Twenty-four questions.

An hour with this list tells a platform owner more than a week of vendor calls. Every question has a yes or a no, and every no has a price attached to it.

Request the 48-hour audit Read the sprint

Nothing here needs tooling to answer. Walk the list with whoever holds the pager, mark every question honestly, and count the noes. Fewer than four is a healthy platform. More than ten means the next incident will be expensive.

Written for teams running production on cloud without a dedicated platform team. It is deliberately not a maturity model: there are no levels, no scores, and nothing to certify.

0/24

Answer the questions to see where the platform stands.

Recoverability

Can the platform come back without heroics?

Can the last deploy be reversed by one person, without a meeting?
Yes No

Every rollback becomes an incident, and the fear of rolling back slows every release that follows.
Has a restore from backup been performed in the last ninety days?
Yes No

An untested backup is a belief, not a control. The first real restore is the wrong time to discover the gap.
Is the environment described in code that can rebuild it from nothing?
Yes No

Recovery time becomes however long it takes to remember, and the memory belongs to whoever is on holiday.
Does anything in production exist only because someone clicked it once?
Yes No

That resource is invisible to review, absent from the plan, and the first thing to break silently.

Delivery

How does a change reach production?

Can a new engineer ship a one-line change on their first day?
Yes No

Onboarding cost compounds. If the path is undocumented, every hire pays the same tax.
Is the same artefact promoted through environments, rather than rebuilt per stage?
Yes No

The thing tested is not the thing shipped, so staging stops being evidence.
Does the pipeline fail the build on a failing test, or is the gate advisory?
Yes No

An advisory gate is a comment. Coverage decays without anyone deciding to let it.
Are deploys routine enough that nobody schedules them for a quiet evening?
Yes No

Batching changes to reduce risk raises it: bigger releases fail harder and are harder to attribute.

Visibility

Who finds out first when it breaks?

Does an alert reach a human before a customer does?
Yes No

Support becomes the monitoring system, and the first signal arrives already angry.
Can a request be followed across services without reading four dashboards?
Yes No

Diagnosis time grows with the number of services, which is the opposite of what the architecture promised.
Does every alert have an owner and a runbook, or do some just fire?
Yes No

Unactionable alerts train people to ignore actionable ones.
Is there a written record of the last three incidents and what changed after them?
Yes No

Without it the same incident is paid for repeatedly and each time it feels novel.

Cost

Does anyone know where the money goes?

Can this month's bill be split by team, service or environment?
Yes No

Nobody can act on a single number. Attribution is what makes cost a decision instead of a complaint.
Is there an alert on the invoice, not just on the infrastructure?
Yes No

Cost incidents are found at the end of the month, by which point they have already happened.
Do non-production environments switch off when nobody is using them?
Yes No

Idle staging is the most common line item nobody has ever defended out loud.
Has anything been right-sized in the last two quarters?
Yes No

Instances are provisioned for the worst day of the first month and never revisited.

Access

Who can do what, and who checked?

Is there a standing human account with production write access?
Yes No

One phished credential becomes total compromise, and the audit trail cannot tell you what happened.
Are secrets held in a manager, or in environment variables and a pinned message?
Yes No

Rotation becomes impossible, so it never happens, so exposure is permanent.
When someone leaves, is there one place that revokes their access?
Yes No

Access outlives employment, quietly, and nobody discovers it until an audit or a breach.
Does the pipeline hold credentials that a person could also use directly?
Yes No

The blast radius of automation becomes the blast radius of every laptop.

Continuity

What happens when the person who knows leaves?

Is there exactly one person who understands how the platform fits together?
Yes No

That is not a staffing risk, it is a single point of failure with a notice period.
Do runbooks exist for the tasks that only happen quarterly?
Yes No

Rare tasks are the ones nobody remembers, performed under pressure, usually at the worst time.
Are architectural decisions written down with the reason, not just the outcome?
Yes No

Every past decision gets relitigated, and the reasoning is reconstructed from guesses.
Could the team operate for a month without any external help?
Yes No

Dependency on a supplier is fine until the supplier is unavailable and the dependency was undeclared.

Counting the noes

0 to 3

A platform in good order. Fix the noes as ordinary work.
4 to 9

Normal for a team that grew faster than its infrastructure. There is a clear order to address these in, and most of it is a few weeks of work.
10 or more

The next incident will cost more than the remediation. Start with recoverability and access; everything else can wait behind those two.

The 48-hour audit takes this list, answers it against the running system rather than from memory, and returns the order to fix things in.

Where this leads

The engagement that closes the noes: four phases from first read to a monitored platform.
Most of recoverability and delivery lives here: cluster topology, the pipeline, the rollback path.

Free 48-hour infrastructure audit

The form takes three answers. What comes back within 48 hours is a written read of the platform: where the spend concentrates, what fails first, and the order to fix it in. No call is required to receive it.

What comes back

Where the cloud spend concentrates, by service and by cause.
The failure modes that would bite first under load or on call.
A remediation order: what to fix now, what can wait, what to leave alone.
A straight no when the fit is wrong, with a name from the wider network instead.

Turnaround: Within 48 hours
Format: Written findings
Charge: None
Next step: Only if asked for

Request the audit

Work email Required

Where the findings land.

Company Optional

Used to place the platform in context.

What is going wrong right now Required

Cost, reliability, release speed, on-call load. Two sentences is enough.

Role

Answers are used to reply to this request and nothing else. No list, no sequence. Privacy.

Request sent.

The request is on its way to CloudWarrior. Written findings follow within 48 hours of it landing.

No further step is needed.

Scripts are off, so this form cannot hand the request on. The same three answers work as plain mail: address, company, and what is going wrong right now. Write to pat@cloudwarrior.io.

Or write straight to pat@cloudwarrior.io.

Contact

Bench open · Remote across the EU

Email pat@cloudwarrior.io
GitHub github.com/cloudwarriorpat