers have always been rare and there were only 309 real data modelers trained by the US Government and the intelligence community before the funding was cut for the training program. That was 25 years ago. A few others were trained by the Ralph Kimball Group.

Unlike Europe, there are no job protection laws in the US. Therefore all workers are subject to harassment by both managers and customers and being forced to work free overtime in slavery-like conditions. It is easy for foreign workers, corrupt managers, or scammers to orchestrate the removal of data security staff inside any company, and then orchestrate a data breach to get personal data they can use to commit fraud, black market organ theft, human trafficking, illegal data brokerage, call center take over for phishing, etc. If an organization loses or alienates the real data modeler, it's unlikely they will ever find another one. After the real data modeler and security professionals are gone, there is theft and fraud.

Examples:

1. Fraud- https://www.deccanherald.com/india/americans-duped-into-losing-10-billion-by-illegal-indian-call-centres-in-2022-report-1175156.html#
2. https://indianexpress.com/article/world/indian-american-tech-fraud-arrested-new-jersey-8918717/
3. Scam Industry - https://www.youtube.com/watch?v=7CZReZ24-to
4. https://www.justice.gov/opa/pr/justice-department-settles-it-recruiter-resolve-immigration-related-discrimination-claims
5. https://www.theregister.com/2022/05/19/it_visa_discrimination/

The easiest way to become a millionaire or billionaire or possibly even a trillionaire is to steal wages. That's why the top jobs in IT, "real" Data Modeler, "real" Business Analyst, "real" Data Architect, are always targeted for theft and fraud. Taking out the people who secure data and replacing them with fake, often compromised resources then allows one to make more money using the aforementioned illegal means.

After they steal what they assume are just someone's wages, and If the real budget is not there and not adjusted for inflation, the only option for management is to orchestrate the creation of fake data models, and then scapegoat someone or a vendor when the project fails. The most common reason that IT projects fail is because someone stole the money that was supposed to be dedicated to creating the initial core data model from the initial project analysis phase. I have seen a single manager steal over 5 million dollars from one company by re-running the same IT project every year, stealing 95% of the data modeling budget each time. In a company that has billions in revenue, a few million dollars may go unnoticed for a while, but a 7+ billion dollar loss in profit due to fake IT software will eventually be noticed. Unfortunately, over 700 US banks noticed it too late and went bankrupt and out of business completely. The major issues are below for managers and staffing agencies who steal the data modeling budget.

1. They don't understand that it's not just for wages. It's for everything, including facilities, hardware, software, legal compliance, legal itself, and wages for a data modeling contractor. These days, just the legal compliance could cost over $1,000,000. Every industry has specific laws that require a legal specialist to work with the business analyst and data modeler. Without a budget and staff for compliance, even accounting and HIPAA-related insurance systems will be constructed in an illegal manner, triggering an average of $110,000,000 in annual fines from courts and governments. https://www.sec.gov/news/press-release/2022-114 https://www.sec.gov/news/press-release/2023-234 Once there is a data breach, even if the crooks are caught, usually less than 20 percent of the money is recovered. So, the crooks stay rich and often spend no time in prison.
2. Middle managers and permanent employees feel jealous because their wages are usually below a living wage and appear to be nowhere near the wages of a data modeler on the surface. However, the cost of a middle manager could average $50,000/year for 30 years, which equals $1,500,000 plus raises, bonuses, benefits, etc. For the life of the data model, the permanent employees cost more but are less capable.
3. They don't understand that a data modeler is only employed on a project for 3 to 36 months because once the model is made, it lasts for over 30 years without additional maintenance, but the data modeler is only paid one time. Therefore, this person is always a contractor with a minimum payment of $1,000,000, which is $2,500,000, adjusted for over 30 years of inflation. The hourly rate would be $1,250 minimum and could go over $10,000/hour for large complex military, bank, health, or ERP systems.
4. Most people are brainwashed by Microsoft and Apple to believe that all computer systems have to have monthly maintenance and updates, but computer systems are supposed to be stable for decades. The reason that updates often damage computer systems is that they are not supposed to be doing the update. The updating is a violation of international engineering standards. The practice was created by an Indian inside Microsoft and later adopted by Apple to wreck previously sold products, forcing customers to always buy a new copy of the same product. They should have waited to complete the product before rushing to market so that adding a new feature is data-driven which requires no change to the underlying code. Similarly, data models do not and cannot be changed after they are made, data is put into them, and applications are built based on them.
5. Changing a data model after applications are made based on the structure causes cascading changes across the entire company that cost billions in labor over just 15 to 30 years. One change to one item in a data model may require tens of thousands of changes in thousands of applications that may be on millions of computers, phones, etc. Most people do not understand how to make enterprise-level stable software. Additionally, operating system updates are not the same as the core data model which enforces security on corporate data in addition to the business rules. Data models are independent of user interface operating systems. If you grant a non-data modeler access to alter the data model itself, they will always cause chaos, security breaches, etc. Always! The data model is the FIRST LINE OF DEFENSE IN CYBER SECURITY. Over 99% of IT staff do not know that they should not EVER change the structure of a corporate or government data model after it is created, reviewed, performance-tested, and frozen. Worse, they also don't know why. So, they fake data models, leave everything insecure, and continuously change them, but charge money through accounting as if they are real and secure.

