Most IT Certifications Are Just Recruiter Bait

I have 14 Microsoft certifications. The honest answer to “why” is not that I needed the knowledge, or that I wanted to prove my competence, or that I love studying. It is that I realized, somewhere around certification number five, that certifications are not what the industry pretends they are, and that once you accept that, they become a lot more useful.

Here is the claim this post is built on. Certifications are not proof of competence. They are not efficient learning tools. They are not a signal that you know the material. They are a ticket past a broken filter, and the broken filter is the only reason they exist in the form they do. If you treat them as anything else, you are either paying too much attention to them or too little, and both mistakes are expensive.

That sounds cynical, but I think it is actually the honest frame. And once you hold certifications up to it, a bunch of confusing behavior in the industry starts making sense, including my own. So let me walk through what the filter is, why the certifications that pass through it are not the ones you would expect, and how you should actually treat them if you want to stop wasting weekends on the wrong ones.

The Filter Is the Whole Point

Start with the mechanism. A job posting goes up. It lists requirements, some of which are real and some of which are not. Somewhere in the requirements there is a line that says “AZ-305 required” or “AWS Solutions Architect Professional preferred”. That line was probably copied from a template, or added by a recruiter who was told to match another job posting in the same category. It did not come from a conversation between the hiring manager and the team about what the candidate actually needs to know. The hiring manager might not even be aware the line is there.

Then an applicant tracking system scans resumes against that line. Candidates who do not match the keyword are filtered out before a human ever sees their profile. The filter does not ask whether the candidate can do the job. It asks whether the candidate has the string “AZ-305” somewhere on their resume. A candidate with ten years of Azure architecture experience and no AZ-305 gets filtered out. A candidate who passed AZ-305 last week and has never touched Azure in production goes through.

That is the filter. It is simple, it is dumb, and it is everywhere. If you have ever wondered why job postings demand certifications that nobody on the actual team has, this is the mechanism. The certifications are not a signal to the team. They are a signal to the filter. The team does not even see the postings until after the filter has done its job.

Once you understand this, certifications stop being confusing. They are not a credential in the traditional sense, where a trusted authority vouches that you know something. They are a cheat code for a pattern-matching system that nobody designed on purpose and that nobody really defends in public, but that runs in almost every hiring pipeline in the industry. The cheat code works. That is the point. That is the only point.

Which Certifications Are Worth Anything

If certifications are cheat codes for a filter, the question is not which ones teach you the most. The question is which ones give you the most filter-clearing power for the time you put in. That is a different question, and it has a different answer.

The certifications that are worth taking are the ones that appear in job postings you actually want. For Azure engineers in 2026, that is basically AZ-104, AZ-305, SC-100, and one or two of the role-specialization ones like AZ-500 for security or AI-102 for AI engineering. These are the strings the filter is looking for, because these are the strings the recruiters copied from each other when they wrote the postings. Getting those certifications opens doors. Not because the certification proves you can do the job, but because the filter stops blocking you when you have them.

The certifications that are not worth taking are the ones nobody asks for. I have a PL-300, the Power BI Data Analyst certification. I passed it in 2023, have never built a production Power BI dashboard, and remember almost nothing about it. If you hired me today to build Power BI reports, I would be googling “what is a measure” on day one. That certification gave me a badge and zero filter-clearing power, because nobody is filtering for PL-300 in the jobs I want. It was a pure waste of two weekends, and I only took it because I was in a “collect them all” phase that I have since abandoned.

The subtler category is the certifications that are worth it for the work, not the filter. My SC-100 falls in this bucket. When I took it I was already doing security work daily: compliance baselines, Defender for Cloud, Sentinel. The study material did not teach me the job. The job had already taught me the job. What the study material did was organize the knowledge I already had, hit the corners I had been quietly skipping, and give me a vocabulary that helped me explain what I was doing to auditors and management. The cert was a side effect. The forcing function of having a deadline and a scope was the real value.

So there are really two kinds of certifications that make sense to pursue. Ones that clear the filter for jobs you want, and ones that force you to organize knowledge you have already picked up in production. Everything else is a LinkedIn decoration that will leak out of your head inside six months.

