Talent Intelligence

Who typically owns talent intelligence?

In larger organizations it usually sits within talent acquisition rather than in the analytics function, because the questions arrive through recruiting and the internal data lives in recruiting systems. Placing it in a central analytics team often produces technically sound analysis that answers a question nobody is currently facing. Where a dedicated role is not justified, the work is commonly absorbed by a recruiting operations or workforce planning lead. The important condition is not where it sits but that it is engaged before requisitions are approved, since intelligence delivered after a search has begun can only explain a difficulty rather than prevent it.

How should talent intelligence be presented to executives?

As a small number of decisions with the assumption behind each made explicit and its source named, not as a data pack. The most useful format states the plan, the constraint discovered, the options, and what would have to be true for the current plan to work. Executives act on a clearly stated constraint and ignore a dashboard. Sourcing discipline matters more here than anywhere else, because a number quoted in an executive discussion detaches from its source within one meeting and reappears later as an established fact, so every figure should carry its origin in the same sentence and modelled estimates should be labelled as estimates.

Can talent intelligence predict future skill demand?

It can describe direction with reasonable confidence and struggles with timing and magnitude. Observing that postings for a skill have risen across several markets over consecutive periods is a defensible statement about the recent past that suggests a trend. Converting that into a specific forecast requires assumptions about technology adoption and economic conditions that are not reliably knowable, and published projections in this area have a poor track record. The honest framing for planning is a range of scenarios with the assumption behind each stated, so a plan can be tested for fragility rather than built on a single predicted number.

What distinguishes talent intelligence from adjacent disciplines?

Three disciplines are routinely confused because they share tooling and often the same analyst. People analytics studies the workforce you already employ: attrition patterns, internal movement, engagement, performance distribution. Talent mapping is a targeted exercise that identifies named individuals in defined roles at defined organizations, usually to support a specific senior search. Talent intelligence sits above both, using aggregate external market information alongside internal hiring data to answer planning questions.

The questions it answers are different in kind. People analytics answers what is happening inside. Talent mapping answers who is out there for this one role. Talent intelligence answers whether a plan is feasible at all: whether the skills exist in the locations under consideration, what competition for them looks like, how the picture is changing, and which assumption in the plan is most likely to be wrong. Because the question is strategic, the output is a recommendation about approach, not a candidate list.

What sources does it draw on?

External sources fall into a few categories. Government and multilateral statistical agencies publish labour force, occupation, and wage data, which is the most rigorous material available and also the slowest to update. Aggregated job posting data shows what employers are advertising for and where, which is fast but reflects demand rather than supply. Professional network and education data indicates where skills are concentrated. Each has known limitations, and any figure taken from one should be attributed to it explicitly when it is used to justify a decision.

Internal sources are the ones organizations underuse. Their own applicant tracking system holds where past hires came from, which searches produced viable candidates, where offers were declined and why, how long comparable roles took, and which requirements repeatedly failed to find a match. That record is specific to the organization and its actual brand position, which makes it more predictive for its own next hire than any external average, and it costs nothing to obtain.

What decisions does it actually change?

Location is the most common. When a plan calls for a number of specialised roles in one city, the intelligence question is whether that concentration of skills exists there, how many other employers are competing for it, and what the alternatives are. The answer may be to distribute the team, to open in a different location, or to accept a longer timeline. Making that call before requisitions open is far cheaper than discovering it after months of unsuccessful search.

The second is build versus buy. Where a skill is scarce everywhere, hiring for it competes with every other employer facing the same constraint, and the realistic options are to train internally, redesign the role to require a more available adjacent skill, or engage contract capacity. Talent intelligence is what makes that a deliberate choice rather than the conclusion an organization arrives at by default after a search has failed.

How does it inform a compensation or location decision?

By separating the two questions that get merged. Whether a range is competitive is a compensation benchmarking question answered with survey data for that role and market. Whether enough people exist in that market to fill the plan is a supply question, and a fully competitive range does not help if the population is too small. Organizations that only ask the first question end up paying at the top of a market that cannot supply the volume they need.

The practical output is a short assessment per planned role: is the skill available here, at what level of competition, and what would change the answer. Where the assessment is uncomfortable it is more valuable, because the alternative is discovering the same constraint through a failed search that consumed a quarter of recruiter capacity.

What are the honest limits of this work?

Most external data describes demand better than supply, because job postings are easy to collect and people are not. Posting volumes also double-count: the same role reposted across several boards can appear as several openings, and postings that were never real appear alongside those that were. Anyone presenting posting-derived counts as a measure of market size should say what the source counts and how.

Coverage is also uneven by geography and by occupation. Data on professional roles in large economies is comparatively rich; data on operational roles in smaller markets is thin, and gaps are frequently filled by estimation whose method is not disclosed. The defensible practice is to name the source next to any figure that informs a decision, state its known weakness, and avoid presenting a modelled estimate as a measurement.

How does a team start without a dedicated function?

Begin with the internal record, because it is free, specific to the organization, and usually unexamined. Where hires actually came from over the past two years, which requirements repeatedly produced no viable candidates, where offers were declined and for what stated reason, and how long roles by level and location genuinely took. Answering those four questions from the applicant tracking system produces a more useful picture than most first external purchases.

Add external sources only against a specific decision that is already pending, such as a location choice or a plan for a scarce skill. Buying broad market data with no decision attached produces reports nobody reads. Starting from the decision keeps the scope narrow enough to finish and makes it obvious afterwards whether the information changed anything.