The One Big Table Trick for Fake Data Models

The most common type of fake data model is one that simply matches the output of a web page to the structure of a single database table. For each page, there may be a stand-alone table or one big table for multiple web pages, with complex SQL to give the appearance of meeting requirements. Complex SQL was rendered obsolete in 1967 with the invention of object-oriented programming, OOP, and object-oriented programming languages such as Java, C++, and Objective-C.

Fake data modelers and developers are pushing non-object-oriented Python and other legacy systems that predate the 1967 standards and the 1970s data standards for relational data along with complex SQL and one big table because it is easy and still in use in third-world countries. However, none of those obsolete technologies can meet modern standards and they were in use before billions of people had access to the Internet and handheld supercomputer phones. Note there is not a single hardware nor operating system vendor that supports Python as their primary programming language. Phone apps, operating systems, and desktop applications are written C++, Java, or Objective-C. However, old technology is still in use in third-world countries and universities that have not revised their curriculum in over 50 years to over 100+ years.

Those old systems did not need the same security that is required today because there was no way to connect to them remotely. One would have to break into a physical building to hack into computer systems if one could even find the building.

Effectively, 93% of companies are paying fake IT professionals to implement new obsolete systems with fake data models. The single big table strategy is over 202 years old and has been obsolete for 57 years.

If the web page has the following fields, the database table will have the same fields in the same output order using a procedural SQL statement or stored procedure. There will be no relationships to nor from this table.

1. Employee ID (PK - Primary Key)
2. Employee Name
3. Employee Salary
4. Date
5. Position ID
6. Job Title

The web page may look like the following. The fake data modeler would simply write an SQL statement, such as "select employeeid, employee_name, employee_salary, current_date, employee_job_id, employee_job_title from one_big_table;", then call it done, charge $100,000 for less than 10 lines of code, then disappear before anyone runs a performance test, production deployment, or before anyone tries to run a business intelligence report. The output would be limited to solely what is below. It would not be able to produce reports with different dates, history, graphs, charts, etc. Some fake data modelers / fake developers will produce a fake static web page that looks like a report in a meeting, but there is no actual dynamic functionality behind it. Those types of people are usually professional liars as well and they may say that something is "old" or an "old way of doing things" if someone asks why certain functionality is not present. They may say, "You don't need that," but it's just a lie. Some people are simply trained manipulators who know which buttons to push. Even if something is actually old, it may still be valid and mandatory for a business to function.

Employee ID Name Salary Date Position ID Title

001 John Smith 250,000 01/01/2024 004 CEO

988 John Smith 25,000 01/01/2024 489 Janitor

002 Jane Doe 100,000 01/01/2024 005 CFO

744 Jane Doe 20,000 01/01/2024 392 Data Analyst

Business Requirements (From a BRD, business requirements document)

1. Data must be filterable by first name, last name, hire date, and position ID
2. Report 1: Report for hire to retire that shows salary history, position history, and time with the company. One employee can hold multiple positions over a career.
3. Employees must be uniquely identifiable by personal data and employee ID.
4. If names change legally due to marriage or court order, show the current name, when the name was changed, and full work history

Why are both the single table field design and structural design bad and why is it fake?

