Case study: how a Bordeaux HR firm cut CV screening time by 70% in 3 weeks

The starting point

Maëlle runs an independent HR recruitment firm in Bordeaux. Eight colleagues, a client base of regional SMBs between 50 and 200 employees, specialising in middle management roles in logistics, industry and B2B services.

A recurring problem: for each open position, the firm received between 120 and 180 applications. Initial screening (reading CVs, checking criteria, ranking by relevance) took two consultants between 4 and 6 hours per position. With fifteen active positions running simultaneously, that was between 60 and 90 hours wasted each week on a mechanical task.

"We were spending more time sorting PDFs than calling candidates. For a relationship-driven business, that's absurd." Maëlle, firm director.

The stack deployed: Claude + Airtable + Make

After a 2-hour diagnostic session, three tools were identified to cover the entire workflow:

Airtable serves as a centralised database. Each position becomes a filtered view. CVs received by email are automatically captured via a dedicated address per role.

Make orchestrates the flows. As soon as a CV arrives in the dedicated inbox, Make extracts the document text, formats it and submits it to Claude.

Claude via API analyses each CV against 5 weighted criteria from the job description, generates a score from 0 to 10, and writes a structured 3-line summary: strengths, gaps, recommendation.

Total monthly cost: €47 (Make Essential + Airtable Team + Claude API usage-based).

The 5-step workflow

1. The candidate sends their CV to the position's dedicated email address
2. Make detects the incoming email and extracts the CV content (PDF or Word)
3. Claude analyses the CV, assigns a score and generates the structured summary
4. The result is automatically created in Airtable with score, summary and original CV
5. The consultant accesses a view sorted by score and focuses on profiles scoring ≥ 7/10

Screening 120 applications drops from 5 hours to 45 minutes. The consultant validates, adjusts as needed, and frees up their schedule for interviews.

Results after 3 weeks

The system was tested on 4 positions in week 1, then deployed across the full active portfolio from week 2 onwards.

• −70% time on initial screening: 63 hours saved in the first three weeks
• 94% validation rate: candidates flagged "to contact" by the AI were confirmed by consultants in 94% of cases
• Time to first contact: down from 4.2 days to 1.1 days after CV receipt
• Unexpected effect: 3 client companies spontaneously noted the improved responsiveness without being told anything

Only one adjustment was needed in week 2: refining the Claude prompt for highly technical positions where certain industry acronyms were occasionally misinterpreted.

What this case study teaches

The value is not in removing human judgement, consultants remain the decision-makers. It lies in eliminating the mechanical work that precedes that judgement.

A significant secondary effect: standardisation. Before AI, each consultant applied their own implicit screening criteria. Now, the same 5 weighted criteria apply to every position, improving consistency and simplifying internal handovers.

In the IMPACT methodology, this automation corresponds to the Implementation phase: identify the highest-volume, most repetitive task and automate it first. The time freed is reinvested in client relationships and direct sourcing, where real value is created.

To identify the 3 to 5 most profitable automations for your firm or SMB, the TransformAudit produces an operational roadmap in a single session.