PRICING
There are three ways to get numbers you trust. Only one of them has a price tag.
You can hire and build it. You can buy tools and stitch them together. Or you can run the Data OS.
Here is what each one actually costs, including the parts that never show up on an invoice.
Every figure about somebody else's tool is sourced and dated at the bottom of this page.
The three options
Most teams are choosing between these without ever writing them down next to each other.
Build it in-house
Salary plus tooling
You have three or more data engineers already, and a genuinely unusual data problem.
- Hire the person, then wait out the ramp
- Pick and wire the warehouse, the loader, the transform layer, and the BI tool
- Model your CRM into something a report can read
- Own every API change, forever
- Total control over the result
- The knowledge lives in one person's head until it is written down
Buy more tools
Per seat, per source
One team needs one report off one system, and nobody else has to agree with it.
- Live in days, for the first source
- Someone else runs the infrastructure
- Each tool reports on the system it can see, and no further
- Definitions live in the tool, so they change when you switch tools
- The bill grows with people and with sources
- Your data lives in their platform
The Data OS
Scoped on the call
You are 20+ people, you have outgrown spreadsheets, and you should not be hiring a data team yet.
- Live in 2 to 4 weeks
- A BigQuery warehouse wired to your billing account, managed by us
- Every source connected, with no per connector fee
- Every teammate with access, with no per seat fee
- One metric glossary that every report and every AI answer reads first
- You keep the warehouse, the reports, and the definitions if we part ways
Ours is the only column without a number, and that is the honest answer rather than a sales tactic. What it costs depends on how many sources you run and how much of your tracking is broken. You leave the first call with a real number either way.
Side by side
| Build it in-houseSalary plus tooling | Buy more toolsPer seat, per source | The Data OSScoped on the call | |
|---|---|---|---|
| Time to a number you trust | Six months before anyone starts building | Days, for one source | 2 to 4 weeks |
| Who connects a new source | Your engineer | Whoever sells a connector for it | We do, included |
| Cost to add the twelfth source | Engineering time | Another line on the invoice | None |
| Cost to add the thirtieth viewer | Another BI seat, whatever yours costs | Another seat | None |
| When a platform changes its API | Your engineer, on a Friday | You file a ticket and wait | We fix it, usually before you notice |
| Who decides what a metric means | Whoever wrote the query | The tool's default | Written in your glossary, before the query runs |
| What your AI reads | Whatever you pipe into it | The vendor's chatbot, if it has one | Your governed tables, definitions first |
| Sources it can join across | Any, once you build it | Usually the one it plugs into | All of them, in one warehouse |
| Where the data lives | A warehouse you stand up | The vendor's platform | BigQuery, on your billing account |
| If the person who built it leaves | You inherit tables nobody understands | Nothing changes | Nothing changes |
| If the arrangement ends | You keep it, if anyone remembers how it works | You keep an export | You keep the warehouse, the reports, and the definitions |
What each one actually costs
Not a receipt, because yours will not match anybody else's. This is the shape of each bill, and the anchors are sourced at the bottom of the page.
Build it in-house
One data engineer averages $123,053 a year1, and spends roughly six months hiring and ramping before building anything. Underneath them you still need a warehouse, a loader, a transformation layer, and a BI tool. Two of those four will not quote you a price without a sales call.
Buy more tools
Reporting at the enterprise tier starts around $2,000 a month2, and the BI seats on top are priced per person, billed annually3. From there the bill grows on two axes at once: every source you add, and every person who wants to look at it.
The Data OS
One flat fee, scoped on the call. No per seat, no per connector, and no usage meter from us. Your only variable cost is the BigQuery you run on your own billing account, where the first tebibyte each month is free4.
Want it line by line, with every vendor's own figure and the date we checked it? That is what the comparison pages are for: building it in-house and your CRM's reporting.
Three of the tools on this page will not tell you what they cost.
Looker says call sales, on every edition5. Salesforce Data 360 quotes consumption credits per account6. Fivetran publishes a rate card with no plan price on it7.
That is not a scandal, it is how enterprise software is sold. It is worth noticing anyway, because "what will this cost us in year two" is a question you cannot answer from any of their websites, and it is the question that actually decides this.
