Updated September 2026 · 11 min read · US multinational technology and consulting corporation
IBM acquired Kenexa in 2012, and for most of the decade that followed, "IBM Kenexa Assessments" was a genuine product sold to employers worldwide. A large body of prep content still treats that as current, and treats it as evidence of what IBM does to its own candidates.
IBM has since exited both halves of Kenexa. In January 2020, SHL announced a partnership under which Kenexa assessment content moved onto SHL's platform, with IBM's VP of Cognitive Applications quoted welcoming the transition. In April 2021, IBM's talent-acquisition suite — BrassRing, Lead Manager and Onboard — was sold to Infinite Computer Solutions.
So the company that once owned the era's biggest assessment vendor now buys assessment services rather than selling them. And critically: no evidence exists that IBM administers Kenexa-branded assessments to its own candidates today. If you are preparing from a guide built around "the IBM Kenexa test," you are preparing from a product history rather than from IBM's current hiring process.
What IBM does publish about its assessments is short, specific and considerably more useful. This guide works from that.
IBM publishes a five-step funnel on its careers site: Application → Screening → Online Assessment → Interview → Decision.
At screening, "IBMers with expertise in your field will review your application, skills and experience." At the interview stage, "you may be invited to schedule an interview (phone, video or in person) or an assessment center visit (group exercise and in-person interview)." IBM notes twice that the process "may vary by location and region," and that caveat does real work — regional variation at IBM is larger than at most employers.
IBM also publishes two role-specific sequences on its careers blog, and they differ in an important way.
Entry-level developer, published March 2025:
Entry-level consultant, published March 2025:
The difference is the fourth stage. Consultants get a case interview; developers get a technical interview. Consultants in many regions get a video assessment; developers get a coding assessment. Confirm which track you are on before you prepare, because the two paths diverge sharply after screening.
For experienced hires, IBM publishes no separate process. The generic five-step funnel applies, with assessments gated by role. In practice, assessment activity concentrates at entry level.
For apprenticeships and new-collar roles — IBM's routes for candidates without degrees, marketed under the line "No degree? No problem!" — IBM publishes the programmes but no distinct assessment sequence.
IBM names four assessment categories, and the specificity is unusual for a large employer:
Worth noting: IBM's assessment FAQ lists all four; its application-process page lists only three and omits work behaviours. Both pages are current. The inconsistency is IBM's, not a transcription error, and it is a reasonable indication that the work-behaviours instrument is not universal.
The work behaviours assessment is the one IBM describes most concretely, and it is a forced-choice format. IBM's guidance: when "asked to pick between two statements," choose "the statement that you agree with slightly more," and "try not to overthink your responses and go with your first reaction wherever possible." That is characteristic of forced-choice personality instruments, which are designed to resist the obvious strategy of picking whichever option sounds most employable — because both options often sound equally employable.
For the coding assessment, IBM does something few employers do: it lowers the difficulty expectation in public. Its entry-level developer post says "the problems won't be too difficult and will be on topics that you're learning in class," and then reframes what is being measured: "it's all about, how you got the answer, how you understood the problem, and why you wrote the program the way you wrote it." Topics named include data structures, networking and technology-specific data-handling tasks.
The competencies IBM says it looks for are communication, technical acumen, problem-solving and creativity. For consulting it adds "flexibility to adapt to different environments" and "collaboration and being a team player," with a pointed caveat attached: "we can only assess what we see and hear."
Coding assessment — IBM does not name a vendor. Candidate reports on Glassdoor consistently describe HackerRank for software engineering roles, including a September 2026 account of an assessment lasting 70 minutes with 2 array-based questions. Treat HackerRank as candidate-reported rather than confirmed.
Video assessment — vendor not named by IBM. A graduate consultant on Glassdoor described the mechanics as being asked 3 questions, with 1 minute to think and 3 minutes to respond to each. That timing is a single candidate report, not IBM policy.
English language assessment — text-based multiple choice, vendor not named.
Work behaviours assessment — forced-choice, vendor not named.
"IBM Cognify" and game-based cognitive tests — these appear widely on prep sites, often attributed to Revelian. IBM's own careers pages never mention gamified or cognitive-ability assessments, and no confirmation from IBM or from the named provider was found. Candidates on Indeed and Glassdoor do describe cognitive testing, and graduate consultants in the UK have reported gamified tests with problem-solving and numerical elements at the online stage. So something exists; what it is called and who supplies it is not established.