1. If a data model does not meet nor consider business requirements it is fake. It does not meet requirement 1. If first name and last name are in the same field/database column, the data cannot be filtered by either first name nor last name. Additionally, if people have the same name, that name data would be duplicated by the number of times people have the same name. How people are named John? Additionally, one may get John the janitor instead of John the CEO when searching the data. If you search for all duplicate names from your data and see any actual duplicates, the model is fake. In modern atomic data, John or any other name would exist exactly one time in the data, and then be assigned using foreign key relationships accordingly avoiding duplication. Fake data models is why many companies have exploding data sizes in the petabyte size that would be in the gigabyte size if the data models were correct. Next, there is no hire date column, just the current date. Because multiple people can have the same name, the name must be made atomic in a first name dimension table and a last name dimension table, both connected to a fact table. Because one employee and multiple employees can be employed with the same job title at the same time or at different times, the relationship between employee and job ID is many-to-many. To resolve that relationship, there should be five tables. Two fact tables, one employee fact and one position/job fact, two dimension tables, and one bridge table between the employee ID dimension and the position ID dimension. One could add a hire date dimension to the employee fact, but consider requirement 2, and the design changes
2. Requirement 2 mandates a third fact table for hiring with a hire date dimension. A new bridge table between the hire date dimension and the employee ID dimension would need to be created. Then the design would change to bridge between the hire date dimension and the position for which the employee was hired.
3. The third requirement mandates more fields in more dimensions connected to the employee fact table such as date of birth, address, phone number, email, etc.
4. The fourth requirement mandates that about half of the data model change, separating names from the employee fact because one employee ID can be connected to the same employee with multiple names at different times. The employee fact must have the employee ID dimension, then a bridge table to a new Name ID dimension connected to a name fact, to which first name and last name dimensions are connected. It might be helpful to add a middle name dimension since there are over 50,000 people named John Smith or Jane Doe. One big table can only show one date and the current job. It cannot keep historical data without duplicating the employee ID and all the other fields. Using one big table always corrupts the data with untraceable duplication until data integrity fails. At that point, no one will be able to find the data they are seeking. Duplicating the employee ID would also mandate removing the primary key designation and unique index, which ensures that the data is unique and searchable with resistance to poorly designed applications that duplicate data or insert the wrong data. MANAGERS ARE NOT MEANT TO SEE THESE DIAGRAMS BECAUSE THEY ARE COMPLEX AND BEYOND THE UNDERSTANDING OF OVER 99% OF THE POPULATION. Because of the complexity, most people try to destroy the real model and go back to the one big table model, which has been outdated since the year 1822 mechanical clock-type computing, punch cards, etc. were replaced with microcomputing. Unfortunately, most universities are still teaching courses based on 1822 technology, punch cards, procedural programming, Fortran, Pascal, procedural Python, etc. They are simply running ancient technology from the 19th century on 21st-century hardware using compatibility applications such as Microsoft Excel. Excel also has little to no security, nor encryption, and the data is easily faked. It can only be accessed by one user at a time.

Mistakes

1. Skipping the analysis phase of the project and NOT creating any business requirement documentation or creating incomplete documentation. The project is over before it begins if this is done. You have ZERO chance of success if the analysis is not done. You are not special! You have ZERO chance of success if the analysis is not done! Everybody thinks they are special, but they are not in the analysis phase of IT.
2. Using ancient obsolete technology rebranded as new with new names. Technology is obsolete when a better technology is created, not just by the age of the technology.
3. Cutting costs from the previously successful cost reduction strategy, which is the entire IT division. Most people under 50 are too young to remember the costs of not having IT and running everything with manual labor, paper, faxes, etc. I have met IT workers under 30 who don't even know what faxes and filing cabinets are.
4. Cutting or embezzling pay and costs from the known successful IT labor roles and strategies such as data modeler, business analyst, data architect, data analyst, and data scientist using Agile, SDLC, or Six Sigma methodologies. Omit any of those roles, deviate from the methodologies, or try to combine any of those roles into just one person, and your project joins the 93% of IT projects that fail, especially if the analysis phase is skipped or conducted without a business analyst who creates the documentation required to employ a data modeler.
5. Embezzlement through staffing agencies, H1B scams, inadequate cheap foreign labor, civil rights violations, and internal fraud. Most people can look up their data modeling budget in their accounting documentation and see one million dollars, which is the standard amount from over 30 years ago, before Y2K, etc. First, the amount was not adjusted for inflation, and secondly, over 95% of it is stolen in kickback schemes between staffing agencies and middle management. The budget is required to create a facility just for data modeling, which includes between $500,000 to over $7,500,000 for hardware, software, facilities, utilities, and cooling for the data visualization and modeling system, plus the pay for the data modeler who knows how to use it. Management and staffing agencies will conspire to steal the biggest sections of the IT budget, then try to hire one person with a $600 laptop to create the data model for $70 / hour, which is less than 5% of the wages for a real data modeler. That means 95% of the data modeling budget is not only stolen but routinely stolen internally with willing accomplices and fake data already set up to fool accountants and upper-level management. Additionally, the internal staff and the staffing agency will try to support the low pay using fake survey data placed on Glassdoor for a data analyst which is falsely equated to a data modeler, knowing that neither the site nor upper levels of management know the difference between a data analyst and a data modeler. So, they get either a fake data modeler who will create something that will never work or a real data modeler who does not have the budget to create a data model that follows business requirements, which may be missing due to other missing parts of the analysis and design budget that were also stolen. The company will pay the money to the staffing agency. The agency will withdraw cash to pay the manager covertly to keep the contract. If the data modeler is real, he or she will be quickly attacked using baseless accusations and group lies, usually, all the people in the group are of the same ethnicity with zero diversity, which always causes fraud, then the real data modeler is replaced with a fake data modeler, usually an H1B foreign worker who will do whatever they are told even if it is wrong and illegal. They may also sell illegal access to the data, hack the system, leave it vulnerable to attack, etc. That is why so many computer systems are hacked every year. Over 50% of US citizens have had their personal data hacked. The EU has limited hacking due to some countries having citizen job protection. For example, it is harder to get a French or German cyber security professional fired so they can be replaced with a corrupt foreign worker desperate enough to break the law using the usual lies and tricks because all workers have government job protection. The EU governments require companies to prove no civil rights were violated before someone is terminated from work. The US has a slavery system in which terminated employees must prove in court that their rights were violated after they have been terminated using their own money which they no longer have due to the termination. US court cases can cost hundreds of thousands to millions of dollars and therefore almost no one has real civil rights. Only the extremely wealthy have rights.
6. Creating systems that are data-rich and information-poor (DRIP). Those systems are created to hide fraud by overloading management with meaningless data. Many foreign companies that use H1B workers are pushing "Data DRIP" as if it is a modern strategy because they know American companies have cut costs by placing non-technical managers in charge of IT. Non-technical managers are easily fooled and impressed by buzzwords that sound cool, although they don't understand the history behind them nor the meaning.