The expensive part was never the software.
It is the six months of salary before anyone starts building, the meeting where four people bring four revenue figures, and the decision that waits a quarter because nobody could defend one of them.
None of that shows up on an invoice, which is exactly why it keeps getting paid.
Two ways this goes wrong that pricing pages never mention
I can do anything. Just let me know what you want to do, because I'm not in charge of the product.
This is the build option's real failure mode, and it is not a technical one. His team can build whatever the business asks for. Nobody in the building can tell him what to build, because the definitions were never written down. A build does not supply direction. It just gives the argument somewhere new to happen.
It kind of sucks at everything equally, but it does everything.
He runs four attribution and reporting tools and still logs into all four before a meeting. Each one is right about its own slice and blind to the rest. Buying a fifth adds a fifth number. What he is missing is not a tool, it is one place where the definitions live.
When the other two are the right call
We would rather tell you no than sell you a bad setup, so here is when we are honestly not the answer.
- Build it if you already have three or more data engineers, a data problem no platform is shaped for, and the budget to keep that team indefinitely.
- Build it if data infrastructure is the product you sell. Then it is not overhead, it is R&D.
- Buy a tool if one team needs one report off one system and nobody else has to agree with the number.
- Buy a tool if you are under about 20 people. The coordination problem we solve mostly has not started yet.
- Wait if nobody internally has asked the same question twice and gotten two answers. That argument is the signal, and until it shows up this is a solution looking for a problem.
Where these numbers came from
Vendor pricing pages and named salary aggregators, nothing secondhand. Where a vendor does not publish a price, we say that instead of guessing.
- Salary.com, Data Engineer salary, United States (states its own date: August 01, 2026) checked 2026-08-14
- HubSpot Data Hub pricing page checked 2026-08-14
- Tableau Cloud pricing page checked 2026-08-14
- Google Cloud, BigQuery pricing page checked 2026-08-14
- Google Cloud, Looker pricing page checked 2026-08-14
- Salesforce Data 360 pricing page (formerly Data Cloud) checked 2026-08-14
- Fivetran pricing page checked 2026-08-14
Questions
So what does the Data OS actually cost?
It depends on scope: how many sources you run, how broken the tracking is, and how much reporting your team needs. We scope before we quote, so there are no surprise invoices, and you leave the first call with a number.
What we can tell you now is the shape. It is a flat fee. No per-seat pricing and no usage meters from us. One-time setup and onboarding get scoped separately. And it lands under the cost of one data hire, who runs $120K+ before they have built anything.
Why won't you put your price on this page?
Because a price without a scope is a number you would have to renegotiate on the first call, and we would rather not start there.
The honest version: our pricing depends on how many sources you run and how much of your tracking is broken, and those vary more between two same-sized companies than you would expect. A page price would be either high enough to scare off the people we are cheapest for, or low enough to be a bait number. The figures on this page are the ones that actually help you decide, and they are about the alternatives, not us.
Where do the numbers on this page come from?
Vendor pricing pages and named salary aggregators, each one listed with a link and the date we checked it at the bottom of this page.
Where a vendor does not publish a price, we say so instead of estimating. Several of them are quote-only, and "quote only" is a real answer that tells you something useful about what the buying process looks like.
Is there a setup fee?
Usually. Onboarding and the initial build get scoped separately from the ongoing engagement, because the first 30 days are the heaviest.
If you would rather do it as a one-time project, we can build the system, hand you the keys, and stop there. That gets scoped as a project instead.
What happens to the cost when we add people?
Nothing. There is no per-seat pricing, and that is deliberate rather than generous.
The whole point of the system is that everyone in the company can answer their own questions instead of routing through one person. A per-seat price would charge you more for the exact behaviour we are trying to create.
What about the Google Cloud bill?
Your data lands in a BigQuery wired to your billing account. We carry the storage cost. You pay Google only for the queries you actually run.
That is why there is no usage meter on our side. We are not reselling you compute with a margin on top, so there is nothing for us to inflate.
Get the number that applies to you.
We map your sources, your tracking gaps, and where your reports lose trust. You leave with a number and a plan, even if it is a plan you run without us.