How do you build a talent pool you actually use?

Because something does get saved, just not the one thing you need later on.

What usually exists is a folder of CVs, or a filter in the applicant system with a label like "interesting". Six months later a vacancy opens, someone opens that folder, sees thirty names without context, and closes it again. Then a new campaign is launched. The folder grows, the vacancy gets filled some other way, and nobody calls that a failure because nobody measures it.

The problem is not that too little gets saved. The problem is that the reason is missing. A CV tells you what someone has done, but not why they missed out at the time: too little experience with one system, a salary expectation that did not fit, a counter-offer from their current employer, or simply a stronger first choice. Those are four very different situations, and three of them have changed within six months.

On our side of the funnel we see the same group appear in every campaign: people who were good enough to talk to and who came second. For the employer the process is then closed. In practice it is a group who already know you, whom you already know, and who once made time for you. Let that group evaporate and the next vacancy starts again at too few applicants.

Three things. No more, because anything that takes longer than a minute does not get filled in.

Why it did not work out this time. One sentence, in plain language, written by whoever had the conversation. "Could not set up the machine unsupervised." "Wanted 4,500, we were at 4,100." "Got a counter-offer." That is the sentence you filter on later, and it is also the sentence that lets you reopen the conversation without sounding like a mail merge.

What the person can demonstrably do. Not a job title but the task: which systems, which machines, which certificates, how many people managed. Titles differ per company and are useless as a filter within a year; tasks are not. This is the same block you use in your role profile, so you have already written it down.

The month you will come back. One date, not "sometime". After a counter-offer we work with six to nine months. After a pay gap it is the moment your own scale changes. After too little experience it is once the person has had time to gain it. Without a date it does not happen, because there is never a day on which somebody thinks: let me open that folder.

What you do not record matters just as much. No loose impressions about someone's personality, no notes you would not read out with that person in the room. How you keep data from a live process is part of how you have set up screening and candidate follow-up: ask permission to approach someone later and attach a time limit to it.

With a reason, not with a rhythm.

The temptation is a newsletter: a quarterly update to everyone on the list. It is cheap and it rarely works, because it is a message to nobody in particular. Someone who was turned down and then starts receiving generic post mostly reads in it that they have become a row in a database.

What does work is one message in the month you wrote down, from the person who had the conversation, with the reason in it. "You mentioned at the time that you did not know the planning system. We now have a role where you would learn it." That is not a mailing, it is a follow-up to a conversation both sides remember.

Whoever does this, does it themselves. This is not a task for automation and not one for an intern: the value lies in the sender having had the conversation. With a bigger list that means you do not maintain everyone, only the people who made you think "shame" at the moment you turned them down. That is a handful per process.

What people on the list see from you in the meantime decides whether that one message works. Someone who hears nothing from you after a rejection except a vacancy reads that vacancy as an advert. That is why a pool you use later runs through your employer branding rather than separately from it. If you want to reach people who never applied at all, you are dealing with passive talent and that is a different craft: that is active sourcing, and it starts with a search rather than with a list.

How big should a talent pool be?

Smaller than you would think. Fifteen to thirty people per job family is enough to have something when a vacancy opens, and small enough to maintain. A list of three hundred names without a reason attached is worth less than twenty names with a sentence and a date, because with three hundred nobody starts calling and with twenty they do.

Does this work for hard-to-fill roles too?

That is where it works hardest, for an uncomfortable reason: with scarce profiles you only speak to a few suitable people a year, so starting from scratch every time is expensive. At the same time the group you build is small, so you have to make do with few names. Do not expect it to pay off in the first few months, and use it alongside your regular recruitment rather than instead of it.

Who maintains it, HR or the hiring manager?

Whoever had the conversation writes the sentence and the date; HR makes sure it happens and that nothing goes in that should not. Hand the task to someone who was not in the room and you get a summary of the CV instead of a reason, and the reason is exactly what makes the pool usable.

What about people who said no themselves?

They belong in it most of all, with their reason. Someone who walked away over the commute, the shift pattern or the offer told you something about your own proposition at that moment. Change that condition later and you immediately have a list of people you can tell that it has changed. That is the only moment a returning message genuinely carries news.

At SocialFind we record the reason at every rejection, because on the next campaign for the same client we often meet the same people again. If you would like articles like this in your inbox, sign up for our newsletter.