Two operational details IBM does publish, both directly useful:
You have 7 days. IBM's instruction is to "complete your assessments within 7 days of receiving the invite." Whether individual tests are timed within that window is left for you to check in your invitation.
You are being recorded. IBM's applicant privacy notice, effective February 2026, states that IBM stores "webcam photos of you completing the exercises" for coding assessments, and "may store the video recordings" of video assessments. This is not hidden — it is in the published privacy notice — but many candidates discover it only when the webcam permission prompt appears.
Software and data science. The coding assessment is the gate. Technical interview follows, then a behavioural interview with upline management. No case interview.
Consulting. Video assessment in certain regions, no coding assessment unless you are applying to a developer or data-scientist consulting role. A case interview at first round. In some regions, an assessment centre with a group exercise. IBM also flags travel expectations for consulting as high — as much as 90% of the time in its own description.
Research. IBM publishes no process for research roles. This is a genuine gap.
Graduate versus experienced. IBM's published guidance is entirely entry-level. Experienced tracks compress to screening plus interviews, with assessments applied selectively.
Regional. IBM flags region-dependence twice, and it is real. The video assessment is explicitly "only for certain regions." Assessment-centre and group-exercise reports cluster in UK and European graduate hiring. Indian campus hiring, as described in candidate aggregations, runs a longer online assessment covering data structures, computer-science fundamentals, debugging and occasional aptitude, followed by two or three technical rounds and an HR round — but the platform is not confirmed, and the timings are candidate-reported.
IBM confirms the format of its assessments but publishes no sample items. What follows are original practice questions written to match the formats IBM describes. They are not IBM questions.
Forced-choice, in the style of a work-behaviours assessment.
For each pair, select the statement you agree with slightly more. Both will often sound positive — that is the point.
There is no scoring key you can reverse-engineer here, and attempting to is counterproductive. IBM's own advice — first reaction, don't overthink — is genuinely the right strategy, because forced-choice instruments are built to detect inconsistency across many such pairs.
Coding practice, at the difficulty IBM describes.
Given an array of integers, return the index of the first element that is strictly greater than the average of all elements. Return -1 if no such element exists.
Then, before you write anything: state your approach out loud, say what you expect the time complexity to be, and name 2 edge cases you would test. IBM's stated interest is in "how you understood the problem, and why you wrote the program the way you wrote it" — so practising the narration matters as much as practising the solution.
English-language practice, text-based multiple choice.
Choose the option that best completes the sentence: The migration was delayed because the vendor's documentation ______ the new authentication requirements. (a) did not account for (b) had not been accounted (c) was not accounting (d) has not accounted to
Video-assessment practice.
Record yourself answering this in under 3 minutes, then watch it back: Describe a time you had to learn an unfamiliar technology quickly in order to finish something. Watch for pace, for whether you actually named a result, and for whether you spent more than 30 seconds on setup before reaching your own actions.
Candidate accounts of IBM's assessments are notably calmer than for most large employers, and the reason appears to be the difficulty calibration IBM describes publicly.
On Glassdoor in September 2026, a software-engineering candidate reported a HackerRank assessment lasting 70 minutes, containing 2 array-based questions. Another Glassdoor account from the same month described an internship assessment containing "one leetcode question and one where you search in the files and fix the error," plus "a 6 question recording with camera and mic assessment" — and their reaction to the video component is the most useful line in the set: "It felt good and more like stuff I would do as if I were working there."
A graduate consultant on Glassdoor described the video mechanics as 3 questions with 1 minute to think and 3 minutes to answer, and characterised the overall process as "straight forward," with testing that "mainly tested problem solving and numerical ability." A separate graduate consultant account was less relaxed: "Tests are hard, try to finish them if you can. Do lots of research on the company."
On Indeed, responses about IBM's cognitive assessment skew mild — "the assessment was fun" from a technical support candidate in 2020, "test was easy enough to do" from a manufacturing candidate in 2023, and "it is as fair as can be" from a customer-service candidate in 2021. One account manager's comment captures the most common complaint, which is volume rather than difficulty: "more tests than i knew of to get a joj."
