Day: November 7, 2026

What the Saville Swift Analysis Aptitude test is Swift Analysis Aptitude is a 24-question, 18-minute cognitive ability test published by Saville Assessment. It is used to screen graduates, management trainees, professionals and managers, usually at the first sift of an application. The test blends three aptitudes into one sitting: verbal analysis, numerical analysis and diagrammatic analysis. Each gets eight questions and six minutes. That is roughly 45 seconds per question, which is why candidates describe it as the hardest part of the process. "Swift" is Saville's label for its combination tests — several reasoning areas measured in one short, unsupervised online session. It is the fast screening tool, not the deep assessment-centre test. Saville also sells the same four aptitudes as full-length standalone tests of 32 questions each. Key facts PublisherSaville Assessment (Willis Towers Watson) Total questions24 Total time18 minutes SectionsVerbal, numerical, diagrammatic analysis Per section8 questions in 6 minutes FormatMultiple choice, online, unsupervised DevicesDesktop, laptop and tablet Scoring1–10 sten scale against a norm group Typical useGraduate, professional and management hiring Average scoreAbout 61% of questions correct The averages are not guesswork. Saville's own Analysis Aptitude Range Technical Summary reports a mean of 60.67% correct across a sample of 95,337 test takers. Most published figures in this guide come from that document rather than from prep-site estimates. Who uses it, and when in the process Swift Analysis Aptitude sits at the online screening stage, before interviews and before any assessment centre. You will usually receive a link after submitting an application, with a few days to complete it in your own time. Saville Assessment publishes client case studies with BNP Paribas, Jaguar Land Rover and Ricoh, covering graduate and apprentice recruitment. Prep and careers sites also report its use at employers including Ernst & Young, easyJet, Fujitsu, Prudential, Nestlé and Camelot. The heaviest concentration is in banking, consultancy, engineering and manufacturing. Two practical points follow from where the test sits. It is a cut-off, not a conversation. At the sift stage employers set a minimum score and filter. A strong CV does not rescue a weak sten score, because nobody reads the CV until you clear the threshold. You may be tested twice. Saville supplies supervised follow-up versions of these tests with fixed content, intended for verification later in the process. If your unsupervised score cannot be reproduced under supervision, that is a problem. Practise to get faster, not to get help. Format and timing SectionQuestionsTimeAverage correct Verbal Analysis86 min63.6% Numerical Analysis86 min53.6% Diagrammatic Analysis86 min64.8% Total2418 min60.7% Numerical is the hardest section by a clear margin. Across 95,337 test takers, candidates averaged 53.6% correct on numerical against 64.8% on diagrammatic. If you have limited preparation time, that gap tells you where to spend it. The testlet structure Questions arrive in testlets — blocks of four questions that carry their own time limit. You cannot bank three minutes on an easy block and spend it on a hard one later. This is deliberate. Saville designed it so a single confusing question cannot wreck the whole section. It also removes the usual strategy of skimming the whole section first, then allocating time. You work block by block. Difficulty rises inside each testlet and across the test. The first block is drawn from easy items, the last from difficult ones. A sudden jump in difficulty near the end is the test working as designed, not a sign you are failing. Every candidate gets a different test Swift Analysis Aptitude uses linear-on-the-fly testing (LOFT). The system builds your test by drawing testlets from a large item bank, matching difficulty to each position in the sequence. Scores are then equated using Item Response Theory so that different test forms remain comparable. The consequence matters: there is no fixed question set to memorise. Sites selling "Saville Swift answer PDFs" are selling you answers to questions you will not be shown. Pattern familiarity transfers; specific answers do not. Section 1: Verbal Analysis Verbal Analysis measures whether you can judge a written claim strictly against a passage. Eight questions, six minutes, average score 63.6% correct. The passages are short business texts — a policy note, a market summary, an internal report. Question types mix True / False / Cannot Say judgements with main-idea and interpretation items. Every question has four options with exactly one correct answer. One rule governs the whole section: answer from the passage, never from what you know. A statement can be obviously true in the real world and still be the wrong answer, because the passage does not establish it. Worked example Meridian Logistics installed a route-optimisation system across its Northern depots in March. Over the following six months, average delivery times at those depots fell by 14% and fuel cost per kilometre fell by 6%. Meridian's Southern depots, still running the previous scheduling method, recorded no change in delivery times across the same period. Statement: The route-optimisation system reduced fuel costs at the Northern depots. A. True B. False C. Cannot say D. True, but only for Northern depots Answer: C. The passage reports that fuel costs fell after the system was installed. It never states the system caused the fall. The Southern depots act as a comparison for delivery times only — the text gives no Southern fuel figures at all, so the causal link is unsupported. Most candidates pick A. They read "installed in March" and "fuel costs fell" and supply the causal arrow themselves. That reflex is exactly what the section is built to catch. The three traps that cost the most marks Supplying causation. Two things happening in sequence is not one causing the other. Scope creep. A claim about one region, one quarter or one product does not extend to all of them. Degree words. "Significantly", "consistently", "most" and "rapidly" need the passage to define them. One data point never establishes a trend. When you are genuinely torn between True and Cannot Say, choose Cannot Say. The error candidates make under time pressure runs overwhelmingly in the direction of inferring too much. Section 2: Numerical Analysis Numerical Analysis is the section candidates score worst on — 53.6% correct on average, eleven points below diagrammatic. It presents tables and charts of business data and asks you to interpret them under time pressure. The arithmetic is not advanced. Percentages, ratios, percentage change, averages and simple rates cover almost everything. What makes it hard is that each question sits on a dense data table, and finding the right two numbers