Part 2 of 2: From first pitch to forecast
Twenty vacancies at a 25% fill rate produce five placements. Ten at a 60% fill rate produce six. The second agency makes more placements from half the assignments, with fewer dead-end searches eating consultant time. Fees and costs decide the final profit, but the point stands: the goal isn’t always more vacancies. It’s more of the right ones.
In Part 1, we covered recruitment delivery. In this part, we cover winning clients, qualifying work, growing accounts and forecasting revenue.
Most agencies already hold the clues: which clients hire repeatedly, which roles they fill, where fees are concentrated and when an account is likely to buy again. But those clues are scattered across CRM and applicant tracking system (ATS) records, emails, notes and consultants’ heads. AI can connect them, if the data is reliable and you know what you’re asking.
As in Part 1, the biggest prize is the reasoning. Your systems record what happened: a meeting, a lost pitch, a stalled vacancy. They rarely record why. AI can prompt consultants to capture it, so the knowledge belongs to the business.
Stage 1: Market mapping and prospect identification
Finding new clients requires more than assembling company lists. AI can analyse public company data, hiring activity and your own CRM records to find organisations that look like your best clients and may be close to a hiring need.
Time saved: Less manual research and list building, and prospects outside consultants’ usual networks.
Up and running: 2 to 4 weeks for a pilot on one target segment. Longer if CRM integration or data enrichment is needed.
The catch: A bigger list isn’t a better one. Public data goes stale, and a job posting doesn’t always mean a vacancy you can win.
In practice: Start with four years of placement and vacancy data: specialism, client value, repeat business, fill rate and, where you have it, cost to fill. Where those overlap is your commercial sweet spot. AI finds similar organisations by sector, size and hiring pattern, then narrows the list using current hiring signals. Consultants check the evidence and judge relationship fit. For a lean agency, the target might be the next 20 genuinely relevant prospects, not hundreds of names. Measure qualified conversations and fees, not list size.
Stage 2: Business development outreach
AI can support research-led outreach by summarising a prospect’s business, picking out relevant hiring themes and drafting emails, call notes or meeting briefs. It can also help coordinate follow-up activity through the CRM, although reminders and follow-up sequences are better left to simple automation.
Time saved: Less preparation and more consistent follow-up, so consultants spend their time in genuine conversations.
Up and running: 2 to 4 weeks for a small pilot, if account records are usable and the proposition is agreed.
The catch: Personalisation isn’t relevance. AI outreach can sound convincing while resting on wrong facts, and high-volume automation can damage the reputation it’s trying to build. Data protection and marketing permissions still apply.
In practice: AI prepares a short account brief and a suggested angle based on verified business needs. The consultant decides whether it’s credible, personalises the message and records why the prospect engaged, or didn’t. Automate the reminders, not the relationship. Judge success by discovery meetings and qualified requirements, not email volume.
Stage 3: Client discovery and vacancy qualification
A vacancy isn’t automatically a good assignment. Consultants need the business problem behind the hire, the essential and negotiable requirements, who decides, urgency, salary range and the realistic chance of a placement. AI can prepare discovery questions, structure notes and flag missing information against an agreed qualification framework.
Time saved: More consistent discovery, faster notes and cleaner handovers between business development and delivery. Better briefs mean fewer failed searches.
Up and running: 3 to 6 weeks, depending on integration and how mature your qualification process is.
The catch: A neatly completed vacancy form isn’t a qualified vacancy. The real problem may be an unrealistic salary, an unclear decision-maker, competing agencies, a shifting spec or a half-committed client. AI can flag gaps. It can’t negotiate priorities or build trust.
In practice: Grade vacancies A, B or C based on client commitment, realistic requirements, candidate availability, fee potential, likely fill rate and cost to fill. Then challenge the brief. A requirement for five years’ sector experience may really be a proxy for a particular stakeholder-management skill. Once you know that, the candidate pool widens. Record the reasoning, so delivery teams can tell non-negotiables from preferences and learn which compromises have led to placements.
Stage 4: Proposals and pitches
AI can assemble a first draft from credentials, case studies, market information and pricing, suggest relevant examples and summarise the client’s priorities. Templates and approval workflows handle the routine production.
Time saved: Faster, more consistent responses to qualified opportunities, leaving more time for commercial strategy and client discussion.
Up and running: 2 to 5 weeks if approved templates and reference material exist. Longer if they don’t.
