More transfers than people expect, but none of it is the part that makes the job hard. Your data model knowledge, your grip on the order of execution, your instinct for what the business actually meant when they asked for a field: all of that carries over intact, and it takes a computer science graduate two years to acquire. What does not carry over is the discipline of writing code that still behaves when someone hands it two hundred records instead of one. Budget six to twelve months of deliberate practice alongside a real job, not a weekend course.
What actually transfers
- The data model. Relationships, roll-ups, sharing, record types. Apex is mostly a way of moving data between objects you already understand better than most developers do.
- Order of execution. If you have ever debugged a validation rule fighting a flow, you hold the mental model that catches out developers arriving from other platforms.
- Governor limits in outline. You have hit the flow element limit already. Same idea, harsher failure mode.
- Requirements sense. Knowing that “just add a picklist” usually means something else entirely. This is the scarcest skill on any delivery team and nobody teaches it.
- Debugging habits. Reproducing before fixing, reading a debug log without panicking.
What does not
Declarative work forgives a lot. The platform keeps you roughly inside the rails, and a flow that processes records one at a time will run for years before volume exposes it. Apex offers no such protection. Three things catch almost every admin who crosses over:
Bulkification. Not the concept, which takes ten minutes. The habit. Writing a query inside a loop feels natural because it reads like the sentence you would say out loud, and it will pass every test you write until a data load runs.
Tests as design, not as a tax. The 75% requirement teaches the wrong lesson. Coverage is a side effect. A test exists to assert that the thing does what you claimed, including when the input is empty, duplicated, or two hundred rows long.
Failure. Declarative tools show an error to a user. Apex running asynchronously fails silently unless you made it not. Knowing what happens to a partial failure in a batch is a genuinely different way of thinking.
The order to learn in
- SOQL properly. Relationship queries, aggregates, the difference between what the query returns and what the sharing model lets you see. You can start here today and it pays back immediately in reports and data work.
- Apex syntax and collections. Maps especially. Most real Apex is building a map, iterating a list, and updating in one DML statement.
- Triggers with a handler pattern. One trigger per object, logic in a class. Learn this before you learn anything clever.
- Tests, including a test data factory. Do this at the same time as triggers, not afterwards.
- Asynchronous Apex. Queueable first, then Batch, then Scheduled. Future is mostly legacy.
- Lightning Web Components. Later than people think. You need JavaScript fundamentals more than you need LWC syntax, and a lot of developer roles touch very little front end.
A realistic timeline
Three months to write Apex that works. Six to twelve months to write Apex that a senior developer would merge without rewriting. Two years before your instincts are reliable on a large org. Anyone selling a shorter version is selling a course.
The variable that moves it most is whether your current job lets you write code. An admin who ships one small Apex change a fortnight improves faster than someone doing tutorials every evening, because production feedback is the teacher. If your role has no path to that, the fastest move is often a sideways one into a team that will let you.
Build the right first thing
Pick an automation in your own org that a flow handles badly. A loop over child records, a callout, a nightly recalculation. Rewrite it in Apex with tests, then be able to explain why the new version is better and what it cost. That one piece of work is worth more in an interview than a certification, because it shows you chose.
Frequently asked
Do I need Platform Developer I first? No. Take it when you can already write the code, as a marker that you have finished the basics. Taking it to learn the basics produces someone who passed an exam and cannot debug.
Should I learn Java first? Not necessary. Apex is close enough to Java that the transfer works in the other direction, and learning a second language to prepare for the first one wastes months.
Should I stop doing admin work? No. The people who are hardest to replace do both, because they can tell you when the answer is a flow and mean it.
