Chapter 3. Funnel, Statuses and Operational Discipline

How to use this chapter

A working scenario. At a vacancy review meeting everyone says "candidates exist." The recruiter sees ten cards, the manager remembers three people, the agency says it is waiting for feedback, and the HR director asks why there is still no offer. After half an hour it turns out: two candidates have already moved to other processes, one has been waiting a week for an interview, four have no rejection reason, and the finalist has no next action assigned.

What actually broke here: the funnel existed as a list of people, but not as a management

model. The team could not reliably answer how many candidates were genuinely in process, where they were waiting, who needed to take the next step, why people were dropping out, where SLA was being breached, and what data could be used for decisions.

The hiring funnel is not a pretty diagram for a report. It is the operational model of candidate movement. This chapter does not teach sourcing in depth — sources, outreach, search strings and channel strategy belong to a separate systematic sourcing methodology. Here sources appear only as an analytical layer: where candidates come from and how source quality shows up in conversion. If the funnel shows a role mismatch at later stages, you often need to go back to Chapter 2 and refine the criteria.

Remember in one sentence: the funnel works only when every candidate has a clear position, a reason

for their status, and a next step.

Rule Why it matters How to verify

Every active candidate has a next Otherwise the process is Weekly list of empty / overdue next

action unmanageable actions

Stage changes on the day of decision ATS reflects reality Candidate movement review

Rejection reason is mandatory Otherwise you cannot improve the Rejection report completeness

process

"Other" requires a comment Otherwise the reason taxonomy Reason taxonomy audit

degrades

Rule Why it matters How to verify

Notes separate fact / assessment / Bias and noise are reduced Random notes audit

decision

Consent / privacy status is checked Privacy risk is reduced PD consent / storage review

before contact / storage

Data hygiene must not be punitive. Its purpose is to preserve manageability and respect for the candidate. If the fields are too complex and recruiters cannot fill them in on time, the problem is in the process design, not only in people's discipline.

The most sustainable way to maintain hygiene is to make "dirty data" visible in a short data-cleanup rhythm. There is no need to wait for a big audit. Once a day or every other day, the recruiter looks only at exceptions: overdue, empty next action, missing reason, stalled on pause, missing consent status.

This takes less time than restoring the process before a report.

Exception How to find it What to do When to escalate

Empty next action field Active candidates without a next Assign an action or close If the owner is unknown

action the process

Overdue feedback Interview stage beyond SLA Remind interviewer / HM, If overdue 2+ working days

record the risk

Stale on pause On pause without a review date Confirm pause, move If pause linked to business

or older than rule forward or close freeze

Missing rejection reason Closed candidates without a Add an approved reason and If the taxonomy does not cover

reason assessment facts the situation

Weak reason "Other", "not a fit", "did not Rewrite via criterion or Decision may pose fair

like them" candidate decision assessment risk

Missing privacy status Contact scheduled but consent Contact is paused until If candidate data has already

or status missing verification been shared

Data hygiene must also account for different levels of detail. For executive search there may be fewer candidates but more decision notes and confidentiality. For high-volume hiring there may be more automated reminders, but the reasons and status taxonomy must be especially strict because errors multiply across hundreds of cards.

Hiring type What is especially important What not to over-complicate

Executive / confidential Access control, decision log, Too many public statuses

client / candidate confidentiality, cautious notes

High-volume Clear reasons, bulk data hygiene Long free-text comments without

checks, fast SLA, fair assessment taxonomy

review

Technical roles Case status, reviewer SLA, assessment Minor sub-stages of every action

facts, reviewer notes

Agency retained search Client status, presentation Consent, shortlist assessment Reporting without real candidate

facts state

In-house mobility Current manager timelines, transition Backroom statuses without transparent

status, fair assessment notes rules

6. Funnel metrics

Funnel metrics help find a bottleneck but do not automatically explain it. Low conversion after screening can mean weak sourcing, rigid criteria, a poor pitch, an inappropriate compensation range, or normal filtering. Therefore every metric should lead to a question, not an immediate blame.

