2027 is coming up fast, and I keep getting the same question from people: what do I actually need to know, and what should I be doing about it?
Most of what you will read about 2027 was written for a hiring boom. The Australian and New Zealand numbers do not show one. So this is the list I would give a friend over coffee. Five things that are actually changing in how you get hired, why each one matters to you, and exactly what to do. Every number has a source at the bottom. Nothing here needs a tool to start.
Let me begin with something most people never hear.
When you apply to a company, that application sits in their system. Not just for that role. For every role they open after it. When a recruiter pulls up your name, the system shows how many times you have applied there, for what, and what happened each time. Someone who has fired off six applications for six different roles in a year is telling that recruiter something before they have read a word of the CV.
I have watched this for 13 years and it still surprises people. A weak application is not free. It costs you the next role at that company too. Keep that in mind for everything below.
A recruiter puts up a role on Monday. Let's say principal level. By Friday there are 300, sometimes 400 applications in the system. From what I have seen over the years, one or two in a hundred actually match what the ad asked for. Call it ten.
The recruiter opens those ten. Books a few interviews. Six or seven people in the process fills the role, and when you have 20 roles on the go, you do not go looking further than that.
The other 290 just sit there, and nobody rejects them or reads them. A few weeks later an automated email says the role has been filled, and all 290 people read that as "my CV was not good enough." It was not a rejection. It was a queue. I call the line that matters the first ten. Above it, a person reads you. Below it, nothing happens.
And one sentence on the market, because it is the reason the queue keeps getting longer: there are fewer roles than last year and a lot more people going for each one. SEEK's July report has Australian job ads down 6% on last year and applications per ad at the highest level they have ever recorded. That is the market you are walking into.
Here is what a recruiter is actually deciding when they open your application. Not "is this person good." It is "can I put this person in front of the hiring manager and defend it in ten seconds." Your job is to write that defence for them. Everything that follows is how.
This is the rule I give every client.
Open the ad. Find the three things it asks for most. They are usually in the first five bullets, and one of them is usually repeated. Write those three down.
Open your CV and look only at your last two roles. For each of the three, find the line a recruiter would read out loud to the hiring manager to justify picking you. It needs the thing itself, a number, and what changed. If that line does not exist yet, write it now. If you cannot write it honestly, that is your answer.
Score it. Three out of three, apply today. Two out of three, apply only if the third is close, and fix the line first. One or zero, do not apply. I know that hurts. But that application goes into the 290, it costs you an evening, and it sits in their system for the next one.
Timing. Apply within two days of the ad going up. A role that has been up for two weeks already has its first ten. If it says "reposted," it has been refreshed, not reopened. The pile is already there.
Run that on the next five ads you look at. My bet is you apply to two instead of five, and both land where a person actually reads them.
How you know it worked: two applications give you a reply rate you can read. Twenty give you a fog.
Here is who actually decides the first ten, because it is not the recruiter. Not at first.
Say the recruiter uses an ATS called Greenhouse, one of the common ones. I walk clients through this on screen because it changes how they apply. Before the recruiter has read a single CV, the system has gone through all 300 and lined them up. Next to each person is the list of requirements from the ad, and against each one it says met, partly met, or missing. Next to "met" it quotes the exact sentence from your CV that it used as proof.
So it is not counting keywords. It reads your sentence, works out what you did, and checks that against what the hiring manager asked for. "Led a cross-functional team of 12 across product, data and legal to ship the new onboarding flow in Q3" gets quoted and marked met. "Strong cross-functional leadership skills" comes back partly met or missing, because there is nothing to quote.
It did not reject anyone. It said "talk to these ten first," and the recruiter does, because they have 20 roles open and that list is the best information they have. LinkedIn does the same thing when you apply through them. Their help pages call it "implicit skills": they read your headline, summary and job descriptions, not just the skills box.
You might have read that keywords are dead and you should just write about your experience. Half right. The words in the ad still matter, because the system is checking your CV against that exact ad. But the word on its own is worth nothing. It has to sit inside a sentence that proves it.
Here is how to get into the first ten. About 20 minutes per role.
Take the three requirements from the relevancy test. Add the next two most repeated ones. Five total. These are the rows the system scores you on.
For each of the five, write one sentence that uses the ad's own word for the thing, then shows you doing it, with a number and what changed. The ad says "stakeholder management"? Your line says who the stakeholders were, how many, and what you got them to agree to. The ad says "Kubernetes"? Your line says what ran on it, at what scale, and what broke less afterwards.
