The God of Fire: Choosing What to Watch

The context

I’m thinking about first days and fresh starts.
Joining a new company. A team reshuffle. An acquisition. A new exec with a new agenda.
The ground shifts and the fires multiply.

In FinOps we have more data than ever and better tools than ever, yet the first problem is still attention. Where do I look? Which fire matters now?

The god of fire

The image came from Forge of Darkness. There’s a scene where a becoming-goddess of fire can step into any flame and listen to the conversations around it. It feels like an incredible power—until you realise how many fires there are. In a medieval world, it’s millions. You could listen to a thousand and still miss almost everything.

That’s our life with dashboards. We can drop into any metric, any bill, any trace. We can be the fly on the wall—only the fire is warmer than the wall and the voices are louder. But even if you can appear in any flame, you can’t stand in all of them. You must choose.

There’s another layer to the metaphor: if you arrive mid-conversation, in a language you barely know, you miss the unsaid. You don’t know the history, the grudges, the private jokes. Data without context is that—loud, detailed, and oddly empty. It’s Bruce Almighty drowning in prayers until someone tells him those are only for one street corner. Scope is mercy.

Power without focus is noise

FinOps floods you with access. Spend, usage, tags, commitments, anomalies, recommendations. It’s exciting right up to the moment it’s exhausting. You go to a conference and return convinced that unit economics is the only truth—until the next talk makes it forecasts, and the next talk makes it re-pricing everything to on-demand equivalents. Shiny fires everywhere.

The trap is thinking access equals progress. It doesn’t. Access asks for focus. The more fires you can enter, the more deliberate you have to be about which one you choose.

Coblan philosophy: context is everything

Tools always make choices for you, even when they feel neutral. Take dashboards I love—Kudos, for example. They surface useful things and nudge you towards actions, like “move to Graviton”. That’s often right, but it’s still a baked-in assumption. Behind every “do this next” sits a product decision, usually shaped by the biggest customers or the loudest use cases. Your context may not match theirs.

So before you accept a tool’s ranking, pause. What is your outcome? What is your constraint? Who are your people?

Stephen Covey’s circles help here: your circle of influence versus your circle of concern. Listen at the fires where you know the actors, where your voice can change the conversation, where you can actually act. If you start by scanning every fire you don’t understand, you will get lost quickly.

Starting somewhere new

When you land in a new place, the temptation is to read everything. Don’t.
Start with what you already know—old friends, old skills. Use those to gain traction, earn trust, and learn the local language. FinOps data is only a small slice at the beginning. The bigger job is understanding the company, the colleagues, the boss, the unwritten rules. Then pick the fire you can help with today.

And remember: “new” happens more than once. A new C-suite, a merger, a pivot—each one resets the map. You go back to choosing again.

Who decides what matters?

We often put this on the practitioner: decide whether savings, risk, or learning is the priority.
In reality that decision belongs to leadership. The practitioner should contribute evidence and options, then align to an explicit choice.

We talk about executive sponsorship, but we need executive decisions: a clear, communicated focus that everyone can gather around.

Once that exists, the work becomes human again. Which meetings will you be in? Finance, engineering, strategy? Which conversations are your fires? Which data will you follow so your presence adds value? When you don’t know the people around a fire, arrive quietly and listen first.

The cost of losing focus

You know focus is slipping when you start throwing dashboards at people. Lists of actions, CSVs, charts, and no change. You’re asking teams to care about things they’re not incentivised to do, at times that don’t fit their cycle, using language that doesn’t fit their world. That’s not a data problem; it’s a context problem.

If the data you’re staring at isn’t answering the question you have, stop. Either change the question or build the tool that fits the question. Sometimes that tool is a simple worksheet; better if it’s something repeatable. But the key is to design for your context, not someone else’s default.

A practical way to choose your fire

1. Name the outcome. Savings, risk reduction, learning, velocity—pick one for this cycle.
2. Get the decision. Ask leadership to confirm the priority. If you are leadership, decide and communicate it.
3. Map the people. Who wins, who pays, who decides? Which fires will they be sitting around?
4. Enter quietly. Listen first. Gather the unsaid. Learn the language.
5. Act once, measure once. Make one meaningful change. Measure. Share the evidence.
6. Re-choose when the world changes. New exec, new strategy, new constraints—reset your focus.

When tools choose for you

It’s not that tool-led priorities are wrong—they’re just not universal.
“Biggest savings first” is perfect when runway is short.
“Highest risk first” is correct when exposure outweighs pounds and pence.
Just don’t let the default become your purpose by accident.

Ask of every recommendation:

• What assumption is hidden here?
• Whose context shaped that assumption?
• Does it match ours today?

Workshops, not waterfalls

When you’ve drifted, run a context workshop. Bring finance, product, engineering, and security into the same room. Share the strategy. Write down constraints. Name the outcome. From there, build a context exchange: a simple way to express what matters right now so your tools and reports reflect reality.

Often you already have the pieces—small scripts, checks, or datasets from your FinOps toolkit. Combine them differently and you’ve got a fit-for-purpose instrument.

Serving humans

Being godlike in data doesn’t remove the fact that we serve humans—sometimes person by person, sometimes the legal fiction we call a company.
Treat the company as human enough to have needs, moods, and scars.

Optimisation time should be recognised in the work (story points count).
Incentives should line up with the outcomes you’re asking for (cost, risk, carbon—pick and reward).
Context has to be shared, not assumed.

The closing reflection — choose, pause, and talk

So here’s what I want you to do after reading this.

Pause.
Really pause.

Step away from the dashboards, the scripts, the forecasts, the tickets.
Open a page — physical or digital — and write down a list of decisions you need answers to.
Not data you need. Not metrics you could calculate.
Decisions.

Things like:

• “Should we keep that service?”
• “Is saving on this cost more important than improving reliability?”
• “Who decides our target for carbon reporting?”

Once you have that list, go and talk.
Talk to the people who can give you context — finance, engineering, product, the execs.
Ask how they see it, what matters most to them, what their priorities really are.

This is the real work of FinOps: to create shared understanding before shared dashboards.

Then, once you have those conversations, go back to your data — step into the fire again, like a god — and look with new eyes.
Now you know what you’re searching for.
Now you know which flame matters.

That’s how FinOps becomes human again.
It’s not just about the data; it’s about what the data means to the people it touches.
It’s about building the little community of understanding that lets tools and humans align around the same fire.

A note from the Malazan world

In Forge of Darkness, and throughout the Malazan Book of the Fallen, gods are everywhere.
They’re powerful, flawed, jealous, brilliant — and they’re always at war.

That’s what data feels like sometimes: divine, immense, and rarely in agreement with itself. Each dataset believes it’s the truth. Each metric fights for dominance.

But what I’ve always loved in those books is this: gods were once human. They became ascendants by doing something extraordinary. They carry that humanity — the pettiness, the grace, the empathy, the hunger — into their divinity.
And they can still die. Only mortals can kill a god.

That line stays with me.
Because it means that power doesn’t make you infallible — it just gives you reach.
And in FinOps, that’s us. We’re ascendants of data: once human, still human, now able to see almost everything.

But the moment we stop listening, stop connecting, stop caring about the fires where people actually sit, we lose the reason we rose in the first place.

So pause.
Talk.
Choose your fire.
And remember: even gods need to listen to mortals.

Inspired by Steven Erikson’s Malazan Book of the Fallen, especially Forge of Darkness.
Guided by the Coblan belief that context is everything.