Dataford · Difficulty Report
The Interview Difficulty Report 2026
We analyzed 913,000+ interview experiences to see which companies are genuinely hard to interview at, how difficulty is distributed, and why the hardest loops are often the ones candidates rate most highly. Here is what the data shows.
Summary · Key findings
Most interviews are not brutal. 57% of experiences land at “average”, and only 14% are rated difficult or very difficult. The horror stories are the tail of the distribution, not the middle of it.
Quant and trading firms are in their own league. Jane Street (55% rated hard), Citadel Securities (48%), and Two Sigma (41%) sit far above everyone else. If you are preparing for a trading firm, calibrate to a different bar.
The gentlest interviews are at the IT-services giants. IBM (11%) and Accenture (10%) sit at the bottom of the difficulty table — an order of magnitude below the quant firms, and a useful contrast to the brutal end of strategy consulting.
Hard and good are not opposites. Jane Street is the single hardest loop in the dataset and also one of the most positively rated, at 71% positive. A high bar that is clear and fair tends to earn respect, not resentment.
AI labs are the exception that proves it. As a cluster they are hard (32% rated difficult) and the least positively rated (39% positive) — the signature of a fast, high-bar process that is still maturing.
“How hard is it, really?” is the question candidates most want answered and the one most likely to be answered by a single scary anecdote.
Dataford collects interview experiences across thousands of companies — what people were asked, how hard it felt, and whether they would do it again. For this report we used 913,000 of them: 880,000 carry a difficulty rating and 889,000 carry an overall sentiment. That is enough to replace anecdotes with a distribution.
Difficulty here is candidate-perceived, not an objective scale, so we read it as how a loop feels from the inside. We normalized the ratings into five buckets and define a loop's hardness as the share of its experiences rated difficult or very difficult. The full method is at the end.
The shape of it
Most interviews sit in the middle
Before naming names, it helps to see the whole distribution. The average interview is, almost by definition, average — and the genuinely punishing tail is thin.
57% of interviews are rated “average”. Only 14% are difficult or very difficult, and under 2% are very difficult.
This is the first useful correction. The internet over-samples the very difficult slice on the right, because that is what people post about. For most candidates at most companies, the realistic expectation is a solid, average-difficulty loop — demanding, but not the gauntlet the forums describe.
Hardest
Where the bar is genuinely brutal
The hard tail is not spread evenly. It clusters in two places: quant and trading firms, and the model-layer companies. Jane Street stands alone at the top, with more than half of its loops rated hard.
Quant and trading firms occupy the extreme; Roblox and the AI-data companies (Databricks, Snowflake) follow.
The contrast at the other end is just as sharp. The large IT-services firms run the gentlest loops in the dataset — roughly a fifth as demanding as the firms at the top.
Strategy consulting splits the difference in a telling way. The case interviews at McKinsey and Boston Consulting Group rate among the hardest anywhere, while the IT-services arm of the same industry sits at the bottom. “Consulting” is not one difficulty; it is two.
Finding
Hard does not mean hated
The intuition is that the hardest interviews leave the worst taste. The data says the opposite is often true. Jane Street is the single hardest loop here and one of the most positively rated, at 71% positive.
Quant firms are hard yet respected. AI labs are the outlier: hard and the least positively rated.
The pattern is about fairness, not ease. A high bar that is clear, well-run, and obviously relevant to the job earns respect even when it is exhausting. AI labs are the cautionary case: a hard process that is still maturing, moving fast, and not yet consistent enough to feel fair. Difficulty is not the thing candidates resent — randomness is.
Outlook
What to do with this
Difficulty is information, and it should change how you prepare rather than whether you apply. The firms at the top of this chart are not unbeatable; they are predictable. A 55%-hard loop is a loop with a known, high standard, which is exactly the kind you can train for.
Match your preparation to the bar. For quant and trading, that means speed and mental math under pressure. For the model-layer companies, it means depth on systems and modeling. And for everyone else, take comfort in the distribution: most loops are average, and average is very preparable.
Train for the bar you are facing
Company-specific interview guides and a question bank that mirrors how each firm actually tests — so a hard loop becomes a known one.
Explore interview guidesFAQ
Frequently asked questions
What is the hardest company to interview at?+
Among companies with enough data to rank, Jane Street is the hardest: 55% of its interview experiences are rated difficult or very difficult. Other quant and trading firms — Citadel Securities (48%) and Two Sigma (41%) — follow closely, well above the rest of the field.
How hard is the average tech interview?+
Less brutal than its reputation. Across 880,000+ rated experiences, 57% land at “average” and only about 14% are rated difficult or very difficult. The genuinely punishing loops are concentrated in a handful of firms.
Are hard interviews a bad experience?+
Not necessarily. Jane Street is both the hardest loop in the data and one of the most positively rated, at 71% positive. A high bar that is clear and well-run tends to be respected. The exception is AI labs, which are hard and the least positively rated cluster.
Which companies have the easiest interviews?+
The large IT-services firms. IBM (11% rated hard) and Accenture (10%) sit at the bottom of the difficulty table — roughly a fifth as demanding as the quant firms at the top.
How was difficulty measured?+
From candidate-reported difficulty across 913,000+ interview experiences. We normalized the labels into five ordinal buckets (very easy to very difficult) and define “% hard” as the share rated difficult or very difficult. It reflects candidate perception, not an objective scale.
Methodology
How this report was built
This report draws on 913,906 interview experiences in Dataford's dataset. Of those, 880,285 carry a candidate-reported difficulty rating and 889,211 carry an overall experience sentiment. We normalized difficulty into five ordinal buckets — very easy, easy, average, difficult, very difficult — and define a company's “% hard” as the share of its experiences rated difficult or very difficult.
Company rankings include only firms with at least 150 rated experiences, to avoid small-sample noise, and use canonical company names. The archetype comparison groups representative firms into quant & trading, AI labs, Big Tech, and consulting & services; the consulting group mixes strategy and IT-services firms, which is why its average understates the strategy case interview.
Difficulty and sentiment are candidate perceptions, not objective measures, and reflect the people who chose to report. Figures are current as of June 2026.