The firms rebuilt the door in front of the interview room. Between 2023 and 2026 they put named online assessments on their own careers pages, staggered and deferred MBA start dates when demand cooled, and invited select candidates into chatbot cases that look nothing like a partner across a table. Candidates rebuilt prep around those gates: more screen time, more AI tools, fewer paid hours with a human coach. The score sheet in the live case did not change.
That is the split most guides miss. Clearing Solve, Sova, the Consulting Career Assessment, or an online case is necessary and new. It is not the interview. The partner still marks structure, math said aloud, exhibit judgment, and a clean close. If you only train for the gate, you arrive fluent in a chat window and mute in the room.
I should say my stake before the timeline. I still score live cases as a Big Four Strategy & Operations partner. I also built CoachNed because one expensive coach and a stack of casebooks does not scale when the funnel is this long. I have a commercial interest in AI tools that force you to speak. I am not a neutral observer of "the rise of AI prep."
What follows: what the firms publish, what named reporting documented about hiring, how prep shifted, and a small arithmetic exercise for how to spend your hours.
What the firms put on their careers pages
By mid-2026, the three most-cited strategy firms all describe a digital screen before or beside live cases. The names matter because invitation emails use them, and folklore invents others.
| Firm | Named screen (as of careers pages checked 2026-09-24) | What the firm states | Live interview they still describe |
|---|---|---|---|
| McKinsey | Solve | Gamified assessment of problem-solving; performance considered with the rest of the application; AI tools, notes, and recording barred except approved disability accommodations | Personal experience interview plus problem-solving interview for most client-facing roles (Interviewing) |
| Bain | Digital assessment via Sova or TestGorilla | About 30–40 minutes; mandatory when required; AI tools prohibited; calculator allowed on the assessment, not in later interviews | Case interviews for consulting roles after the digital step (Interviewing) |
| BCG (US Associate) | Consulting Career Assessment (CCA) plus, for select candidates, an online case (~35 minutes) | CCA within 30–90 minutes of apply, ~30–35 minutes, 48-hour window; online case for a select group ~2 weeks after the deadline | Case study interview: structure, questions, data, calculations, clear reasoning (Case interview preparation) |
McKinsey's Solve page is explicit about integrity: no AI tools, no screenshots, no help from others, with consequences that include exclusion and rescinding offers. Bain's digital-assessment FAQ says the same in one line: AI tools are strictly prohibited. BCG's US Associate page also answers "Does BCG use AI in its hiring process?" by saying technology enhances, does not replace, human decisions. Read that carefully. The firms are experimenting with AI-shaped screens. They are not handing the offer decision to a model.
Office variation is real. BCG Mexico describes an Online Case Experience of about 30 minutes plus a CCA of about 30 minutes before two rounds of case interviews (BCG Mexico). Treat the invitation as the source of truth for your office.
Deferred starts and thinner classes: what named reporting showed
The assessment stack grew while the class sizes and start calendars flexed. That is not folklore; it is dated press and school data.
In February 2023, the Wall Street Journal reported that McKinsey intended to eliminate as many as 2,000 jobs, largely in support roles (WSJ, 21 Feb 2023). In April 2023, the same paper's Lindsay Ellis reporting — carried widely, including by Fox Business — described McKinsey, Bain, and BCG delaying some MBA start dates into 2024, with Bain offering cash incentives (up to $40,000 in one communication) for later starts (Fox Business / WSJ, 18 Apr 2023). In October 2023, Ellis again reported that tech, finance, and consulting were dialing back MBA offers on campus (WSJ, 16 Oct 2023).
School placement tables show the soft year in the share of acceptances, not in the salary line. Wharton's published industry placement series puts consulting at 28.8% of accepted offers for the Class of 2023, 25.1% for 2024, then 28.2% for 2025, with consulting median base salary stuck at $190,000 across those years (Wharton full-time industry choices). Harvard Business School reports 21% of the Class of 2025 seeking employment went into consulting, with a $190,000 median base (HBS consulting function, Class of 2025).
What that supports: demand for new consultants was choppy from 2023 into 2024, start dates moved, and campus recruiting felt thinner. What it does not support: a permanent collapse of consulting as a destination, or a claim that "MBB stopped hiring." Shares recovered at Wharton by 2025. Salaries did not fall. The practical effect for candidates was a longer, more selective funnel with an extra digital gate, not a new interview format.
