SECTION 01The short answer
A working in-house data team is about seven people and runs roughly $1,107,000 a year to $1,572,000 a year once everyone is on payroll. That is $92,250 a month.
Build the same seven seats from Bureau of Labor Statistics median wages and load them with benefits at the federal ratio and you land at $1,196,571, which is $99,714 a month.[1]
- $667,000 a month. A billion dollar company’s team, 40 to 80 people, $8,000,000 a year.
- $92,250 a month. A full team, in house, 7 people, $1,107,000 a year.
- $25,000 a month. One senior data scientist, 1 person, $300,000 a year.
- $5,000 a month. Your Kixik data team, every four weeks, at the entry band.
Two different routes to the same place, which is the useful thing about this question: the answer is not controversial once the benefits multiplier is in it. Almost every figure you will find online leaves that multiplier out, and is therefore about 30 percent too low.
SECTION 02The seven seats
National medians, May 2025, with the federal occupation code beside each so anyone can check them.[1]
| Seat | SOC code | Median wage | Cost to employer |
|---|---|---|---|
| Analytics lead | 11-3021 | $175,140 | $250,200 |
| Data engineer | 15-1243 | $139,500 | $199,286 |
| Data scientist | 15-2051 | $120,230 | $171,757 |
| Data scientist | 15-2051 | $120,230 | $171,757 |
| Analyst | 15-2031 | $88,940 | $127,057 |
| Analyst | 15-2031 | $88,940 | $127,057 |
| Database administrator | 15-1242 | $104,620 | $149,457 |
| Seven people | $837,600 | $1,196,571 |
A few things about that table are worth saying out loud.
These are medians, so half of the people in each occupation earn more. They are also national. In San Francisco, Seattle, New York or Boston every row goes up, often by a third or more.
And the team is not seven senior people. It is one lead, one engineer, two scientists, two analysts and somebody keeping the database alive. Loaded, that averages $170,939 a seat, which is below what a single senior data scientist costs. A team is cheaper per head than a star and does far more.
SECTION 03The multiplier everyone forgets
A salary is not what an employee costs. It is about 70 percent of what an employee costs.
The Bureau of Labor Statistics measures this directly and publishes it quarterly. For private industry workers in June 2026, total employer compensation averaged $46.89 an hour. Wages and salaries were $32.82 of that, or 70.0 percent. Benefits were the other 30.0 percent.[2]
So the multiplier is one divided by 0.700, which is 1.43.
| If the salary is | It costs the employer about |
|---|---|
| $88,940 | $127,057 |
| $120,230 | $171,757 |
| $175,140 | $250,200 |
That 30 percent is not padding. It is the employer half of payroll taxes, health insurance, retirement contributions, paid leave, workers compensation and unemployment insurance. It is money that leaves the business every month and never appears in the conversation about what to offer a candidate.
If anything, 1.43 is conservative here. The same release puts the wage share for full-time private industry workers at 68.5 percent, which is a multiplier of 1.46, and a data team is full-time.[2]
SECTION 04What the number still leaves out
Even at $1,196,571 the figure is the floor, because it is payroll and nothing else. Still to come:
- Recruiting. Agency fees run 20 to 25 percent of first-year salary, and seven hires is a year of somebody's attention even without them.
- Tools. A warehouse, the pipes to fill it, and the licences to query it. Real money and it grows with the data.
- Management. Seven people need somebody above them whose actual job is deciding what the team works on. In a small business that is the owner, and it is not a small tax on their week.
- The ramp. A new data team does not produce anything useful for months. They are learning the business, finding where the data lives, and discovering that three systems disagree about what a customer is.
- Turnover. These roles move. Every departure takes the context with it and restarts the ramp for that seat.
None of that is an argument against hiring. It is the difference between the offer letter and the cost.
SECTION 05The cheaper versions, and what each buys
Almost nobody doing $2 million to $50 million a year is choosing between seven people and nothing. They are choosing between four real options, and the useful question is what each one actually covers.
| Option | Roughly | What it covers, and what it does not |
|---|---|---|
| One senior data scientist | $300,000 a year | Real analysis, one person's worth. Gets data out of systems badly, because that is engineering and not their trade. No cover for holidays or resignation. |
| A business intelligence contractor | project work | Builds a specific thing well, then leaves. Excellent for a one-off. The knowledge goes with them and the next question starts a new engagement. |
| A fractional finance chief | part time | Strong on the money: cash, margin, forecasting, structure. Not their job to go into transaction-level operating data and find where it is leaking. |
| A bookkeeper doing more | modest | Knows the accounts better than anyone. Almost always working in the accounting records rather than the systems where the operating answers live. |
Those are four different jobs that get discussed as though they were four prices for the same job. The most expensive mistake is hiring one of them to do another one's work.
SECTION 06Why this was never available below a certain size
Put the arithmetic next to a real business. A company doing $8 million a year at a 10 percent net margin makes $800,000. The in-house team costs more than that.
It was never that smaller businesses did not need this. The capability had one price and one shape, and the shape did not fit anything under a certain size.
So the work did not get done smaller. It got skipped. Forty years of businesses running on a very good gut and a very good bookkeeper, which is exactly the correct decision when the alternative is a seven-figure hire.
SECTION 07What we charge
We are a data team small businesses hire like staff, and we run $5,000 to $11,000 every four weeks, priced to the size of the business. At the entry band that is $60,000 a year.
The in-house build above is 19.9 times that. The analytics lead alone, one person, loaded, is $250,200, which is 4.2 times the entire service.
We are not going to claim seven people's output for one price, because that would be nonsense. What we will say is what the shape actually is: shared rather than dedicated, so the fixed cost of a data team gets spread across the businesses using it instead of being carried by one. That is the whole trick, and it is the same trick that made an accountant affordable to a corner shop.
The fit, plainly
We work with owner-operated businesses doing $2 million to $50 million a year. Cancel any month with one email and keep everything we made.
Before anyone pays us anything, we will put together a short brief on your business from what is public, so you can see how we think first. Write to data@kixik.tech, or read what we actually do.