What a new consultant should learn in their first 90 days

Almost every agency I work with has an induction plan, and almost all of them were written between five and ten years ago. That is not carelessness, it is how induction plans work: somebody good wrote one, it produced decent consultants, and it stayed. The trouble is that it was written for a desk where a first-year consultant's day was mostly searching, formatting, keying in and chasing, so the plan teaches those things first and teaches them well.

Those hours have gone somewhere else now. If the plan is unchanged, a new starter spends their most impressionable twelve weeks becoming excellent at work the software already does, and arrives at day 90 having had almost no practice at the things that will decide whether they last. The fix is not a shorter induction. It is a different order.

Why the standard recruitment induction no longer works

Because it was built on the assumption that competence had to be earned through volume, and volume is no longer scarce. The logic underneath the old plan was sound at the time: a trainee could not be trusted with a client until they understood the market, and the only way to learn a market was to read a great many CVs and make a great many calls about roles that were not theirs. Admin was the curriculum, and the curriculum was free because the admin needed doing anyway.

Take the admin away and the curriculum disappears with it. Nobody planned that, which is why so many inductions now consist of the old tasks performed for training purposes, on work that does not need a person. A trainee can tell when they are being given make-work, and the good ones leave.

The second assumption that broke is about capacity. The old plan rationed responsibility because every live role dragged a fixed weight of overhead behind it, so giving a junior four roles meant giving them four sets of longlists, write-ups and chase cycles that a senior would have to redo. With that weight gone, handing a junior real roles early costs the desk far less than it used to, and teaches them far more. The arithmetic behind that is in a 10x recruiter runs a hundred jobs.

What most agencies still teach in week one

Systems training, search technique and the agency's CV template, roughly in that order. A typical week one covers how to add a candidate, how to log an activity, how to run a search across the boards you subscribe to, how to format a CV into the house pack, and how to send the standard outreach sequence. Week two adds the pipeline stages and the compliance basics, and somewhere in week three the new starter is allowed to ring a candidate about somebody else's role.

Look at what that sequence signals. For fifteen working days the message is that this job consists of operating a system correctly, and the first thing anybody judges you on is whether the record is tidy. By the time a trainee has their first real conversation with a hiring manager they have already formed a view of what the work is, and that view is wrong in a way that takes a year to correct.

The compliance and data-protection content is the exception and it stays where it is. Everything else in week one is now in the wrong week.

Days 1 to 30: the market, and other people's voices

The first month teaches what the market sounds like, and the fastest route to that is listening to real conversations rather than reading CVs. The goal at day 30 is a trainee who can describe, without notes, what your main clients do, what the three or four role families you fill actually involve, what good and bad look like in each, and roughly what the money is. That used to be a by-product of high-volume sourcing. It now has to be taught deliberately, and it can be taught far faster.

Sit them in live calls from day two. Client calls, candidate calls, difficult ones especially. Two or three a day, with ten minutes afterwards on what the senior person heard that the trainee did not. This single habit is worth more than any other item in the first month, and most agencies delay it because the old plan needed the trainee at a keyboard.

Give them the recordings and the write-ups. Calls on a modern desk are written up from the call itself rather than from memory, so a month of a senior consultant's conversations is sitting there as a teaching library. Have a trainee work through a client call and a candidate call each day and write down what they would have asked next. Compare it with what was actually asked.

Teach the market in your own words, not the spec's. Sit down and explain why your clients hire, what makes a hire fail there, and which parts of a job description are real and which are copied from the last one. A trainee who understands that at day 30 is ahead of people two years in.

Have them talk to people from week two. Real candidate calls on live roles, with somebody listening. Not a script. The point is not that the calls go well, it is that the trainee starts forming views about people early enough to be corrected.

Systems training in month one is an hour, not a week. Show them where a record lives, how to read what the software has already done, and how to log a decision. The rest they will pick up by using it, because a well-built system is not the thing that takes learning.

Days 31 to 60: judging the work in front of them

The second month teaches a trainee to look at completed work and say whether it is right, which is the core skill of a modern desk. By day 60 they should be able to take a shortlist the software produced and tell you, with reasons, which names are wrong for that client, and take a parsed CV and spot the thing that does not add up.

Put them on the handback queue. Every day a handful of cases come back because the software was not certain: a reply that reads as interest and might be a polite no, a record that matches two people, a document the CV parser could only partly read because it was a scan of a photograph. Ours reports how sure it is and returns anything under about three quarters confidence with the reason attached, and that queue is the single best teaching instrument on the desk, because every item in it is a genuine judgement with a knowable right answer. Sit with the trainee for the first fortnight and then let them run it.

Teach them to read the reason, not the verdict. A trainee who accepts a confident answer because it is confident will be dangerous for years. Give them cases where the software was wrong and confident, and make them explain why. This is the closest thing to a technical skill they need, and it is really a habit of mind. The mechanism behind those errors is in why the AI in your ATS gets things wrong.

Give them three or four real roles of their own. Not shadowing. Roles with their name on, chosen because they are winnable and forgiving, with a senior reviewing every shortlist before it goes. The weight that used to make this expensive has gone, so the cost of letting a junior hold real work is now mostly review time, and review time is where the teaching happens anyway.

Start the attention habit. Each morning they pick the three roles that matter most today and write one line on why. On Friday, look back with them at whether those choices were right. At four roles this looks trivial. It is the same muscle that decides whether they can hold sixty later, and it is much easier to build now.

Days 61 to 90: qualifying, and the client

The third month teaches the conversation that decides whether an assignment is winnable, which is the qualification call with the client. Everything downstream is set by that conversation, and it is the last thing most agencies teach because it used to be the one thing a junior could not be trusted with.