Solutions

1. Do not allow IT team managers to hire anyone themselves. The IT staff should only be hired by independent IT staff connected to HR with actual practical exercises proving skills and experience. The biggest problem with hiring is that HR is still designed solely for manufacturing, which employs slave-type manual labor. HR itself needs to be modernized and diversified ethnically with strict penalties for discrimination, which is usually severe within HR. I could write an entire article on modernizing HR for 21st-century IT.
2. Do not rely on education credentials for IT because most universities cannot afford to properly train IT professionals. Most university courses are for training people for manufacturing even if it is labeled as an IT profession. For example, it costs over 1 million dollars to train a new data modeler in all 30 sets of skills, which is now actually 56 sets of skills, up 26 skills due to advances in technology over 30 years. Over 75% of people who try to become a data modeler fail because they do not have the patience to train for over 5 years, and then work 2 to 5 more years beside a master data modeler before they are ready to work and get full pay. The path to becoming a real data modeler is equivalent to three Ph.D. courses, but you don't get the title of Doctor when it is complete. Universities cannot afford to risk losing 75% of 100 million dollars while training data modelers. Tuition cannot cover the costs. So, the universities say that corporations should do the training. The corporations say the universities should prepare students for real-world jobs. In the end, there are no new data modelers created. Currently, no one is training new data modelers, and 98% of those claiming to be data modelers are fake and usually from a country that did not have access to the original standards and training, which were classified inside the United States. Another example is the fact that Computer Science degrees are primarily designed for manufacturing computer hardware and do not cover the full 30 to 56+ sets of skills required to become a data modeler. However, most non-technical managers and HR staff assume it does without looking it up.
3. Do not rely on survey sites like Glassdoor for wage calculation because their data is just a bunch of fake surveys from unvalidated sources, mostly from India.
4. Do not try to pile multiple jobs or roles onto one person. Each role is separate and requires a separate resource. Using one person for multiple roles comes from manual labor slavery practices for manufacturing and has the opposite effect in IT or any intellectual labor role.
5. Use strict work/life balance for all teams. The quality of all intellectual labor goes down when resources are overworked.
6. Limit meetings and contact between staff for all code developers and data-related staff. They should have a single point of contact, which is the lead business analyst.
7. Hire at least one real senior business analyst with living wages. Make sure he or she can write real business requirement documentation and conceptual data models.
8. HIre subordinate business analysts to collect requirements. There should be one for each domain of business.
9. Pay high wages to people who make your core data systems to remove any incentive to create backdoors or ways to compromise your data system. If you don't pay fair wages, most people will start stealing. One cannot pay a contractor the same wage as a permanent employee and expect the same quality as if the real contractor-level wages were paid. Most data breaches begin with internal civil rights, labor, and wage issues.
10. Hire an encryption specialist with atomic data experience on staff permanently.
11. All IT managers should have at least 10 years of actual hands-on IT experience even if you must retain IT staff with business degrees. Managers without technical skills cannot manage IT teams. They are designed for manufacturing which is opposite to IT.
12. Mandatory diversity for every domain of business. If a company has staff in any section that is solely from one gender and ethnic group, there is a near 100% chance of fraud.
13. Search job descriptions for "complex SQL", "stored procedures," and other legacy technologies and stop them from being implemented. SQL is still used, but it is usually non-complex fast, and basic one-line SQL used to load an object inside an OOP language. Once the object is loaded with data, it runs in memory, making your data available for real-time reporting and proactive management. Complex SQL can only do overnight or multi-day batch processing. It is meant for only one user. Most modern systems have thousands to over a billion users per day. Stored procedures are only used to automate internal database administration tasks, not applications with more than one user.
14. Do not allow aggregate data which is easy to fake. All new systems should have real-time atomic data, which is the only way to make proactive business management possible; especially anti-fraud and business intelligence. If someone is stealing money from a company, they simply run the aggregate data batch process, then change the data after the process completes to hide the theft. There is no way to hide theft on a real-time data system. It will work even faster than bank systems.
15. Look for one person who is assigned more than one project role. Managers and staffing agencies routinely allocate funding for roles, assign multiple roles to one person, force them to work free overtime, and then steal 100% of the funding allocated to the additional roles assigned to one person. In accounting, everything looks fine, but it is actually slavery, in violation of labor laws and the 13th Amendment of the US Constitution. It also falls under human trafficking. Lastly, it guarantees that the project will fail and be refunded the following year when more money will be stolen.
16. Do not sign contracts with any staffing agencies that do not disclose the pay of its resources and how much is being paid to do the job versus how much they are putting into their own pockets. If they are hiding the pay, they are stealing most of it.
17. Do not sign non-performance guarantee agreements. If an agency cannot make IT systems that perform, they are fake and are just running a scam. You should know your concurrent user count by hour, by day, by week, by quarter, and by year. That must be used to create a minimum performance benchmark.
18. Pay for performance testing.

