This post covers how to work a large candidate database with the team you already have: why size is not an asset by default, why adding more records makes it worse, the three jobs that actually need doing, and how to tell whether any of it is working.<\/p>\n
<\/span>A large database is not an asset by default<\/span><\/h2>\nMost firms describe their database in record counts, which is a measure of how much has been collected rather than how much can be used.<\/p>\n
The useful question is different. Of those 50,000, how many have a current employer you would trust, an owner who still works here, and enough context that a recruiter could decide in thirty seconds whether to call them? For most firms the honest answer is a few thousand.<\/p>\n
The rest are not assets. They are storage.<\/p>\n
<\/span>Adding more makes it worse<\/span><\/h2>\nThis is the counterintuitive part, and the data is unambiguous. Top-quartile firms add fewer candidates than everyone else, 800 per recruiter against 930, and place 5.21 against 1.38 (Source: The Economics of Recruiting<\/a>).<\/p>\nAdding records is the easiest activity to measure and the easiest to be busy with. It also dilutes the signal in everything already there, and every new record adds to the maintenance load nobody is doing.<\/p>\n
The firms placing four times more are not sourcing harder. They are converting better, and a database they can actually work is a large part of why.<\/strong><\/p>\n<\/span>The three jobs that need doing<\/span><\/h2>\nMake the records answer questions<\/h3>\n
The reason a 50,000-record database feels unusable is that it can only be queried by the fields someone filled in, which is usually title, employer and location.<\/p>\n
Everything that would actually decide a shortlist sits in the notes: what the candidate said they wanted, why they turned something down, the project they described on a call. That material is stored and unreachable, which for practical purposes is the same as not having it.<\/p>\n
This is the single highest-leverage fix, because it changes the answer to “who do we already know for this” from a handful of names someone remembers to everything the firm has ever learned. Around 71% of placements come from candidates already in the database before the job opened (Source: The Economics of Recruiting<\/a>), so this is not a marginal gain.<\/p>\n![\"The](\"https:\/\/recruiterflow.com\/blog\/wp-content\/uploads\/2026\/05\/5-1024x342.png\")
<\/a><\/p>\nKnow when something changes<\/h3>\n
A database decays quietly. People move, get promoted, leave the market. A record that was accurate in 2023 is a wrong answer in 2026, and nothing in the system announces that.<\/p>\n
Job change detection is the highest-value piece of automation available to a recruitment firm, because a candidate who has just moved is both a placement risk and a client lead. Firms using the AIRA Job Change Alert Agent reduce time to first submittal by 34% and see 12% higher placements on average (Source: The Economics of Recruiting<\/a>).<\/p>\nThat is the clearest example of the principle: the work gets done without anyone doing it.<\/strong><\/p>\nRe-engage on a cadence, not a campaign<\/h3>\n
Most firms run database re-engagement as an occasional project. Someone exports a segment, sends a mailout, gets a handful of replies, and the exercise is not repeated for a year.<\/p>\n
A cadence is different. A defined touch on a defined segment at a defined interval, running continuously, owned by the system rather than by whoever has a quiet week. Five hundred contacts a month on a rolling basis beats fifty thousand once.<\/p>\n
<\/span>What this looks like with six recruiters<\/span><\/h2>\nNobody is reviewing 50,000 records. The work is structured so that the database surfaces candidates rather than being searched.<\/p>\n
\n- Segment once.<\/strong> Split by discipline and seniority, not by source or date added. Four or five segments is enough.<\/li>\n
- Set a cadence per segment.<\/strong> Placed candidates on a guarantee-period touch. Interviewed-but-rejected on a six-month nurture. Five-star candidates quarterly. These run automatically once configured.<\/li>\n
- Let the alerts drive the priority list.<\/strong> Job changes, contract end dates approaching, candidates who opened something. These are the records worth a human call this week, and there will be a manageable number of them.<\/li>\n
- Reserve the human hours for the shortlist.<\/strong> Screening to submission is the largest leak in the funnel at 11.3%, and it is the one stage that cannot be automated. Every hour saved on database maintenance should be spent there.<\/li>\n<\/ul>\n
<\/span>How to tell if it is working<\/span><\/h2>\nOne measure, tracked monthly: what proportion of your submissions came from candidates already in the database before the job opened.<\/p>\n
