How to find candidates already in your database

The average recruitment agency database contains thousands of candidates. A significant share of them will never be surfaced again, not because they are unsuitable, but because nobody searched the right way. Your next placement is almost certainly already in there. The question is whether your search method is good enough to find it.

Most consultants default to keyword searches: job titles, qualifications, specific tools. It is fast, familiar and, for a large portion of your database, completely inadequate. A candidate who listed "stakeholder management" in one place and "client liaison" somewhere else is, to a keyword search, two different people. They are not. And your competitor who figures that out first gets the fee.

Why keyword search leaves money on the table

Keyword matching works by finding an exact string in a document. That sounds reasonable until you remember that candidates do not describe themselves consistently. One person writes "P&L responsibility"; another writes "managed a budget of £2m"; a third puts "financial accountability". All three could be equally suited to a Finance Director role. A keyword search for "P&L" retrieves only the first.

The problem compounds over time. Every year your database grows, but the candidates added earliest are the least likely to appear in search results, simply because their CVs predate the vocabulary currently in fashion. That is not a data-quality problem. It is a search-method problem, and it has a direct cost. Research by the Aberdeen Group found that sourcing a candidate from an existing database costs roughly six times less than sourcing from scratch. If your search method is missing half the relevant profiles, you are paying over the odds every time you post a job externally and wait.

Searching by capability, not job title

Capability-based searching shifts the frame from what someone called themselves to what they can demonstrably do. It asks: what skills, behaviours and outcomes does this role require, and which candidates in the database evidence those things, regardless of the exact language they used?

This is where semantic search becomes genuinely useful. Rather than matching strings, a semantic search engine reads meaning: it understands that "managed cross-functional teams" and "led multi-disciplinary projects" are describing similar capabilities, and it surfaces both. The practical effect is that you retrieve candidates you would otherwise have scrolled past or, more likely, never seen at all.

The technology is only as good as the configuration work behind it, of course. A semantic search tool set up without careful thought about how your consultants describe roles, and how your candidate pool tends to describe themselves, will still return noise. The setup is where the real investment sits. But when it is done well, the difference is not marginal. It is the difference between sourcing from five hundred relevant profiles and sourcing from fifty.

The hidden value of passive candidates in your own records

Registered candidates who applied for a role two or three years ago occupy a curious position. They are not actively looking, so they will not appear on job boards. They are not unknown to you, so you do not need to pay a sourcing fee. They are, in effect, a warm audience that most agencies treat as a cold one.

This matters because the labour market is not static. A candidate who was unavailable eighteen months ago may have since been made redundant, may have completed a qualification, or may simply have decided they are ready for a move. As recruiters, we rarely go back to ask. The assumption that an old registration means an unavailable candidate is, in most cases, untested.

A structured re-engagement programme, informed by a proper capability search rather than a keyword trawl, can surface a meaningful number of placeable candidates from records that have been sitting dormant. It is not glamorous work. It is, however, reliable.

How to run a capability-based search in practice

Before running any search, it helps to build a capability map for the role. This is simply a list of the things the successful candidate must be able to do, expressed in plain language, with synonyms and related phrases noted alongside each one. It takes perhaps fifteen minutes per role and it transforms the quality of your search inputs.

From there, consider structuring your database search in layers:

• Start with the core capability, not the job title. Search for what the person must do, not what they have been called.
• Layer in sector or domain knowledge as a secondary filter, not a primary one. Many skills transfer across sectors; filtering by sector too early eliminates transferable candidates.
• Use location and availability as final filters, not first ones. You want to know who is capable before you start narrowing on logistics.
• Review the bottom of your results, not just the top. Semantic matching ranks by relevance, but a candidate ranked fifteenth may be a better practical fit than one ranked third.

It also helps to review your CV parsing settings periodically. Parsed data is the raw material of any database search, and if your parser is discarding or mis-categorising information at the point of ingestion, no search method will recover it later. CV parsing quality is one of those configuration details that is easy to overlook and consistently worth revisiting.

Making your database a living asset

A database that is never refreshed becomes a liability rather than an asset. GDPR obligations aside (candidates must have a lawful basis for remaining on your system and must be able to exercise their rights), a database full of stale, uncontacted records is one that consultants stop trusting. Once trust goes, so does usage. The database becomes a formality rather than a tool.

Keeping your database genuinely useful involves a few habits that, when built into the weekly rhythm of the desk, are not onerous. Consider flagging candidates for re-engagement at set intervals after their last contact. Consider updating records when a placement falls through, capturing the reason, because that context is often valuable the next time a similar role comes in. And consider treating every inbound application as an opportunity to enrich existing records, not just to fill the current vacancy.

The agencies that get the most from their candidate data are not necessarily the ones with the biggest databases. They are the ones whose consultants actually trust what is in there. Seeing how eBoss handles candidate matching is a reasonable next step if you want to understand what that looks like in practice.

What to do about this now

The practical place to start is a single role you are currently working, ideally one where external sourcing has felt slow or expensive. Build a capability map for it, as described above, and run a database search using capability terms rather than job titles. Compare the results to what your standard keyword search returned. The gap between those two lists is a reasonable proxy for how much your current method is costing you.

If the gap is small, your search setup is working well. If it is large, you have just found your next sourcing efficiency project, and it will not require a single external job board posting to fix it. Good recruiters with the right tools will always outperform the same recruiters with the wrong ones. The tool, in this case, is already in front of you.