The AI interview experiments that are real
Call them experiments, not the new standard.
BCG's named online case for select US Associate candidates is the clearest firm-published chatbot-adjacent step: about 35 minutes, after application, for a subset of the pool (US Associate opportunities). Other BCG markets describe an Online Case Experience of similar length. That is a structured case delivered through a digital interface. It is closer to a live case than Solve or Sova are. It is still not a partner who can interrupt you, change the prompt, or ask why you ignored a line on the exhibit.
McKinsey's experiment is different: Solve as a gamified screen, plus a hard ban on AI assistance during it. Bain's is Sova or TestGorilla as a cognitive and preference screen, also with an AI ban. None of these pages claim the live case is optional.
Candidate folklore often collapses "I got a chatbot" into "firms hire by AI now." The careers pages do not say that. BCG's own language is that hiring decisions remain with teams under human oversight. When you hear otherwise on forums, treat it as rumor until the firm names it.
How prep shifted: more screens, fewer coaches per candidate
Prep followed the gate.
Candidates spend hours on assessment simulators because the gate is timed and binary. That is rational. The second shift is quieter: fewer minutes of human coaching per candidate. Demand for practice rose when applications stayed high and classes felt selective; coach supply did not. My estimate, labeled as such: a candidate who would have bought eight to twelve human hours in 2019 now often buys two to four and fills the rest with AI tools, peer swaps, and typed drills. I cannot prove a market-wide series; CoachNed exists partly because that arithmetic is obvious from the demand side.
AI tools are useful for volume and for a scored debrief when no human is free. They are weak at two things the sheet still rewards: thinking under interruption, and judgment when the data is incomplete. A chatbot that never cuts you off will let you rehearse a monologue. I stop listening in the room when the candidate sounds like they are reading a prepared tree.
The honest division of labor in 2026:
| Prep surface | What it trains well | What it does not replace |
|---|---|---|
| Named assessment sims (Solve-style, Sova-style, CCA-style, online case) | Format familiarity, timing, calm under a clock | Spoken structure under pushback |
| Typed drills | Math speed, exhibit reads, structure drafts | Delivery and recovery mid-sentence |
| AI voice case | Full arc with a debrief when humans are scarce | A partner's taste and the social pressure of the room |
| Human coach or sharp peer | Interruption, judgment calls, PEI follow-ups | Unlimited cheap volume |
Use all four. Do not confuse the one that feels modern with the one that decides the offer.
What did not change: the live score sheet
BCG's published case-prep page still lists the same job: structure the approach, ask thoughtful questions, analyze data, do quick calculations, communicate clearly; quality of reasoning over a single right answer (BCG case interview preparation). McKinsey still pairs a personal experience interview with a problem-solving interview. Bain still runs case interviews after the digital assessment.
The boxes I fill in have not been rewritten for the chatbot era. Opening restatement. Useful clarifying questions. A structure built for this question. Math set up aloud before digits. An exhibit read for the one line that matters. A close that answers first. Behavioral examples that survive four follow-ups. That sheet is older than Solve, older than Sova, and older than Casey-style interfaces. Candidates who treat 2026 as "the year of the online assessment" and under-train the spoken case are optimizing the wrong bottleneck.
Ned's rule. Budget for two exams. Spend enough hours to clear the named gate on your invitation; spend most of the remaining hours speaking a full case aloud until a stranger can score you without generosity. The gate is new. The sheet is not.
Worked example: sixty hours, two wrong budgets
Take a candidate with sixty focused hours before first-round interviews, an invitation that names a digital assessment and then two rounds of cases, and no spare cash for a coach every week.
Budget A (gate-obsessed). Forty hours on assessment sims and forum write-ups. Fifteen hours of typed drills. Five hours of one spoken case with a friend who never interrupts. Result: calm on the screen, brittle in the room. The first partner who asks "why that branch?" watches the structure collapse.
Budget B (sheet-first, gate-sufficient). Twelve hours on the named assessment until timing is boring. Thirty-six hours of spoken cases and matched drills (structure, math, exhibits, synthesis), at least half of them out loud. Twelve hours of PEI stories with follow-ups. Result: the digital step is a formality you have seen before; the live cases are the product you actually ship.