UK graduate consultants have described an assessment centre running roughly half a day, combining a group discussion, logic tests, a one-to-one competency interview and a group pitch.
After the online assessment, IBM does not publish results, scores or bands to candidates, and does not publish a retake policy. If you are progressed, the recruiter screening interview follows.
After the recruiter screen, developers move to a technical interview; consultants move to a case interview.
After the final behavioural interview, IBM's published funnel ends at "Decision." No SLA is published.
Two things are worth knowing about IBM's handling of your data through this process. IBM's applicant privacy notice discloses that IBM "use[s] AI technology for the purpose of completing IBM pre-employment verification checks," with an explicit limiting principle: that AI "will not be used to substantially assist or replace discretionary employment decisions." IBM also commits to giving "supplemental notice" where "processing activities involve artificial intelligence."
That is a more precise commitment than most employers publish, and it cuts against a common misreading. IBM's CEO Arvind Krishna has spoken publicly, and widely reported, about IBM replacing hundreds of HR roles with AI — its AskHR agent automated a large majority of routine HR enquiries. That concerns internal HR operations, not candidate screening. Reporting it as "IBM screens candidates with AI" overstates what any source shows, and IBM's own privacy notice draws the opposite line for hiring decisions.
First, establish which track you are on. Developer and consultant paths diverge after screening. Preparing case-interview structure for a software role, or data structures for a consulting role, wastes the time you have.
Do not prepare from Kenexa-branded material. IBM exited that business in 2020 and 2021. Guides built around "the IBM Kenexa test" are not describing IBM's current hiring.
Take the difficulty calibration seriously. IBM says the coding problems "won't be too difficult" and are on topics "you're learning in class." Candidate reports support that. Grinding hard algorithmic problems is less valuable here than being fluent and articulate on standard ones.
Practise explaining your code, not just writing it. IBM's stated interest is in your reasoning. If you can solve a problem but cannot say why you chose that approach, you are prepared for the wrong half of the assessment.
For the work-behaviours assessment, answer quickly and honestly. Forced-choice instruments include consistency checks. Trying to construct an ideal candidate across dozens of item pairs is both hard and detectable.
Set up for the webcam. IBM stores webcam photos during coding assessments and may store video-interview recordings. Sit somewhere with plain background and decent light, and expect to be seen.
Use the full 7 days, but not the last hour of them. The window is generous by industry standards. Sit the assessment when you are alert, on a stable connection, not at 11pm on day 7.
Do not use AI during the assessment. IBM prohibits AI use "to complete assessments, technical tests, or coding challenges" while explicitly permitting it for CV improvement and interview preparation. That line is published and unambiguous.
No evidence supports it. IBM moved Kenexa assessment content to SHL in January 2020 and sold its talent-acquisition suite to Infinite Computer Solutions in April 2021. Prep guides built around an "IBM Kenexa test" are describing a product IBM no longer owns.
IBM names four types: coding assessments, video assessments, English language assessments, and work behaviours assessments. Its application-process page lists only the first three, so the work-behaviours instrument does not appear to be universal.
IBM instructs candidates to complete assessments within 7 days of receiving the invite. Whether individual tests are timed within that window is stated in your invitation.
IBM does not name a vendor. Multiple candidate reports on Glassdoor describe HackerRank for software engineering roles, including a September 2026 report of a 70-minute test with 2 array questions.
IBM's own guidance says "the problems won't be too difficult and will be on topics that you're learning in class," and emphasises that reasoning matters more than the answer alone. Candidate reports are broadly consistent with that.
Yes. IBM's applicant privacy notice, effective February 2026, states IBM stores webcam photos of candidates completing coding exercises and may store video-assessment recordings.
IBM discloses using AI for pre-employment verification checks, with a published commitment that AI "will not be used to substantially assist or replace discretionary employment decisions." Reports of IBM replacing HR staff with AI concern internal HR service operations, not candidate scoring.
No. IBM permits AI for CV improvement and interview preparation but explicitly prohibits using it to complete assessments, technical tests or coding challenges.
Yes, materially. Consultants face a case interview at first round and, in some regions, a recorded video assessment. Developers and data scientists face a coding assessment and a technical interview instead.
Use TestSolve to crack the IBM assessment
AI-powered answers and reasoning sent to your phone. Free first solve, no signup.
Try it free →See pricing