How to choose a recruitment CRM

Choosing a recruitment CRM is one of the few decisions in an agency that is genuinely hard to reverse. You will be inside it every day for years, your whole team's habits will form around it, and your data will accumulate in its shape. Most agencies do the decision badly, not through carelessness but because the process everyone falls into is designed by the vendors.

That process is a shortlist from a comparison site, three demos, a feature checklist, and a negotiation on price. Every step of it is a step the vendor is good at. Here is a better one.

Start with your week, not with a shortlist

Before you speak to anybody, write down what actually happens in your agency in a normal week. Not the process diagram from the induction pack. The real one, including the spreadsheet somebody keeps and the reason they keep it.

Sit with two consultants for an hour each, one of your strongest and one of your newest. Write down every tool they open, every time they type something that already exists somewhere else, and every moment they stop and ask someone a question the software should have answered. Do the same with whoever runs your back office.

You now have something no vendor can supply, which is a description of your own agency in enough detail to test against. Everything that follows is a test against this list rather than against a feature grid.

Write down the four or five things that would actually change your P&L

Most requirement lists run to eighty items and are useless, because a list of eighty things cannot distinguish between vendors. Every serious product does seventy-five of them.

The useful list is short and specific to you. It might be that your consultants lose an hour a day to formatting CVs. It might be that your forecast is assembled by hand from three places every Monday. It might be that contract timesheets are the reason your finance person works Saturdays, or that nobody can tell you which of your clients has gone quiet.

Four or five of those, written as a sentence each, with the hours or the money attached. That is your actual requirement. The other seventy-five items are table stakes and you can check them in an afternoon.

Run the demo, do not watch it

A demo is a rehearsed tour of the parts that work. The person giving it has given it four hundred times and knows exactly which roads are paved. You will come away impressed and no better informed.

So take the controls. Send your four or five sentences in advance and ask them to show you those, with your data if possible and theirs if not. Then ask to drive for fifteen minutes yourself, doing one ordinary thing end to end, like taking a job from a client email through to a shortlist sent out.

Three things to watch while you do it. How many clicks the ordinary thing takes, because you will do it forty times a week. What happens when you type something slightly wrong. And how the software behaves when it does not have the information it needs, which is the moment that tells you the most about the people who built it.

Ask the five questions that are hard to prepare for

Question	Why it is hard to fake
What shipped last week, and how much of it was fixes rather than features?	It needs a dated changelog and an honest ratio. Most have neither
What is the oldest part of the product, and what is the plan for it?	Every product has one. Only the honest ones will say so
Which capabilities are yours, and which are a partner's?	Reveals what happens to those parts if a partnership ends
Show me your AI doing something it is unsure about	Cannot be rehearsed, and shows whether it guesses
What does it take to get all our data out?	The answer tells you how they behave the day you want to leave

Ask all five of every vendor, including the one you already like. Write the answers down, because by the third demo they blur.

Talk to a customer they did not choose for you

Reference calls arranged by a vendor are worth very little, because the vendor picks a happy customer and briefs them. Do it anyway, and then do the useful version as well.

The useful version is finding an agency of roughly your size and shape that uses the product, through your own network rather than theirs, and asking three questions. What did they underestimate about the switch. What do they still work around. And whether they would choose it again today knowing what they know.

The third one is the whole conversation. People will be diplomatic about the first two and honest about the last.

Cost it properly before you negotiate

The licence is the part everybody compares and the smallest part of the answer. Before you talk about price, build the real number: licences, the costs that arrive per use like call minutes, credits and signature envelopes, the implementation and migration cost, and the hours your people spend retyping between systems that do not talk.

Then express the whole thing per recruiter per month, and again as a share of your average placement fee. The second version is the one that makes the decision obvious, because a monthly figure always sounds small and the same figure set against a fee does not.

A system that costs more per seat and removes two of your other tools is usually cheaper. A system that is cheap per seat and needs three integrations is usually not.

Check that it will still be good in three years

This is the part that gets skipped, and it is the part that decides whether you are doing this again in thirty-six months.

Ask for the changelog with dates and read three months of it. Look for two things: whether it moves every week, and whether every module moves. A product where the candidate screens change constantly and the phone system has not moved in a year is telling you which parts were built to complete a feature list.

Then ask what the newest thing in the market is and when they had it. A vendor who adopts slowly is not behind once. They are behind on everything that arrives for as long as you are a customer, and the gap widens every year.

The traps

Choosing on the demo. The demo shows finished parts. Your four sentences and fifteen minutes at the controls show the rest.

Choosing on the feature grid. It cannot separate serious vendors, so it quietly hands the decision to price.

Letting the loudest consultant decide. Whoever shouts most in the evaluation is usually the person most attached to the current way of working.

Ignoring the back office. Placements, timesheets, invoicing and compliance are where switching goes wrong, and they are never in the demo because they do not look exciting.

Believing a roadmap. A roadmap is a list of intentions with no dates behind them. A changelog is a list of things that happened. Every agency I know that ended up trapped on dying software got there by reading roadmaps.

A three-week shape

Week one, write your own week down and produce the four or five sentences. Week two, three demos where you drive, the five hard questions, and your own reference calls. Week three, build the real cost, check the changelog, and decide.

That is faster than most agencies manage and it is also better, because almost all the wasted time in a normal evaluation goes on watching demos and ticking boxes that cannot separate anybody.

Run this at us as well as at everyone else. If our answers are not better, buy the other one. That is why I say Recruitly is the best recruiting CRM in the world, and why I would rather you tested the claim than took it.

---

Gowri is the CEO of Recruitly.