Industry

Digital, software and AI for recruitment and staffing

A two-sided market where the database is the entire asset, and where automated screening is now a regulated activity.

Recruitment is one of the few businesses where the core asset is literally a database, and where its quality decays continuously — people change jobs, contact details go stale, and the value of a record falls the longer nobody touches it.

It is also the sector where AI has moved fastest from opportunity to regulatory question. Automated screening and ranking of candidates now falls within high-risk classifications in the EU AI Act and within specific rules in several US jurisdictions, which changes what may be built and what has to be documented.

Why this sector is moving now

The two-sided problem shapes everything. An agency has to market to clients and to candidates simultaneously, with different messages, different channels and different systems, and most agencies do one well and the other by accident.

Database decay is the quiet cost. Contact data goes stale at a meaningful rate every year through job changes alone, which means a database that is not actively maintained is losing value faster than most agencies account for.

Speed is the competitive dynamic in contingent recruitment, much as it is in property. The agency that submits a credible shortlist first wins a disproportionate share, which makes matching and outreach speed the operational priority.

The pressures behind it

Database decay: Contact records going stale through job changes at a rate that erodes the core asset annually.
Two-sided marketing: Clients and candidates needing different messages, channels and systems simultaneously.
Submission speed: A credible shortlist first winning disproportionately in contingent work.
Regulated screening: Automated candidate ranking now classified as high-risk under EU rules and regulated in several US jurisdictions.
Margin compression: Client procurement pushing fees down and preferred supplier lists narrowing.
Consultant turnover: Relationships and knowledge leaving with the consultant who held them.

Where the work usually starts

The systems and channels a recruitment agencies typically needs joined up. Most engagements begin at one spoke and widen only if it earns it.

Usually data hygiene, because every other improvement is built on the database and most agencies have never systematically deduplicated or enriched it. It is unglamorous and it raises the ceiling on everything that follows.

Matching and outreach automation come next, then candidate-facing communication. Automated screening that ranks or filters people is deliberately last, and only with the regulatory position established first.

Marketing and brand for recruitment agencies

Brand Strategy & Development: Agencies position on sector and speed, which every competitor also does. The differentiator that survives a procurement conversation is usually depth in a narrow discipline or an unusual guarantee, and it requires turning down work outside it.
Brand Management: Consistency matters across two audiences with genuinely different expectations. A tone that works for candidates can read as unserious to a procurement team, which is an argument for two message frameworks rather than one voice.
Social Media Strategy: LinkedIn is where both sides are, which is convenient and creates the two-audience problem in one feed. Consultant-led content outperforms agency accounts, as it does across professional services.
Social Media Management: Consultants posting in their own voice about their own market is the model that works. The agency's role is enabling it — templates, prompts, compliance boundaries — rather than producing content centrally that consultants will not share.
Content Creation & Creative Production: Salary and market data from your own placements is the asset competitors cannot copy, and it earns attention from both audiences at once. Most agencies have the data and never publish it.
Digital Marketing: Two funnels that need measuring separately: client acquisition and candidate attraction. Blending them produces a number that describes neither, and agencies routinely report candidate volume as if it were commercial progress.
Paid Advertising: Job advertising and client acquisition are entirely different disciplines with different platforms and economics. Job boards and programmatic job advertising are their own specialism; client-side advertising is ordinary B2B marketing.
Search Engine Optimisation: Job pages create a structural problem — thousands of short-lived, near-identical pages that can become an index bloat issue rather than an asset. Sector and salary guide content is the durable half, and job page indexing needs deliberate handling.
Email, SMS & WhatsApp Marketing: Candidate communication is where the volume is and where consent is most often assumed rather than documented. SMS and WhatsApp get answered where email does not, particularly for contract and temporary placement.
Lead Generation & Prospecting: Client-side prospecting from hiring signals — job postings, funding rounds, leadership changes — is genuinely effective because the trigger is public and timely. It is one of the clearest cases for event-driven outbound in any sector.

