A hiring process audit reviews every step from job ad to signed offer. It shows where candidates drop out, how interviewers score and what a real applicant meets. SmartHiring delivers it as a written report in week 1. Below is one in full. The company is a composite of loops I have audited. Names and numbers are changed.
No pitch if the loop is already fine.
SmartHiring's hiring process audit is one written report in six parts, delivered in week 1. A summary ranks the 3 leaks by cost. The interview process review and the scorecard and loop audit follow. In the mystery application, a CV for your job ad goes through your real pipeline. The top gaps and a 4-week action plan close it.
For the audit, SmartHiring needs the job ad, the scorecards and 3 recent debrief notes. An ATS export of the last 90 days shows where candidates drop out. With the CTO, Elman Huseynov writes the CV the job ad asks for. If your screen rejects it, that is finding number 1.
Example report
In 90 days the example fintech's loop saw 212 applicants and ran 31 first rounds. It made 3 offers and closed 1. The 3 places it lost most are ranked below by what they cost, with the section that holds the evidence.
| # | Leak | Evidence | Cost in 90 days |
|---|---|---|---|
| 1 | The CV screen rejects the candidate the ad describes | Mystery application rejected in 3 days, section 4 | The 2 strongest profiles in the ATS never reached a human |
| 2 | Two interviewers, same candidate, 2 levels apart | 4 of 6 debriefs, section 3 | 2 offers argued over for a week, 1 lost to a faster company |
| 3 | The take-home takes 9 hours and decides nothing | Round 3 drop-off 41%, section 2 | 13 candidates left before a human saw their code |
The example fintech's loop is shown round by round, with drop-off from the ATS export. A round without a scorecard is marked, because a round nobody scores is a round nobody can defend in the debrief.
| Round | Owner | Length | Scorecard | Candidates in | Drop-off after |
|---|---|---|---|---|---|
| CV screen | Recruiter, keyword filter in the ATS | ~2 min | No | 212 | 85% (180 rejected) |
| 1. Recruiter call | Recruiter | 30 min | No | 32 | 3% |
| 2. Hiring manager | 1 of 2 EMs | 45 min | Free text | 31 | 29% |
| 3. Take-home | Candidate, then 1 engineer | 9 h median | Pass/fail | 22 | 41% (9 never submitted) |
| 4. System design | 2 engineers | 60 min | 1 to 5, no anchors | 13 | 23% |
| 5. Founder | CTO | 30 min | No | 10 | 3 offers, 1 accepted |
Reading. 5 rounds, 3 of them without a scorecard. The round that removes the most candidates after the screen is the take-home, and it removes them by attrition: 9 of the 13 lost never submitted. The 2 who reached the founder round and declined both wrote "timeline" in the ATS notes. Screen to offer took 34 days median; the offers that win in this market land in 14.
I read the example fintech's 3 scorecards and 3 debrief notes. All 6 cover the same 3 candidates. The ratings can then be set against the decision.
The panel does not disagree about candidates. It disagrees about the bar, because the bar is not written down anywhere.
With the example fintech's CTO I wrote the CV of the engineer the job ad asks for: 6 years backend, Go and Postgres in production, one system carried through a scaling incident, no Kubernetes line because the ad calls it "nice to have". It went through the public careers page on a Tuesday under an agreed name. Only the CTO knew.
This is leak number 1 because it sits before every other round and costs almost nothing to fix: the filter is 2 checkboxes in the ATS.
| Gap | Fix | Effort | When |
|---|---|---|---|
| ATS filter rejects the ad's own profile | Must-haves in the filter match the must-haves in the ad; the recruiter reads every CV that passes | 1 hour | Now |
| Round 4 ratings 2 levels apart | Anchored scorecard per role: what strong, mixed and weak evidence looks like on each line; ratings locked before the debrief | Half-day workshop | Week 2 |
| Debrief decided by the opening line | Ratings read out first, lowest rater speaks first, hiring manager decides last | One meeting rule | Week 2 |
| Take-home loses 41% | A 90-minute paired session on a bug in a repo like theirs; AI allowed, reasoning observed | 1 day to build | Week 3 |
| Round 2 free text | Same anchored scorecard, hiring-manager lines only | Covered by the workshop | Week 2 |
| Rejection feedback is a template | One specific gap from the scorecard, sent by the hiring manager within 2 days | 10 minutes per candidate | Week 3 |
Measured in 90 days: panel agreement within 1 level, screen-to-offer days, paired-session drop-off, offers accepted. The same 4 numbers from the same ATS export, so before and after compare.
This report is an example. The company is a composite of engineering teams I have audited; the name, dates and every number are changed. Your report is written about your loop, from your scorecards, debriefs and ATS export.
Start with a free 30-minute call with Elman Huseynov. Tell him how your team interviews today. He will tell you where the loop loses signal and what to fix first. He will also say whether the audit, the workshop or calibration fits your stage.
No pitch if the loop is already fine.