Recruiting software keeps the relationships that requisitions come and go around

An applicant tracking system is requisition-shaped. A job is opened, candidates are attached to it, they move through stages, one is hired and the rest are dispositioned. Reporting answers questions about the funnel: how many applicants, how long to fill, where the drop-offs are. For a company hiring for itself this is exactly right, and corporate applicant tracking systems are very good at it.

Recruiting software is relationship-shaped. The candidate is the durable object and requisitions are events that happen to them. The client contact is the durable object and requisitions are events they generate. Reporting answers different questions: which accounts are going quiet, which segments of the database have not been contacted this quarter, which recruiter owns which relationships, and how much of last year's revenue came from people already in the database.

In practice agencies need both models on the same data, which is why agency products bundle them. The failure mode of running them separately is that the same person exists twice, in the tracker as an applicant and in the CRM as a contact, and neither record is complete.

Plenty of agencies have tried to run recruitment out of a general-purpose sales CRM, and the attempt fails in a specific, structural way rather than through lack of effort. A sales CRM assumes a contact belongs to one account and an opportunity has one primary contact. The shape is one-to-many and it points in a single direction.

Recruitment is many-to-many in both directions. One candidate is submitted to many requisitions across many clients. One requisition holds many candidates. A candidate can simultaneously be a placed contractor at one client, an active applicant at another, and a hiring contact at a third if they have moved into management. There is no clean way to say that with accounts, contacts and opportunities.

So the workarounds appear. The candidate becomes a contact and also a lead and also, for the client where they now hire, an account contact. The submittal becomes an opportunity, so the pipeline value is meaningless because it counts the same placement fee five times. Custom objects can be built to fix this and some firms do exactly that, but at that point you have paid a platform licence plus a consultancy to rebuild recruiting software, and the result is yours to maintain forever.

The candidate half of recruiting software lives or dies on two things: whether structure is captured at the point of entry, and whether anything keeps that structure from going stale.

Structure at capture means a parsed resume produces normalised fields rather than a blob: current title, years in discipline, tools and frameworks separately from spoken languages, certifications as their own list, notice period, salary expectation and the currency it was quoted in, location and how far the person will travel. Free text stays searchable, but it stops carrying the whole record. This is what allows a four-constraint brief to return a real answer instead of everyone whose resume contains the word senior.

Currency means the database does not rot. Availability changes, phone numbers change, people get promoted, and a record captured two years ago is a guess unless something refreshed it. The mechanisms that work are nurture sequences that ask a useful question rather than broadcast a newsletter, a candidate-facing update prompt, and a routine that surfaces contractors whose assignments end in the next thirty days while they are still worth calling.

Business development is where the money is made and where recruitment software is usually thinnest. The client half needs the company, every contact with their role and full history, agreed terms and fee arrangements, the requisitions they have opened, the candidates presented, what came back, and the date anyone from your firm last spoke to them.

Two capabilities separate a working client CRM from a directory. First, contact portability: when a hiring manager changes employer, the relationship should follow the person, because that person is a warm introduction at a new account rather than a lost record. Second, cooling signals: an account that billed well last year and has not been contacted in sixty days should appear on someone's list without anybody remembering to check.

Client-level analytics belong here too. Fill rate, submittal-to-interview conversion and time-to-first-submittal by client tell you which accounts deserve your strongest recruiters and which are consuming capacity at a loss. They are only computable when the client is a record rather than a text field on a job.

The useful applications in recruiting software are narrow and real. Parsing an incoming resume into structured fields removes the data entry that otherwise never happens. Scoring a candidate against a live requisition and showing which requirement drove each point turns a long list into a working shortlist. Searching your own database before you pay for new sourcing is the single highest-return use, because rediscovery is cheaper than acquisition. Transcribing a screening call and filing the summary on the record preserves information that used to live in a notebook.

What to distrust is a score with no explanation, because recruiters stop trusting it after the second bad shortlist and then you are paying for a feature nobody opens. Distrust automation that sends without a human in the loop, distrust any claim that a model can assess fit better than a competent recruiter, and read carefully anywhere a product implies a screening decision is being made rather than assisted.

There is also a governance dimension. Where an automated employment decision tool is in scope under rules such as New York City's Local Law 144, the product should record which candidates were scored and export the data an independent bias audit needs. Commissioning that audit and issuing candidate notice remain the employer's job, not the vendor's, and any vendor implying otherwise is overselling.

Almost every agency has a story about a CRM that was excellent for six months and unusable after two years. The cause is nearly always structural rather than behavioural. If a system permits every important attribute to be recorded as free text, the database becomes unsearchable no matter how disciplined the team is, because forty recruiters will write the same fact forty ways.

The preventive measures are unglamorous and they work. Typed custom fields rather than notes. Required fields at stage boundaries, so a candidate cannot be submitted without right-to-work captured. Coded dispositions from a fixed list rather than free-text rejection reasons. Duplicate detection that runs during import rather than as a clean-up project afterwards. Retention windows that flag dormant records for review instead of letting the database fill with people who left the industry.

Ask about all five during a trial. They are the least exciting part of any demo and the strongest predictor of whether the system will still be useful in year three.

You need recruiting software if your revenue depends on relationships that outlast a single vacancy: staffing agencies, recruitment firms, executive search practices, RPO providers and embedded talent teams that run requisitions for multiple business units. You probably do not need one if you are a small employer hiring a handful of people a year, where a lighter applicant tracking tool or your existing HR platform will do the job at a fraction of the cost.

Surhires is thirty-nine, forty-nine and ninety-nine US dollars per user per month, or thirty-two, forty-one and eighty-two billed annually, with a three-seat minimum and a fourteen-day trial that does not ask for a credit card. There is no free-forever tier. That pricing is provisional pending buyer validation, which means it may change before general availability.

What is shipping today includes AI matching, resume parsing, auto-shortlisting, assessments, offer letters, WhatsApp messaging, interview scheduling and the mobile recruiter app. The client portal, source-of-hire attribution, the Chrome sourcing extension and multi-board posting are in build. Every feature page carries its own status, and the status on the feature page is the one to believe.