IT, software and AI for recruitment agencies

Website Design & Development: The job search experience is the product for candidates and largely irrelevant to clients, who are checking credibility. Most agency sites optimise entirely for the first and leave the second to a generic about page.
CRM & Sales Systems: The ATS and the CRM are frequently the same system used badly for both purposes. Client pipeline, candidate pipeline and the relationship between them need modelling deliberately rather than inherited from a template.
Business Process Automation: Compliance document collection, reference chasing, timesheet reminders, placement paperwork and interview scheduling. High-frequency administrative work that consumes consultant time that should be spent on conversations.
AI Automation Systems: CV parsing and enrichment are safe and valuable. Ranking or filtering candidates is a regulated activity in the EU and in several US jurisdictions, and we will not build it without the client having established their regulatory position and, frankly, taken advice.
AI Knowledge Bases & RAG: Client requirements, past placements, market intelligence and consultant notes, retrievable with citations. It directly addresses the knowledge-leaves-with-the-consultant problem, which is the sector's main structural risk.
AI Voice & Customer Communication: Candidate screening calls are technically feasible and reputationally risky, because a candidate who feels processed by a machine tells other candidates. Availability checking and interview scheduling are the defensible uses.
Custom Software & Platforms: Justified for agencies with an unusual model — statement-of-work delivery, complex contractor compliance, sector-specific credentialing — that standard ATS products do not handle. Not justified as an ATS replacement.
Data Engineering & BI: Fill rate, time to submit, submission-to-interview ratios and margin by client and consultant. Most agencies measure placements and revenue and cannot see which part of the process is actually failing.
Cloud, DevOps & Infrastructure: Candidate data includes CVs, right-to-work documents and sometimes background check results, which carry retention obligations and heightened sensitivity. Where it sits and how long it is kept are compliance decisions.
Systems Integration: ATS to job boards, ATS to CRM, payroll and timesheets to accounting, compliance documents to the placement record. Job board posting in particular has to be idempotent, because duplicate listings are publicly visible.
Digital Transformation Consulting: The audit usually finds consultant time concentrated in scheduling and compliance chasing rather than in the conversations that make placements. It also frequently finds the database is worth substantially less than assumed.
Maintenance & Ongoing Support: Job board APIs change, compliance requirements change, and a broken posting integration is discovered when applications stop arriving. Monitoring matters because the failure is silent rather than loud.

What is specific to this sector

Automated screening or ranking of candidates falls within the high-risk category of the EU AI Act, which brings documentation, human oversight and conformity obligations. Several US jurisdictions regulate automated employment decision tools separately, including bias audit requirements. This is the clearest example in our work of a use case that is technically straightforward and regulatorily heavy, and we treat it accordingly.

Candidate data is personal data collected for a specific purpose, and retaining a CV indefinitely on the theory that someone might be placed later is a common practice with a weak lawful basis. Retention policy and a genuine re-consent mechanism are worth building rather than assuming.

Right-to-work and background checking obligations differ by country and carry penalties for the employer and sometimes the agency. Systems that track document expiry and block placement on lapse are doing compliance work rather than administration.

Temporary and contract placement brings employment and payroll obligations that vary sharply by market — worker classification, agency worker rights, and in several countries a licensing regime for labour supply. Systems that model permanent placement only will mishandle the contract book, which is frequently the larger part of the business.

Not legal or regulatory advice. Sector rules described here are scoping context, current to our latest review. Confirm what applies to your business with a qualified adviser.

Questions

Can we use AI to screen candidates?

Parsing and enriching CVs, yes. Ranking or filtering people is high-risk under the EU AI Act and regulated separately in several US jurisdictions, and we will not build that without you having established your regulatory position and taken advice. It is the one area where we routinely say no.

What is the highest-return project?

Usually database hygiene, which is nobody's idea of an exciting engagement. Deduplication and enrichment raise the ceiling on matching, outreach and reporting simultaneously, and most agencies have never done it systematically.

How should we handle job pages for SEO?

Carefully. Thousands of short-lived near-identical pages can become an index bloat problem rather than an asset. Sector and salary content is the durable half, and job page indexing needs a deliberate policy rather than a default.

Should we market to clients or candidates?

Both, separately, with different messages and separate measurement. Blending them produces a number that describes neither, and candidate volume gets reported as commercial progress when it is not.

How long can we keep candidate data?

Long enough to justify against the purpose it was collected for, which is shorter than most agencies assume. A retention policy with a genuine re-consent mechanism is worth building, and indefinite retention is hard to defend.

What about WhatsApp for candidates?

It gets answered where email does not, particularly in contract and temporary work. Bringing it into the ATS rather than leaving it on consultants' phones is both an operational and a data protection improvement.

What does it cost?

Quoted per phase after a discovery call. Where an engagement touches automated screening we would scope the regulatory work explicitly rather than absorbing it, because pretending it is free serves neither of us.

Should we buy a new ATS?

Usually not. Most agencies are using perhaps a third of what their current system does, and the gap is configuration and data hygiene rather than capability. We would establish that before recommending a migration that will consume a year.

How do we stop knowledge leaving with consultants?

Structurally, by making the record the asset rather than the relationship — notes, client requirements and history captured in the system as a matter of process. A retrieval assistant over that record then makes it usable, but the capture discipline has to come first.