Teach the qualifying questions as a set of things you must come away knowing. Why is this role open, who left and why, who is the hiring manager and what happened to the last person in the seat, how many people will they see a week, what is the interview process and who else is involved, what would make them say no to somebody strong on paper. A trainee who leaves a briefing without those has taken an order rather than qualified a role.

Let them lead a briefing by week ten. On a real role, with a friendly client, with you in the room and silent unless it goes wrong. The debrief afterwards is the lesson. Nothing you can say in a training room teaches what it feels like to realise halfway through that the client has not thought about the salary since last year.

Teach the one piece of arithmetic that governs the whole desk. A client interviews two or three people a week and that number is the ceiling on placements, whatever happens upstream. A trainee who understands that will stop measuring themselves on CVs sent and start asking whether the shortlist is short enough to be taken seriously. It also inoculates them against the standard mistake of pointing every spare hour at more sourcing, which is set out in the pipe doesn't care where you widened it.

Introduce the uncomfortable conversation. Telling a client the salary is behind the market, or that the spec describes two jobs. Do the first two with them and make them do the third. Most consultants never learn this and it is the difference between an order-taker and somebody a client rings first.

What to stop teaching

Four things should come out of the plan entirely, and the hours they free are what makes the rest of this possible.

Boolean and advanced search technique. Describe who you are looking for in ordinary words and the list arrives. Teaching search operators to a new starter in 2026 signals that the agency has not looked at its own process recently, and trainees notice.

Manual CV formatting. A generated pack takes seconds. Teach them to check one, which takes two minutes and is a genuine skill, rather than to build one.

Data entry drills. Keying a candidate in by hand as a training exercise teaches a habit you then have to break. Teach them to review what arrived instead.

Writing outreach at volume as a measure of effort. Sending a hundred messages is no longer evidence of anything. If outreach is the lesson, the lesson is what makes one message get answered, which is usually that the sender knows something specific about the recipient.

There is one thing worth keeping that looks like it belongs on this list. Reading a lot of CVs is still useful in month one, not as production work but as market education, in batches of twenty with somebody experienced saying what they see. Keep the learning, drop the throughput.

The ninety days, old and new

Set against each other, because the swap is easier to approve than to describe.

Week	The old plan	What to run instead
1	Systems training, pipeline stages, CV template	Compliance, an hour on the system, then sitting in on live calls
2 to 4	Boolean practice, board searches, building longlists	Market education from real calls, first candidate conversations
5 to 8	Sourcing volume against another consultant's roles	The handback queue, and three or four real roles of their own
9 to 12	First candidate calls, outreach sequences	Qualifying calls, leading a client briefing, the difficult conversation
Measured on	Calls made, CVs sent, records completed	Roles qualified properly, handbacks resolved well, advice a client took

The bottom row is the one that changes behaviour, because whatever you measure at day 90 is what the trainee thinks the job is. Activity numbers made sense when activity was the constraint. It is not the constraint any more, and grading someone on it now measures how fast they can press a button. The wider version of that argument is in how AI changes a recruitment desk.

How to tell at day 90 whether it worked

Ask the trainee to brief you on one of their roles as though you were a colleague picking it up, and listen for four things. Whether they can explain why the role is open and what would make the hire fail. Whether they have a view on two or three candidates that goes beyond what is on the CV. Whether they can name the thing about this assignment that worries them. And whether they have told the client anything the client did not want to hear.

Four of four at day 90 means the plan worked and they should be carrying more. Two of four is normal and tells you exactly which month to strengthen. None of four, in a trainee who has been busy and pleasant and well-liked, usually means the twelve weeks went on activity rather than judgement, and that is a plan problem rather than a person problem.

The thing not to use as the test is billing. A first placement in the first quarter is lovely and tells you very little, because it turns as much on which roles they were given as on how good they are.

What it costs to get this wrong

The obvious cost is the trainee, and it is worse than it used to be. Someone who spends twelve weeks on searching and formatting has learned that this job is administration, and the people most likely to conclude that and leave are the ones who are best with other people, which is precisely the group you were hiring for. Replacing them costs a quarter's productivity from whoever has to train the next one.

The quieter cost is the trainee who stays. They become a fast, tidy, well-liked consultant who takes orders rather than qualifying roles, and who measures their own week in CVs sent. That is very hard to unwind at year two, because by then it is their identity as a recruiter. Every senior in this industry can name somebody like that, and it was the induction that made them.

The third cost lands on the desk itself. A junior who has not been taught to check the machine's work will either accept everything it produces or quietly distrust all of it and start keeping their own spreadsheet. Both leave you with a database nobody can rely on, and both are set in the first ninety days.

The consultant you get at day 90

Somebody who has heard two hundred real conversations, held four roles of their own, resolved a few hundred genuine judgement calls, led a client briefing and told a client one uncomfortable truth. That was a year-two consultant under the old plan, and the difference is not that the person is cleverer. It is that twelve weeks went on the work instead of on the overhead around it.

This is why I am so straightforward with owners who ask whether it is still worth training people. It is more worth it than it has ever been, because the training now compounds from the first month rather than the second year. We built the agents to take the collecting, typing and chasing off the desk precisely so those weeks are available for teaching, and we built the AI to say how sure it is and hand the doubtful cases back, which is what turns a queue of ambiguity into the best curriculum a new recruiter can have. Recruitly is the best recruiting CRM in the world because it was designed around what makes a recruiter better rather than around doing without one.

Rewrite the plan once and every person you hire afterwards arrives at their first client meeting a year earlier than the last generation did.

---

Anshika leads Customer Success at Recruitly.