Put all five lines in the top half of page one. Under your most recent role, or in a short summary under your name. The system reads the whole document. The recruiter does not. If they have to scroll to find their defence, they move to the next person.
Do the same to your LinkedIn: headline and the first two lines of your About, same five things, same words.
Test it. Hand your CV and the ad to someone who does not know your work. Give them 30 seconds to point to where you cover each of the five. Every one they cannot find, the system marks missing and the recruiter never sees.
How you know it worked: within a week of the LinkedIn edit, your "search appearances" number in profile analytics moves. That is the software finding you.
The advice everywhere is: use AI to apply to more jobs, faster. Here is what it looks like from the recruiter's side of the desk.
A big share of the 300 are now CVs generated in one go by a free chatbot, from a prompt like "rewrite my CV for this job." Any recruiter can tell within two lines. Not because there is a detector. Because they all read the same. Same verbs, same shape of bullet, same summary that could belong to anyone. When 200 CVs read like one person wrote them, nobody can tell them apart, and a CV nobody can tell apart never makes the first ten. Nobody built a machine to catch you. They built one that cannot tell you from the other 289, which is worse.
The employers are moving on this too. SmartRecruiters has shown a prototype tool that flags suspicious application patterns from how and where applications come in. Greenhouse's CEO said on a podcast in August that as candidates use AI to apply in bulk, employers use AI to screen faster, and both sides end up worse off.
AI is genuinely good at writing CV bullets. The problem is not the tool, it is the brief. Think of it like a barrister. A good barrister can argue almost anything, but only from a brief. Hand them the facts, the numbers, what was at stake, what you decided and what changed, and they build a case a jury believes. Hand them just the charge sheet and say "win this," and you get a generic speech that could be about anyone. "Here is the job ad, rewrite my CV" is the charge sheet with no brief. So is "rewrite this so I hit 99% on the ATS score." The AI has nothing to argue from, so it reaches for the same words everyone else gets.
Brief it before it writes. For each bullet, tell it what kind of line this needs to be (a scale line, a result line and a what-I-changed line are different jobs) and give it the raw material: the numbers, the size of the team or system, the problem you walked into, what you actually did, what was different afterwards. Messy notes are fine.
Then let it write, and check every line against one question: could a recruiter read this out loud to the hiring manager and defend picking you? No number, no scope, nothing changed means the brief was thin. Add to the brief. Do not let it pad.
Let it read as well as write. Paste the ad and your CV in and ask: "List the five things this ad asks for most. For each one, quote the line in my CV that proves it, or say NONE." That is the exact job the screening software will do. Every NONE goes back to step 1.
Let it research. Ask what the company shipped this year, who runs the team, what the role probably exists to fix. Use one specific thing from that in your first paragraph. Not "I am passionate about your mission." Something like "I noticed the payments team split off in March, and I have run that exact split twice."
Cover the header and reread. If a stranger could tell which company and which role it is for from the body alone, it is one of the ten. If it could go to anyone, it is one of the 290.
Six applications built like that beat sixty one-shotted from the ad. It is not close.
How you know it worked: the recruiter who calls you can quote a line from your application back to you. That happens to people who write their own lines. It never happens to a template.
One honest caveat before the last step. This does not land equally on everyone. Indeed Hiring Lab Australia's numbers put AI mentions in job ads at 43% for software and data roles, 27% for IT systems, 18% for industrial engineering, 17% for marketing, 16% for legal and 2.7% for government. If you are in software, data or IT, everything above applies at full strength. A senior engineer and a logistics coordinator are heading into different 2027s, and most advice never says which one it was written for.
Quick check: open five current ads for your target title and count how many mention AI tools or AI work. Three or more, you are in the high lane and steps 2 and 3 are your priority. One or none, step 1 is your whole job for now.
Here is the thing about the pile: the best candidates usually are not in it. The recruiter found them first. Most recruiters I know spend more of their week searching than reading applications, and if one has already found you and messaged you, you never queue at all.
The search is simpler than people think. I have run it for 13 years and I show it to clients live. The recruiter opens LinkedIn Recruiter and types a title first, almost always. "Data engineer." That alone might give them 130,000 people. Then they add the two or three things the hiring manager said matter, "Spark," "streaming," "Sydney," and the list drops to a few thousand. The people at the top have those exact words in their headline and current title. The system bolds the match. They look at the first two pages. If you are on page nine, you do not exist.
Two more things you probably do not know. The system highlights people who have turned on Open to Work, and recruiters like that, because those people reply. And when a recruiter saves you to a project, that is remembered. They leave notes. They can see whether you replied to a recruiter last time.
