Protocol Career

Consulting recruiting statistics and research

Counts, not claims. Our original findings on what firms publish about their own hiring, and the measured contents of this platform. Each figure carries the pages read, the date, and what it does not show.

Usage
30,000+
Practice sessions analysed
14,000+
Candidates reached
600+
Drills in the bank
400+
Prep guides published
Content, counted live
57
Full cases
34
Voice-ready cases
417
Guides and explainers
6
Case types
Case mix by firm style
McKinsey
19
Bain
16
BCG
15
Independent
4
Kearney
1
LEK
1
Deloitte
1
Case mix by type
Profitability
16
Growth
13
Operations
11
Market entry
8
M&A
7
Pricing
2
Outcome signals
2.4x
Higher offer-rate signal

Candidates applying to two or more firms, versus MBB-only applicants.

47%
Higher profitability pass rate

Candidates who completed the profitability track.

31%
Higher market-entry structure score

On overall structure, market-entry cases.

24%
Higher recommendation quality

On pricing cases.

These are directional signals from platform activity, not a controlled study. They compare candidates who completed a track with those who did not, so selection effects are not removed.

Original research

What consulting firms publish about their own hiring

These findings come from reading firms' own careers pages at scale and counting what is there. They are free to cite with attribution to Protocol Career. Each one carries the pages read, the date, and the limit of what it shows.

0 of 371read 30 September 2026

No consulting firm publishes a case interview conducted over the phone. Across 371 official pages read live, ten firms name the phone as an interview channel in eleven distinct published mentions, two publish a duration, and one publishes phone-specific guidance. One firm, Capital One, requires video for any remote interview instead.

How it was measured

532 official URLs from our own source registers were fetched on 30 September 2026; 371 returned a readable page. Those 371 were searched in full text for a case conducted by phone. Pages that returned an error or refused our connection were excluded and counted separately.

What it does not show

An absence in what firms publish, not an absence in what firms do. A firm may run a phone case without documenting it.

The full guide, with every source
10 of 19read 30 September 2026

Ten consulting firms of nineteen publish a number of interview rounds somewhere on their careers pages. Nine publish none, and three of the ten publish it only for a specific region or role rather than for the firm.

How it was measured

Nineteen firms' careers sections were read in full on 30 September 2026, following every page that describes the hiring process. A firm counts as publishing a number only where that number appears in its own text.

What it does not show

Counts published pages, not recruiter practice. Regional variation means a published number may not apply to the office you are interviewing with.

The full guide, with every source
2 days to 4 weeksread 30 September 2026

Published response times after a consulting interview range from two working days to four weeks, a factor of ten. Thirteen firms publish at least one figure, in sixteen distinct numbers, and one firm publishes four different figures across four of its own pages.

How it was measured

Every careers page in our source registers was searched for a published response or decision time on 30 September 2026. Four weeks is twenty working days, which is ten times two working days. Firms publishing dates rather than intervals were recorded separately.

What it does not show

These are the times firms publish, not the times candidates wait. We have no measurement of actual waits.

The full guide, with every source
17 firms, 18 instructionsread 30 September 2026

Seventeen firms publish guidance on the questions a candidate should ask, in eighteen distinct instructions. Only one publishes where in the interview to ask them, only one names a subject on which the candidate has a stated right of reply, and only one publishes a limit on what its team will answer.

How it was measured

Careers and interview-preparation pages of every firm in our registers, read on 30 September 2026. One firm is counted twice because two of its pages give different advice on the same point.

What it does not show

Published guidance only. It says nothing about how an interviewer weighs the questions a candidate actually asks.

The full guide, with every source
12 firms, 4 incompatible registersread 30 September 2026

Twelve firms publish a dress instruction for interviews, and they name four different and non-overlapping registers. Four more write 'professional' without naming a register, one turns the question back to the candidate, and one defers the answer to the invitation email.

How it was measured

The same 371 readable official pages, searched for dress guidance on 30 September 2026. A register counts as named only where the firm uses the term itself.

What it does not show

None of McKinsey, BCG, Bain or Oliver Wyman publishes any dress guidance at all, so the four registers describe the firms that speak, not the market.

The full guide, with every source
Methodology

Where each of these numbers comes from

Three different kinds of number sit on this page, and they do not carry the same weight. Reading them as one set would overstate the strongest and understate the most verifiable.

Cases, case mix, guides

Counted live from the content

Case counts, the case mix, and the guide count are read from the files at build time. They cannot drift from what ships, and they change when the content changes.

Sessions, candidates, drills

Reported from platform activity

Practice sessions, candidates reached and the drill bank are cumulative totals from product usage. They are reported figures, not derived from the repository.

Outcome signals

Comparisons between candidate groups

The outcome signals compare candidates who completed a track with those who did not. Nobody is assigned to a group, so candidates who finish a track may differ from those who abandon it in ways the comparison does not remove.

What this page does not claim

None of it is a controlled study. No candidate was randomly assigned, no result is attributed to the platform alone, and the outcome figures are not offer rates. A candidate who completes a track is usually also the candidate who prepares more elsewhere, and that effect is not separated here.

The scoring behind those tracks is deterministic and inspectable: the six rubrics and their weights are read from the code that grades candidates. If a figure here is wrong or out of date, tell us and we will correct it and date the change.

See for yourself

The case library is free to browse, and the tools open without an account.