When Can You Actually Estimate Software Work?

A system review gives you direction. It does not give you certainty.

The Question Everyone Asks

After a system review, the next question is almost always:

“Okay — so how long will it take to fix?”

It sounds like a reasonable question.

It’s also based on a misunderstanding of what a review actually provides.

What a Review Actually Gives You

A system review can tell you:

• where the problem is likely coming from
• what parts of the system are interacting incorrectly
• what the major constraints are
• what direction the solution will take

That’s valuable.

But it’s not the same as knowing how long it will take to fix.

The Gap No One Talks About

After a review, it feels like:

“We understand the problem — so we should be able to estimate it.”

But there’s a missing step.

Understanding where a problem originates is not the same as understanding:

• how far it extends
• how many systems it touches
• what breaks when you change it
• how stable the fix will be

Why Reviews Don’t Create Certainty

A review is a map.

It shows you where to go.

It does not show you what you’ll encounter when you get there.

So What Actually Creates Certainty?

Not more review.

Not more theorizing.

Work does.

The First Real Work Is What Matters

Once you start:

• running full flows
• observing real behavior
• testing under real conditions
• making changes and seeing the impact

You begin to answer the real questions:

• does the fix hold?
• what else breaks?
• how deep does this go?

That’s when estimates start becoming real.

What About Large “Investigation Phases”?

Some teams propose large upfront investigations:

• 20 hours
• 40 hours
• sometimes more

This can help organize the system.

But it still has a limit.

At some point, analysis stops improving certainty.

Only interaction with the system itself does.

The Real Progression

Software work doesn’t go:

Review → Estimate → Fix → Done

It actually goes:

Review → Direction → Initial Work → Feedback → Clarity → Continued Work

When Do You Actually Get a Real Estimate?

You start to get meaningful estimates when:

• the system behavior has been observed directly
• key flows have been stabilized
• major unknowns have been surfaced
• the problem stops expanding

In other words:

after you begin working through it

The Business Tension

Businesses want:

• predictability
• timelines
• cost control

That’s completely valid.

But software — especially existing systems — often begins in uncertainty.

The Real Solution

You don’t eliminate uncertainty upfront.

You manage it correctly.

That means:

• working in focused blocks
• making progress visible
• increasing clarity over time
• making decisions as the system reveals itself

What You’re Actually Buying

At this stage, you’re not buying:

• a fixed outcome
• a guaranteed timeline

You’re buying:

a faster path to understanding and stabilizing the system

Final Thought

A review tells you where the problem is.

Only working through the system tells you what it actually takes to fix it.

One-line version

You can’t estimate what you haven’t uncovered yet. You uncover it by working through the system.

Next Step

If this resonates, read how I work → How I Work