Metric Formula What it shows Follow-up question

Stage conversion Moved to next stage / entered Is this quality, criteria, market

stage or process?

Time at stage Exit date – entry date Who is accountable for the delay?

SLA breach rate Overdue candidates / candidates Is the issue systemic or a one-off?

with SLA

Candidate drop-off share Candidate withdrew / active Where is candidate interest lost? Communication, timelines, offer,

candidates motivation?

Rejection reason share Reason count / total rejections Is the reason specific enough?

Source-to-stage conversion Candidates from source reaching Is source quality low for this Not deep-diving into sourcing here,

stage / candidates from source role family but seeing the signal

Offer acceptance Accepted / extended Final alignment match Pre-offer alignment, comp,

timelines, competing offers?

Minimum dashboard for the weekly debrief:

Block Indicator

Funnel health Candidates by stage by vacancy priority

Speed Overdue candidates and average time in key stages

Quality Conversion by stage and rejection reasons

Accountability SLA breaches by owner group

Candidate experience Candidates waiting response, candidate drop-offs

7. HarmonyATS example

Using HarmonyATS as an example, the funnel can be measured through several reports. This is a neutral example: the method applies to any ATS with stages, statuses, reasons, SLA and reports.

Report / object What to use it for What decision does it Caution

support

Vacancy report Distribution of candidates across Priorities, workload, funnel Do not compare roles without

active vacancies and stages health accounting for complexity

Funnel with details Conversions, bottlenecks, Where to improve screening, Check the sample size

losses by stage interview, offer, communication

Candidate movement Actual moves, recruiter Activity does not equal

activity, stage-to-stage movement quality

Best practice: review reports together. For example, Funnel with details shows a drop at HM interview, the rejection report shows a role mismatch, the SLA report shows feedback delays. Together this may mean that screening criteria do not match the hiring manager's expectations and feedback arrives late. Example of linked analysis:

This analysis is more useful than a blanket "the funnel is bad". It turns a report into a specific question: which stage, which reason, which owner, which action. If data is sparse, do not pretend the report proves a trend. For a small sample, use "signal to investigate" rather than "problem proven". For example, two rejections on compensation for an executive role may be an important signal, but not statistics. Ten similar screening rejections on a high-volume role are closer to a process pattern. In both cases it is useful to look not only at the number but also at specific cards, notes, timelines, candidate communication and operational context. This is how the team preserves accuracy, trust in data and speed of management decisions.

8. Personal data, fair assessment and ethics of notes / reasons

The funnel contains not only operational data but also personal data. Notes and rejection reasons can become a source of legal and ethical risk if they include personal traits, rumours, age hints, family circumstances, health, or subjective labels.

feedback arrives late.

Example of linked analysis:

Observation Report / data Possible hypothesis Verification

Many candidates in Screening, Funnel with details, screening Screening is too broad Read 10

few HM interviews notes or criteria are unclear rejected/proceeded CVs

HM interview conversion Funnel + Rejections Hiring manager is looking for Calibrate on rejected

low a different profile examples

Offer acceptance is low Offer outcomes, time-to-hire, Compensation, delay, weak Check motivation notes

candidate notes pre-offer alignment and competing offers

Source A gives many CV sources + stage Source quality is low for Do not deep-dive into

candidates but few cases conversion this role family sourcing here; pass the

signal to the sourcing debrief

SLA breaches at one stage SLA report Owner overloaded or Check calendar load

stage is redundant and stage value

This analysis is more useful than a blanket "the funnel is bad". It turns a report into a specific question: which

stage, which reason, which owner, which action.

If data is sparse, do not pretend the report proves a trend. For a small sample, use "signal to investigate"

rather than "problem proven". For example, two rejections on compensation for an executive role may be an

important signal, but not statistics. Ten similar screening rejections on a high-volume role are closer to a process

pattern. In both cases it is useful to look not only at the number but also at specific cards, notes, timelines,

candidate communication and operational context.

This is how the team preserves accuracy, trust in data and speed of management decisions.