The catch: AI can write persuasively about a service you can’t deliver. Unverified case studies, unsupported claims and inconsistent pricing damage credibility and margin. Commercial terms need human sign-off.
In practice: Build an approved library of service descriptions, evidence and case studies. AI drafts, and the account owner checks deliverability, terms and the strength of the offer. When you win or lose, record why: price, delivery confidence, an incumbent supplier or something else. Measure conversion and profitability, not production speed.
Stage 5: Account management
In established accounts, knowledge is scattered across consultants’ relationships, meeting notes and placement history. AI can pull it into account briefs, surface patterns and help prepare client reviews.
Time saved: Better-prepared meetings, more consistent follow-up and a shared view of each relationship.
Up and running: 4 to 8 weeks where multiple data sources need connecting or cleaning. A simple account-brief pilot may be quicker.
The catch: CRM activity isn’t relationship strength. A client can look active while losing confidence in you, or look quiet while holding real potential. AI can’t read silence, frustration or an informal conversation.
In practice: Every account plan needs a category, an objective, an owner, agreed activity and measures of success. Combine recency, frequency and value with hiring patterns, fill rate, service issues and relationship context, and AI can flag clients likely to hire again, valuable accounts gone quiet and openings across service lines. The account score tells you where to look. The consultant works out what it means and what to do. Record why a client chose a supplier, changed a brief or cooled on you, so that knowledge survives account handovers. Automation does the noticing. People manage the relationship.
Stage 6: Retention and account growth
Existing clients bring repeat business, new service lines and introductions. They also bring concentration risk. A team can be flat out servicing one big account while its pipeline withers. AI can flag changes in hiring patterns and service usage early enough to act.
Time saved: Earlier sight of risks and opportunities, instead of waiting for the periodic review or relying on a consultant’s memory.
Up and running: 4 to 8 weeks for an account-health pilot, depending on the quality of historical data.
The catch: Fewer vacancies don’t necessarily mean an unhappy client, and more hiring doesn’t mean they’ll buy more from you. AI shows where to look, but it can’t tell you the cause of a relationship change without human context.
In practice: A client places permanent roles with you but buys contract recruitment elsewhere. AI flags the service gap. A consultant finds out whether it’s unmet demand, procurement rules or another supplier. Or picture a once-active client whose instructions fall away. AI pulls together recent delivery results, feedback and account notes, and the consultant discovers frustration with shortlist speed, not a hiring freeze. Fixing that may save the relationship. Recording it sharpens account intelligence.
Ask yourself: if your five largest clients disappeared tomorrow, could you name the next 20 most likely to buy?
Stage 7: Forecasting
A useful forecast looks beyond open vacancies to fill rate by client and role, likely fee, hiring patterns and cost to fill. AI identifies inconsistencies and stalled opportunities. Consultants supply the context on probability and timing.
Time saved: Less manual reporting and earlier warning of at-risk opportunities.
Up and running: 4 to 10 weeks, depending on data quality, historical outcomes and integration.
The catch: Patchy data creates false confidence. A model can rate an opportunity highly without knowing a hiring freeze is coming.
In practice: AI flags vacancies stuck longer than comparable assignments, weak conversion, or forecasts out of line with past results. Consultants explain the cause: budget approval, client hesitation, candidate availability or a changing brief. Each needs a different response, even when the pipeline status looks identical.
Where to start with AI for client-side recruitment
There’s no universal first step. It might be market mapping, vacancy qualification, account intelligence or forecasting. Whatever you choose should connect to delivery: better briefs improve sourcing, and delivery results shape retention.
Start with four questions:
Where are your fees concentrated?
Which clients and job types have the strongest fill rates?
Which clients are likely to hire next?
Where are consultants spending time on assignments unlikely to convert?
Then set a baseline, prioritise by impact, complexity and readiness, fix the data, pilot with consultant oversight and scale what works. At every key decision point (a vacancy accepted, a prospect prioritised, a proposal lost, an account flagged), have AI prompt a short explanation and file it where others can find it.
Don’t automate the relationship, automate the work around it.The goal is more of the right jobs, better client conversations and more profitable placements.
We help recruitment businesses find where AI and automation create real value across delivery and client work, from workflow and data assessment to tool evaluation and phased implementation. Contact us to discuss your AI implementation strategy.
Missed the previous article? Read Part 1 of 2: Where AI earns its keep in recruitment delivery, from candidate sourcing to start date.