Summary

Because of INTERNAL labor fraud and embezzlement, most companies pay over $1,000,000 to over $30,000,000 for data modeling but get less than $5,000 in actual work and quality. They then expect to be able to use BI, KPI metrics, real-time data, reports, etc., but they get a fake data model that just outputs to a fake web page. If the data modeling fails, everything else will also fail. Below are the top seven tips to get it done right.

1. Follow the money. Make sure 100% of the budget actually goes to data modeling. This may require independent hiring outside the IT teams.
2. Check for one big table and tables that match web page output.
3. Check for duplicate names and other duplicate data.
4. Check if you can change the date ranges in reports.
5. Get professional performance testing.
6. Use technical managers.
7. Work/Life Balance and separation of duties for all.

There is no one big table data model that will satisfy even the most basic requirements because life, people, jobs, etc. are all too dynamic. The four requirements required a minimum of 19 tables to follow the business rules. A full set of requirements for an entire corporation could mandate over 10,000 tables, which is why a multi-million dollar budget is required for real enterprise data visualization and modeling systems, facilities, pay, etc. Data models are based solely on the documented business requirements and do not need maintenance for 30+ years if and only if the analysis was completed and all business requirements were captured in the documentation.

A con artist would say, "A data model doesn't have to be perfect," to get out of doing the work. Your answer should be, "Yes it does," because the definition of an imperfect data model is a data model that does not meet 100% of the documented business requirements. What will be sacrificed? Supply chain? Accounting? Row-level security perhaps? How about a major business domain component such as a sales platform and advertising optimization? Maybe an auxiliary component such as demographics could be omitted. That would then disable sales optimization by customer type, location, price by location, etc.

If any one component is missing, it causes cascading failures across the entire model. That's why the analysis and documentation is so important. Again, beware of the words, "We don't have time," it will guarantee a failed analysis and a failed project. If any employee does not have time to document the analysis, that employee must be replaced. Also beware of the words, "Change the data model," after it has been completed tested, and frozen. Whoever says that is likely a hacker.

Data models only change when the industry or the company has major changes in how they work, which usually takes over 100 years or longer. Retail storefronts of the 15th century are about the same today. The only major change has been electronic payment methods. In the example above the one big table fake data model omitted 98 percent of the requirements and functionality, however, one big table is the most commonly requested design by novices.

The most absurd request I get on data modeling jobs is, "Can you create a simple data model we can maintain ourselves without a data modeler?" Data models are based solely on business requirements, not the lack of skills for the permanent staff. It is absurd because people who are not data modelers can only think like a spreadsheet, which is one big table.