What changed this year: LinkedIn's Hiring Assistant, the AI tool recruiters are moving to, has a memory. Their own researcher described it as giving the search "a concept of state." In plain terms, the search remembers what the recruiter wanted last time and builds on it, instead of starting from zero. Under the old system, polishing your profile the night before you applied did the job. Under one that remembers, your profile is a running signal, and a stale one keeps sending the old message. Politely, forever.
Twenty-five minutes tonight, ten minutes a month after that.
Headline. Your target title, the one recruiters type, then your two strongest specialties, then your city. "Senior Data Engineer | Spark, streaming pipelines | Sydney." Not "Passionate technologist." Nobody searches for that.
Current job title. Make it match the title the market uses for what you do, if it is honest. If your company calls you "Technical Lead III" and the market calls it "Engineering Manager," the market word goes in the headline at minimum.
First two lines of your About. Same five things from step 2, same words, before the "see more" fold.
Skills. Your top three pinned skills should be the three words a recruiter would type. Not "Leadership." The actual tool or domain.
Once a month, one small update. Put the reminder in your calendar now. Under a search with memory, small and regular beats big and once.
How you know it worked: write down your "search appearances" number today and check it a week after the edit. It moves within days. That is recruiters finding you before there is an ad.
On a 482 visa, the first ten gets filtered a second time by sponsorship, and no CV edit changes that. Target the employers that sponsor at your level before you tailor anything. Contracting? Agency recruiters search SEEK and LinkedIn far more than they read application piles, so step 5 matters more to you than step 2. Under five years in? The ad's must-haves are the whole game, so keep the CV to what the ad asks for and nothing else.
Everything above is what I tell people across the desk, and none of it needs anything but an evening.
One last thing, kept to the end on purpose. A lot of this is the manual version of what we built Careersy AI to do. It reads your CV against the exact ad the way the screening software does, requirement by requirement, and shows you which lines are missing. It checks how visible you are to the AI tools recruiters actually search with, names what is keeping you out of the results, and tells you what to change. And it helps you work out which roles to go for in the first place, so you are aiming at the two, not the twenty. You still write every line and send every application yourself. If you would rather do it by hand with the steps above, that works too. That is why they are here.
No. Australian job ad volume was down 6.0% year on year in July 2026 and applications per ad hit a record high, per SEEK's own Employment Report. ABS vacancy data shows the same direction. AI hiring tools are expanding while the market contracts.
Systems including SEEK Assist, Workday's Recruiting Agent, RightMatch AI inside Greenhouse, and SmartRecruiters' Winston Match score and rank candidates against the job's stated criteria before a recruiter reviews anything. The recruiter then reads from the top of that ranked list, usually only the first ten or so, until they have enough people to interview.
The evidence points that way, though it is an early signal from two data points rather than a measured trend. SmartRecruiters has demonstrated a prototype fraud detection tool built to flag suspicious application patterns, and Greenhouse's CEO has described AI mass-applying and AI screening escalating against each other. The larger cost is that one-shot AI CVs read alike, and a CV that looks like 289 others does not rank in the first ten.
Unevenly. Indeed Hiring Lab Australia data shows AI-mention share ranging from 43% in software and data roles down to under 3% in government postings, while some of the fastest-growing categories against pre-pandemic levels, cleaning, logistics and personal care, are in-person roles with low AI exposure.
Ongoing. LinkedIn's Hiring Assistant runs on a persistent memory system that retains a recruiter's search preferences across sessions, so a profile that is updated over time carries more weight than a single edit made right before applying.
Sources: SEEK Employment Report (July 2026, June 2026); ABS Job Vacancies, Australia (release dated 25 June 2026); SEEK NZ Employment Report via interest.co.nz (12 August 2026); Indeed Hiring Lab Australia (published 1 April 2026, 30 January 2026 and 22 April 2026); SEEK Assist product page; Workday blog citing Gartner (8 May 2026); Greenhouse support documentation (updated 27 March 2026); SmartRecruiters GlobeNewswire release (7 April 2026); LinkedIn business and help documentation; Surface.ai "Below the Surface" podcast (recorded 4 August 2026, published 12 September 2026); Stack Overflow Blog (25 August 2026).
Enhancv, Resume.io and Grammarly all check your CV against the ATS. None was built for an Australian or New Zealand CV, and none publishes a rule about what it writes when it cannot ground a line. An honest comparison, with prices checked September 2026.
17 min read
Filling one role takes four structurally different searches. The AI sourcing tool I tested runs all four, a busy recruiter usually runs one, and most profiles can only be reached by that one.
12 min read