8. Personal data, fair assessment

and ethics of notes / reasons

The funnel contains not only operational data but also personal data. Notes and rejection reasons can become a source of legal and ethical risk if they include personal traits, rumours, age hints, family circumstances, health, or subjective labels.

10. When to use and when not to use

Use Why

Regular hiring Without a funnel you cannot see bottlenecks

High-volume roles Small taxonomy errors scale

Executive / senior roles Decision log and privacy discipline needed

Agency delivery Client status and feedback SLA needed

Distributed teams Single source of truth needed

Do not over-complicate Why What to do instead

Very small one-off hire 20 statuses are not needed 4–5 stages, next action, reason

Process without clear criteria Funnel will show noise Go back to Chapter 2 and refine the role

Sourcing channel problem Funnel will show signals on sources, Go back to sourcing and channel strategy

but not fix sourcing strategy

11. SOP: funnel procedure fragments

SOP: stage movement

Event Rule

Candidate added Source, vacancy, privacy / consent status and next action filled in

Screening completed Stage updated, CV completed, reason noted on rejection

Interview completed Feedback / assessment scorecard completed within SLA

Candidate on pause Reason stated, owner and review date set

Offer sent Terms, deadline, motivation risks and owner documented

Process closed Rejection / rejection reason, status communication and archive / talent pool rule filled in

SOP: rejection reasons

Rule Example

Reason linked to a criterion or candidate decision Skills gap, compensation mismatch, candidate accepted

another offer

Comment is short and assessment-based "Did not demonstrate X in case Y"

"Other" used rarely and with comment "New reason needed: relocation policy mismatch"

Protected / sensitive data is not written No age, family, health, origin

Candidate-facing feedback is separated from internal notes Internal assessment facts are not turned into blunt

language

SOP: taxonomy management

The stage, status and reason taxonomy must have an owner. If every recruiter can add their own options, duplicates will appear within a month: "salary", "comp", "money", "too expensive", "did not meet salary requirements". If adding reasons is entirely forbidden, people will start writing everything in comments. A controlled process is needed.

Action Rule Owner

Add a new reason Only if existing reasons do not cover a Recruitment ops / lead

recurring situation

Delete a reason If the reason is unused or duplicates Recruitment ops

another

Merge reasons Preserve old-to-new value mapping Ops + analytics

Change a stage Check impact on historical reports Recruitment lead

Change SLA Agree with the owner group and Recruitment lead + HM / client

business expectations

Run an audit Monthly review of "Other", missing, Recruitment lead

stalled and biased notes

SOP: weekly funnel cleanup

The data-cleanup meeting differs from the recruitment debrief. The debrief answers "what is happening with hiring." Cleanup answers "are the data clean enough to trust".

Block Question Action

No next action Which active candidates do not have a Assign, close or pause with date

next step?

Overdue SLA Where is the overdue and who is Reminder or escalation

accountable?

Weak reasons Which rejection reasons are useless? Rewrite and update taxonomy if

needed

Stale stages Who has been at a stage too long Confirm status or close

without movement?

Privacy gaps Where is there no consent / status / Stop processing and run a review

storage clarity?

Report trust Which data cannot be used in a Mark caveat and fix source

dashboard?

12. Implementation plan

Horizon What to do Result

1 day List current stages / statuses / reasons Team sees the mixing of entities

and remove obvious duplicates

1 week Introduce stage rules, SLA and weekly Active candidates receive a next action

cleanup

1 month Review conversion, SLA, reasons and Funnel becomes an improvement tool

update taxonomy

1 quarter Connect the funnel debrief with economics Hiring is managed through business

and the operating rhythm impact

13. Maturity model

Level How it looks Next step

1. Chaos Candidates in chats, statuses are stale Make ATS the single source

of truth

2. Basic discipline Stages and reasons exist but notes have gaps Introduce next action and

SLA cleanup

3. Managed funnel Conversion, SLA, reasons reviewed weekly Improve criteria and

interview design

4. Data-driven management Bottlenecks linked to owners and actions Add economics and

quality review

5. AI-assisted hygiene AI helps find dirty data Strengthen privacy, fair

assessment and