The Order Matters More Than Anything

The order in which you take a certification relative to the hands-on work completely changes its value. That took me years to notice, and it is the single biggest lever I have found on whether any given cert is worth the money.

A certification you take before you can do the work is a waste. You memorize concepts you have no hooks for, pass the exam by pattern-matching the question format, and forget everything within months because you never apply any of it. My PL-300 is exhibit A. But there are thousands of people with AZ-104 who cannot actually administer an Azure subscription, and tens of thousands with AWS Associate certs who cannot actually deploy anything to AWS. They did the cert first and the work never happened, and now the cert is sitting on their LinkedIn signaling competence they never had.

A certification you take after doing the work is a completely different product. You go into it already knowing the material in operational terms. The exam forces you to map what you know onto the vocabulary the certification uses. You discover two or three corners you had been quietly avoiding. You walk out with a cleaner mental model of the thing you already knew, plus a badge that clears the filter. Same exam, same price, completely different outcome.

Almost nobody frames it this way, because the people selling certifications want you to buy before you have done the work, not after. The whole course-and-exam industry depends on the assumption that certification is a path to competence. It is not. Competence is the path to certification. The industry has it backwards, and the reason it has it backwards is that backwards is more profitable.

Once you see this, the rule for yourself becomes simple. Do not buy a certification you cannot already do the work for. If you cannot sit down in a live environment and configure the thing the certification covers, you are not ready. Go do the work first. Read the docs, deploy a few test resources, break something, fix it, and come back when you have the scars. Then take the exam and use it to organize what you learned. The exam is cheap at that point. It is expensive as a learning tool.

What to Do About Other People’s Certifications

The same frame changes how you should read other people’s certifications, including when you are interviewing them. A wall of badges on a LinkedIn profile is not evidence of competence. It is evidence that the person understood the filter and played it. That is not nothing. It tells you they can read the rules of a system they did not design and act rationally inside it, which is approximately forty percent of what any IT job actually requires. But it does not tell you they can do the work.

When I interview candidates, I do not read certifications as a competence signal. I read them as a sorting signal. A candidate with zero certifications and ten years of visible production work is more interesting to me than a candidate with eight certifications and a thin resume. The first candidate did the work without needing the badges, which suggests they understood the filter was broken and routed around it. The second candidate optimized for the filter, which is rational but does not tell me anything about their ability to ship.

The actual competence questions have nothing to do with certifications. They have to do with what the person built, what broke, how they fixed it, and whether they can explain a decision they made and why it was a trade-off rather than a best practice. Those are the questions the filter cannot screen for, which is exactly why they are the ones worth asking. A portfolio of real work, a git history with decisions in it, a story about a production incident told with specific technical detail, any of those are worth more than a wall of certifications in the interview, because they are the things the filter could not have handed you.

This is why I have written before about how the AI hiring tools make the same broken filter worse. The filter was already broken before AI got involved. AI just scales the brokenness by applying it to more candidates faster, with less human override, which means career switchers and self-taught engineers get filtered out in bulk before anyone can notice the filter was looking for the wrong thing. Certifications are the workaround for engineers who got filtered out. That is an honest frame for what they are: not a credential, not a learning tool, but a workaround for a system nobody fixed.

The Honest Recommendation

Let me bring this together into the single thing I want you to take from this post. Certifications are a ticket past a broken filter. Treat them that way. Stop studying for them to learn things, because the learning does not stick unless you are already doing the work. Stop avoiding them on principle, because the filter is real and avoiding it costs you interviews. Start taking them strategically, which means: do the work first, identify which certifications the filter is asking for in the jobs you want, take those certifications as receipts for work you have already done, and ignore the rest.

I will probably get my fifteenth certification next month. That is not inconsistency. My argument was never that certifications are bad. It was that the conventional framing is wrong, and the honest framing turns them into a tool with a specific job. I use the tool for its specific job, and I do not buy more of it than the job requires. Less glamorous than “certifications prove skill”, less cynical than “certifications are useless”, and more accurate than both.

The cert is not the learning. The work is the learning. The cert is the receipt for work you already did, priced so you can hand it to a recruiter who will not read anything else. Stop buying the receipt before the work, and stop apologizing for buying it after.