Reproduce the split on a spreadsheet if you like. Assign each hour to "gate" or "sheet." If gate exceeds one-quarter of the total before you have twelve spoken cases on the calendar, you are over-fitting the screen.
A second small calculation, on coach scarcity. Suppose a strong human coach sits four cases a week at $200 an hour for a 90-minute block ($300 per case). One hundred candidates each wanting eight cases need 800 slots — fifty coach-weeks. The cohort does not get it. AI and peers fill volume; humans should fill the quality check. That is why I built a product that scores spoken cases, and why I still tell you to find one human who will be rude to you before final round.
Practice this week
Do one full spoken case with a timer and a recording, then one timed pass on the assessment your invitation names.
- Run the CopperCup voice interview (~18 minutes, seven-score debrief) or a peer case with the phone recording. Listen once without pausing. Mark the minute you went quiet, the minute the structure drifted, and the close.
- On the assessment side, use the matching simulator only: McKinsey Solve practice and Red Rock, BCG CCA, Bain Sova-style, or the Casey-style chatbot case. Stop after one honest timed attempt.
- Put the next five calendar slots on spoken work, not another assessment video.
Everything on CoachNed is open for seven days, no card; then $120 for a recruiting season or $49 a month. CoachNed is not affiliated with McKinsey, BCG, Bain, Deloitte, or any firm named above.
Frequently asked questions
Did consulting firms replace case interviews with AI in 2026?
No. Firm careers pages still describe live problem-solving or case interviews after digital screens. BCG states hiring decisions remain with teams under human oversight. Chatbot online cases and gamified assessments are additional gates, not substitutes for the partner interview.
Which online assessments should I prepare for?
Only the ones named in your invitation. McKinsey commonly uses Solve; Bain may assign Sova or TestGorilla; BCG US Associate candidates typically face the CCA and, if selected, an online case of about 35 minutes. Office pages differ. Prepare the named tool; do not prepare every rumor.
Why did MBA start dates get deferred in 2023?
Named reporting in April 2023 described McKinsey, Bain, and BCG staggering or delaying some MBA starts into 2024 amid softer demand, with Bain offering cash incentives for later starts. That was a capacity and demand response, not a change to interview scoring. Check your offer letter and recruiter for your class year; do not assume 2023 patterns apply unchanged.
Is AI practice enough if I cannot afford a coach?
It is enough for volume and for a scored baseline. It is not enough if nobody ever interrupts you. Use AI for reps; use one skeptical human or a harsh peer for at least a few cases before finals. Mix assessment sims with spoken cases so you do not overfit the gate.
What still decides the offer?
The live score sheet: structure, quantitative reasoning said aloud, exhibit judgment, communication, and behavioral examples that survive follow-ups. Clear the digital gate, then treat the room as the exam that pays.
Sources
- Solve, McKinsey's assessment game — firm description of Solve and AI/integrity rules. Checked 2026-09-24.
- Interviewing at McKinsey — PEI and problem-solving interview framing; assessments may include a test or game. Checked 2026-09-24.
- Bain digital assessment — Sova or TestGorilla, timing, AI prohibition, calculator rules. Checked 2026-09-24.
- Bain interviewing — digital assessment then case interviews for consulting. Checked 2026-09-24.
- BCG US Associate job opportunities — CCA timing, select online case (~35 min), AI-in-hiring statement. Checked 2026-09-24.
- BCG case interview preparation — live case expectations unchanged in published guidance. Checked 2026-09-24.
- BCG Mexico careers — OCE and CCA described for that market. Checked 2026-09-24.
- WSJ: McKinsey intends to eliminate as many as 2,000 jobs — Feb 2023 headcount reduction reporting. Checked 2026-09-24.
- Fox Business (WSJ Lindsay Ellis): McKinsey, Bain delay some M.B.A. start dates to 2024 — April 2023 deferred starts and Bain incentives. Checked 2026-09-24.
- WSJ: M.B.A. job offers in short supply — Oct 2023 campus recruiting pullback. Checked 2026-09-24.
- Wharton full-time industry choices — consulting share of acceptances 2023–2025 and $190k median. Checked 2026-09-24.
- HBS consulting function, Class of 2025 — 21% into consulting, $190k median base. Checked 2026-09-24.