Back to blog
Tech

When an Analytics Platform Costs Too Much: Building Your Own Views on the Data You Already Have

Nicolás Levy Nicolás Levy
|
Published Nov 18, 2026
|
Read Time 8 min
|
Share
Qualabs blog cover: When an Analytics Platform Costs Too Much: building your own views
Summary

Analytics platforms priced by volume cost more every time a media company's data grows. An in-house tool that builds views on top of the data the company already stores turns that recurring bill into an engineering decision it controls. At Qualabs we built one for a large US media group: a tool that creates analytical views over any data source.

Analytics bills grow with your data, not with the answers you get

The cost of an analytics platform usually follows volume: plays, sessions, events, or gigabytes ingested and scanned. For a media company, all of those grow together with the catalog, the audience and the number of systems that produce data.

That pricing works well while volume is small and the questions are standard. It becomes a problem when the company needs views that the platform does not offer, over data the platform does not hold. At that point the team pays twice: once to send the data to the platform, and again in engineering time to get the views it actually needs.

The questions that matter to a media company also change over time. A new device, a new workflow or a new partner brings a new question, and the answer often sits in data the company already stores: a database, the logs of its own services, the response of an internal API, files in cloud storage.

Building your own views pays off when the data is already yours

An in-house analytics tool is worth considering when three conditions hold: the platform price grows faster than the value it brings, the data already lives in systems the company controls, and the views the team needs are specific to its own operation.

In that situation, the tool does not need to collect or store anything new. It reads data where it already lives and turns it into views. The cost stops following volume and follows something the company decides: the infrastructure it already runs and the engineering time to build and maintain the tool.

Building is not free, and it is not always the right call. A platform brings collectors, device coverage and a team that maintains it. A tool of your own brings flexibility and a cost you control, and someone has to own it. The decision depends on which of those your company needs more.

A large US media group needed analytics without the platform price

Our team works inside the engineering organization of a large US media group. The company needed an analytics platform, and the option it was evaluating came with a cost it could not justify.

Our engineer on the team proposed an alternative: an in-house tool to build analytical views over any data source. The client's technical lead approved it, and our team built it.

The tool works with the sources the company already has: a database, logs, the response of an API, files in cloud storage, or a spreadsheet. An analyst describes the view in plain language, and the tool builds it at the speed of a prompt and saves it so anyone on the team can open it again.

Our team presented the tool to the client's engineering organization, and the engineers there showed strong interest in it.

How the cost compares when you build on data you already have

The savings come from where the cost grows. A platform priced by volume charges more for every new session, event or gigabyte. A tool that reads data already stored adds a view without adding a new meter: its cost follows the infrastructure the company already runs and the time of the team that maintains it.

Cost factorAnalytics platform priced by volumeIn-house views on data you already have
What drives the billPlays, sessions, events or gigabytes ingested and scannedEngineering time and the infrastructure already in place
When data growsThe bill grows with itCost follows the infrastructure and team the company decides on
A new viewLimited to what the platform supportsDescribed in plain language and saved
Where the data livesCopied into the provider's systemsRead where it already lives
Data sourcesThe ones the platform integratesAny registered source: databases, logs, APIs, cloud storage, spreadsheets
Upfront costLowEngineering time to build the tool
Who maintains itThe providerYour team or a partner

The table shows the trade-off without hiding the cost of building. A platform is cheaper at the start. A tool on your own data keeps its cost under the company's control as volume grows, as long as someone is in charge of maintaining it.

An AI model writes each view, and the platform controls every read

The tool is neither a traditional BI dashboard nor a chatbot. Business intelligence tools usually start from a dashboard that someone builds by hand, and a chatbot gives an answer that disappears with the conversation. Here, the analyst describes the view, an AI model writes the program behind it, the platform runs that program in an isolated environment, and the result is saved as a reusable view.

Each view is a graph of small steps. Steps that read data come first, shared steps compute each measure once, and each panel draws its result. Because every measure lives in a single shared step, two panels of the same view cannot contradict each other. When an analyst changes one part, only that part is rebuilt, and the rest reuses its results.

Control stays with the platform. Administrators register the data sources, and analysts choose only among the ones their permissions allow. The generated program cannot reach the network or hold credentials: every read goes through the platform. Running a saved view again executes the same program without calling the AI model.

This is how we approach Product Engineering at Qualabs: engineers who understand the client's problem, propose a solution and build it.

Five questions to answer before you buy or build analytics

  1. Map what drives the bill, whether it is plays, sessions, events, gigabytes or seats, and project it with your expected growth.
  2. Check where your data already lives, because data you already store is data you do not have to pay to move.
  3. List the views your team needs and the platform cannot give, since that gap is where the hidden engineering cost sits.
  4. Name the owner of an in-house tool before you build it, so maintenance is part of the plan.
  5. Keep control of the data in the platform, with registered sources and permissions, whether you buy or build.

Frequently asked questions

Our analytics bill keeps growing with our data. What can we do about it?

Look at what drives the bill. If the platform charges by plays, sessions, events or gigabytes, the cost will keep rising as your audience and catalog grow. At Qualabs we helped a large US media group change that equation: instead of paying to move data into a platform, we built an in-house tool that creates views over the data the company already stores, so the cost follows infrastructure and engineering decisions the company controls.

We don't have the team to build and maintain our own analytics tool. Is building still an option?

Yes. Building does not have to mean hiring a new team. Qualabs engineers work inside the client's engineering organization: they understand the problem, propose a solution, and build it. Ownership and maintenance are agreed from the start, so the tool keeps delivering value after the first version.

Is it safe to let an AI model write analytics code over our company data?

It is when security does not depend on the AI model. In the tool Qualabs built, the generated program runs in an isolated environment, cannot reach the network or hold credentials, and can only read sources that an administrator registered and the analyst is allowed to use. Every read goes through the platform.

How can we trust the numbers in an AI-generated view?

By making the logic consistent and reusable. In the tool Qualabs built, every measure is computed once in a shared step, so two panels of the same view cannot contradict each other. Once a view is saved, running it again executes the same program without calling the AI model, so the same question always gets the same logic.

Our data is spread across databases, logs, APIs and cloud storage. Do we need to centralize it first?

No. Copying everything into a new system adds cost and delay. The tool Qualabs built reads each registered source where it already lives, whether a database, service logs, an internal API, cloud storage or a spreadsheet, and saves the view, not a new copy of the data.

Our BI dashboards can't answer the new questions the business keeps asking. What's the alternative?

BI tools work best for well-known metrics that someone designed a dashboard for. When a new device, workflow or partner brings a new question, an analyst using the tool Qualabs built simply describes the view in plain language. The tool builds it at the speed of a prompt and saves it so anyone on the team can open it again.

Stay Updated

Get the latest on media tech trends, events, and insights delivered to your inbox.