If it is flat while your record count grows, you are collecting rather than recruiting, and a seventh recruiter will not fix it.<\/strong> Put it alongside your other recruiting metrics<\/a> and review it with the same seriousness.<\/p>\nBefore any of it, run a database audit<\/a> so you know what proportion of records have a reliable current employer and owner. Cadences built on stale fields reach the wrong people convincingly.<\/p>\n<\/span>Start with the records you already have<\/span><\/h2>\nAIRA Search<\/a> reads across the notes, calls and transcripts your team has recorded rather than only the extracted fields, so a candidate who described the right experience on a call two years ago is findable on that basis. Every result cites the roles, companies and dates behind it, so you can see why someone surfaced before you pick up the phone.<\/p>\n\nBring one live brief to a demo<\/a> and we will run it against your own database, and you can see how many of the answers were already in there.<\/p>\n<\/blockquote>\n
This is the counterintuitive part, and the data is unambiguous. Top-quartile firms add fewer candidates than everyone else, 800 per recruiter against 930, and place 5.21 against 1.38 (Source: The Economics of Recruiting<\/a>).<\/p>\n Adding records is the easiest activity to measure and the easiest to be busy with. It also dilutes the signal in everything already there, and every new record adds to the maintenance load nobody is doing.<\/p>\n The firms placing four times more are not sourcing harder. They are converting better, and a database they can actually work is a large part of why.<\/strong><\/p>\n The reason a 50,000-record database feels unusable is that it can only be queried by the fields someone filled in, which is usually title, employer and location.<\/p>\n Everything that would actually decide a shortlist sits in the notes: what the candidate said they wanted, why they turned something down, the project they described on a call. That material is stored and unreachable, which for practical purposes is the same as not having it.<\/p>\n This is the single highest-leverage fix, because it changes the answer to “who do we already know for this” from a handful of names someone remembers to everything the firm has ever learned. Around 71% of placements come from candidates already in the database before the job opened (Source: The Economics of Recruiting<\/a>), so this is not a marginal gain.<\/p>\n A database decays quietly. People move, get promoted, leave the market. A record that was accurate in 2023 is a wrong answer in 2026, and nothing in the system announces that.<\/p>\n Job change detection is the highest-value piece of automation available to a recruitment firm, because a candidate who has just moved is both a placement risk and a client lead. Firms using the AIRA Job Change Alert Agent reduce time to first submittal by 34% and see 12% higher placements on average (Source: The Economics of Recruiting<\/a>).<\/p>\n That is the clearest example of the principle: the work gets done without anyone doing it.<\/strong><\/p>\n Most firms run database re-engagement as an occasional project. Someone exports a segment, sends a mailout, gets a handful of replies, and the exercise is not repeated for a year.<\/p>\n A cadence is different. A defined touch on a defined segment at a defined interval, running continuously, owned by the system rather than by whoever has a quiet week. Five hundred contacts a month on a rolling basis beats fifty thousand once.<\/p>\n Nobody is reviewing 50,000 records. The work is structured so that the database surfaces candidates rather than being searched.<\/p>\n One measure, tracked monthly: what proportion of your submissions came from candidates already in the database before the job opened.<\/p>\n If it is flat while your record count grows, you are collecting rather than recruiting, and a seventh recruiter will not fix it.<\/strong> Put it alongside your other recruiting metrics<\/a> and review it with the same seriousness.<\/p>\n Before any of it, run a database audit<\/a> so you know what proportion of records have a reliable current employer and owner. Cadences built on stale fields reach the wrong people convincingly.<\/p>\n AIRA Search<\/a> reads across the notes, calls and transcripts your team has recorded rather than only the extracted fields, so a candidate who described the right experience on a call two years ago is findable on that basis. Every result cites the roles, companies and dates behind it, so you can see why someone surfaced before you pick up the phone.<\/p>\n Bring one live brief to a demo<\/a> and we will run it against your own database, and you can see how many of the answers were already in there.<\/p>\n<\/blockquote>\n<\/span>The three jobs that need doing<\/span><\/h2>\n
Make the records answer questions<\/h3>\n
<\/a><\/p>\nKnow when something changes<\/h3>\n
Re-engage on a cadence, not a campaign<\/h3>\n
<\/span>What this looks like with six recruiters<\/span><\/h2>\n
\n
<\/span>How to tell if it is working<\/span><\/h2>\n
<\/span>Start with the records you already have<\/span><\/h2>\n
\n
![\"Book](\"https:\/\/recruiterflow.com\/blog\/wp-content\/uploads\/2026\/07\/2Notetaker-1-1024x280.png\")
<\/a><\/p>\n