A CRM built for how recruiting actually works

• A search is not a sales deal. A B2B deal closes in weeks; a retained search runs 9 to 16 weeks from kickoff to accepted offer. Force brief, sourcing, longlist, shortlist, interviews, offer and placed into a Deal pipeline and you invent stages like "in delivery" that are really a second Closed-Won, which breaks every revenue report. BD and delivery need separate objects.
• A candidate is not a contact. A candidate has a function, level, current company, last comp band, location, work-permit status, last meaningful conversation, source, and a submission state on every active search. A flat People table cannot hold that without a dozen custom fields that go stale the moment they change jobs.
• Linear ATS does not fit retained search. An ATS is built around one posting, a queue of applicants and a linear funnel. Retained search is the opposite: a small number of high-value passive candidates kept warm for years and re-engaged when a relevant mandate opens. The workflow is cyclical, which is why most firms keep two tools and still lose context between them.
• The same person plays multiple roles. A founder you placed three years ago is now a buy-side hiring manager; a candidate from a closed search is the warm intro for the next one. Without many-to-many links between People, Companies and Searches, the network value of every prior placement evaporates.
• Investor and referral attribution. "Which searches did this fund send us this year" is a single filter if Investors are tracked as Companies linked to the Search — and a manual reconstruction in a system where every company is the same kind of entity.

• Companies — the accounts, split by relationship type: Client, Prospect, Investor, Portfolio, Past Client. Industry, headcount band, stage, account owner, primary contact, relationship status, lifetime fees billed. An Investor company links to every Search that partner has sent, so attribution is one filter away.
• People — identity across every role a contact has played, with a lifecycle status: Candidate, Active Buy-Side, Hiring Manager, Investor Partner, Past Client, Referral Source, Off-Limits. A linked-records view shows every Company, Search and Application they have touched. Candidates can stay here behind a Candidate Status field, or split into their own object when the firm wants stricter privacy on the bench; for most boutique firms one People object with a flag is enough.
• Deals — the BD pipeline only: Lead, Qualified, Proposal Sent, Verbal, Signed, Lost, with the source tagged Inbound, Outbound, Investor Referral, Past Client Referral or Event. Closed Won creates a Search and the Deal exits. The Deal never tracks delivery.
• Searches — a custom object, one record per active mandate: client, hiring manager, role title, function, level, location, fee structure and value, billing schedule, kickoff and target placement dates, status from Brief through Placed, drop reason, lead consultant, source investor, linked Slack channel.
• Applications — a custom object joining candidate to search, with per-shortlist status: submission date, current stage from Sourced through Placed, drop reason, last interaction. The same candidate sits on multiple Applications across multiple Searches without losing state on any of them.

The trap small firms fall into is cramming everything into one pipeline and losing visibility on both. Keeping them separate means the number on the BD dashboard is the number the partner actually cares about.

• BD pipeline (Deals): Lead, Qualified, Proposal Sent, Verbal, Signed. Won deals leave the board entirely and create a Search. Typical cycle is one to six weeks for a known client, four to twelve for a cold one.
• Search pipeline (Searches): Brief, Sourcing, Longlist, Shortlist, Interviews, Offer, Placed, Dropped, owned by the lead consultant. Drop reason is a multi-select rather than free text, so year-end pattern analysis actually works.
• Bench (People filtered to Candidate Status): not a pipeline but a status table, sorted by last meaningful conversation and filtered to Active Bench. The view a partner opens before any sourcing call.
• All three sit on the same Companies and People. A consultant opens an account and sees every Deal in flight, every Search in delivery, every candidate ever submitted to that client, and the people on both sides — one scroll, the whole relationship.

The activity-capture automation alone changes daily life. A two-partner firm running six to eight active searches recovers four to six hours a week that previously went to manual logging.

• Closed-Won creates a Search, inheriting client, hiring manager, fee value, kickoff date and lead consultant from the Deal. No copy-paste, no lost context.
• Email and calendar capture by default, attaching the right thread and meeting to the right Candidate, Search and Company. The single highest-leverage automation for a solo or two-partner firm — it stops the relationship living in your inbox.
• Outbound sequences flow back onto the right Person record, so reply rates surface per search rather than per sequence, which is what the consultant needs to know.
• A per-search Slack channel surfaces on the Search record, so anyone opens the Search and finds the conversation without hunting for the channel name.
• A stalled-search alert: a scheduled workflow scans Searches in Sourcing or Shortlist with no candidate movement in 10 days and posts to the partnership channel. Quiet searches do not become quiet client churn.
• Investor referral attribution: every Search inherits the source Investor from the Deal, and a view on the Investor record shows searches sourced, searches placed and lifetime fees attributable to that relationship.

The consultant Monday view: active Searches by stage with last activity, Searches in Sourcing or Shortlist with no movement in 10 days, Applications waiting on a client decision, and bench people not contacted in 30 days. Five minutes and the week starts. The firm-level dashboard: BD pipeline value by stage, active Searches by stage and days since kickoff, time-to-shortlist by consultant, placements per quarter against fee target, drop reasons by category.

Two AI pieces do real work here. A Research attribute on Companies auto-fills industry context, recent funding, team size, recent leadership hires and known stack when a new client lands, so the consultant opens a fresh prospect record and the briefing is already there — one attribute replacing the $300 to $400 a month enrichment tool most firms bolt on. And an agent scoped to the Search object scans the linked emails, calls and submission notes from the past 14 days and writes a one-paragraph status summary on the record, so the lead consultant reads the brief before a client check-in instead of reconstructing the search from memory.

The line: Attio is the relationship layer for a relationship-led firm. It tells the consultant who the clients are, what searches are in flight, which candidates are stuck, and where the next mandate is coming from. The ATS, the parser and the compliance tool handle execution that needs structured workflow at scale. A core build takes about 7 to 10 working days; the blocker is rarely Attio, it is the partnership agreeing what each search stage means.

• High-volume contingency staffing — 200 open reqs, 20 recruiters, automated CV parsing, compliance workflow, structured scorecards across thousands of candidates a month. Bullhorn, Recruiterflow and Loxo are built for this. If the bottleneck is throughput against a CV pile, keep the ATS.
• Public job boards and apply-flow. Attio has no native job-board posting or apply-form pipeline. If most of the pipeline is inbound applicants, the ATS still owns that workflow and Attio holds the relationship layer beside it.
• Resume parsing and CV ranking at scale. Attio stores a resume as an attachment; it does not parse, score or rank. Rarely matters for executive search, which is relationship-led, but a contingency desk running on parsed CV search needs a real index.
• Background checks and compliance. ID verification, right-to-work, structured reference logging and EEO reporting live in a dedicated tool, with status syncing back to the Application record.