takes longer than the calculation itself. You are permitted a calculator on Saville aptitude tests. Use one, and use it well — on-screen calculators are slower than the physical one on your desk. Worked example Q3 sales performance, European regions (€000): RegionActualTarget Nordics1,8401,600 Benelux1,2751,500 DACH2,3402,080 Question: Which region exceeded its Q3 target by the largest percentage? A. Nordics B. Benelux C. DACH D. Nordics and DACH equally Answer: A. Nordics beat target by 240 on 1,600, which is 15%. DACH beat target by 260 on 2,080, which is 12.5%. Benelux missed by 225, or −15%. DACH has the larger surplus in euros. Nordics has the larger surplus in percentage terms. The question asks for percentage, and under time pressure plenty of candidates answer the question the data made easiest rather than the one on screen. How to protect the section Read the question before the table. Know which two figures you need, then go and get them. Check the units line every time. Thousands, millions, percentages and index values are swapped between questions deliberately. Estimate first, calculate second. If the options are 8%, 15%, 24% and 31%, a rough figure often eliminates three of them before you touch the calculator. Watch percentage of versus percentage change. They produce different answers and both usually appear among the options. Section 3: Diagrammatic Analysis Diagrammatic Analysis measures logical rule-following, not visual flair. Candidates average 64.8% correct, making it the highest-scoring section — but only for people who have seen the format before. Questions show inputs passing through operators — symbols that each apply one fixed transformation — to produce outputs. You are given worked paths that let you decode what each operator does, then asked to supply a missing piece. Saville's own documentation names three item types in this bank: Item typeWhat you are givenWhat you supply Identifying RulesOutput and operatorsThe missing input Comprehending ProcessesInput and outputThe missing operator Understanding Logical SequencesInput and operatorsThe resulting output The practical read: you will be asked to work forwards, backwards and from the middle out. Candidates who only drill forward-direction questions get caught by the other two. Worked example Two decoded paths: W X Y Z → [△] → Z X Y W W X Y Z → [○] → X W Z Y △ swaps the first and last items. ○ swaps the items within each pair. Question: P Q R S passes through △ then ○. What is the output? A. Q S P R B. S Q P R C. Q S R P D. P R Q S Answer: A. Apply △ to P Q R S: swap first and last, giving S Q R P. Apply ○ to S Q R P: swap within each pair, giving Q S then P R — so Q S P R. Option B is what you get if you forget the second operator. Option C is what you get if you apply ○ before △. Wrong options in this section are almost always built from a specific procedural slip, not from random alternatives. Method Decode operators on the simplest path first — ideally one where a single operator acts alone. State each rule in one short sentence before applying it. "Swaps first and last" is testable; a vague mental image is not. Verify the rule against every path it appears on. One confirming path is not proof. For missing-input questions, run the operators in reverse, then check by running your answer forwards. How it is scored Your raw answers are converted to a sten score from 1 to 10 and compared against a norm group, not against a pass mark. For Swift Analysis Aptitude the standard comparison group is Professionals & Managers — an international sample of over 13,000 people, per Saville's example score report. StenYou scored higher than 1about 1% 2about 5% 3about 10% 4about 25% 5about 40% 6about 60% 7about 75% 8about 90% 9about 95% 10about 99% Sten 5 and 6 straddle the average. Most graduate employers set their cut-off at sten 5 or 6; competitive schemes in banking and consulting often sit at 7 or above. Employers set their own thresholds and rarely publish them. The total score is what decides You receive a total score plus three sub-scores. Saville states plainly in its technical summary that the total is the decision-making score, and that sub-scores should not be used in isolation. The reason is reliability. The total score has an internal consistency of .83; the sub-scores range from .62 for verbal to .68 for numerical. Eight questions is simply too few to measure one aptitude precisely. For you as a candidate this is good news. A weak section does not sink you if the other two are strong — the sections are summed before anyone makes a decision. Pace is measured separately Saville's reports show an aptitude score and a pace score for each area. Pace compares how quickly you answered against the average response time for the same questions. It is reported alongside accuracy, not folded into it. So rushing does not buy you a better score. Answering 18 questions correctly beats answering 24 questions with 14 correct, and the report will show the assessor exactly which pattern you produced. What the score actually predicts Saville's validation study of 308 people found the total score correlated .54 with third-party ratings of workplace performance after correction for criterion unreliability. By sub-test, verbal contributed most (.48), then numerical (.34), then diagrammatic (.24). That is a respectable figure for a single 18-minute test, and it explains why employers weight it heavily at the sift. It also explains why verbal performance matters more to your profile than the section's length suggests. Swift Analysis versus the rest of the Saville range Candidates routinely prepare for the wrong Saville test. Check your invitation email for the exact test name before you buy or practise anything. TestMeasuresQuestionsTimeTypical level Swift Analysis AptitudeVerbal, numerical, diagrammatic2418 minGraduate, professional, manager Swift Executive AptitudeVerbal, numerical, abstract2818 minSenior manager, director Swift Analysis Verbal & NumericalVerbal, numerical3224 minGraduate, professional Swift Comprehension AptitudeVerbal, numerical, error checking—9.5–16 minEntry level, operational Verbal Analysis Aptitude (single)Verbal only3224 minFollow-up or assessment centre Numerical Analysis Aptitude (single)Numerical only3224 minFollow-up or assessment centre Diagrammatic Analysis Aptitude (single)Diagrammatic only3224 minFollow-up or assessment centre Abstract Reasoning Aptitude (single)Abstract only3216 minFollow-up or assessment centre The distinction that trips people up most is Analysis versus Executive. They share identical verbal and numerical items drawn from the same bank. The only difference is the third section: Analysis uses diagrammatic reasoning, Executive uses abstract reasoning. The two formats correlate at .72 and measure the same underlying construct, so preparation transfers well between them. But the question presentation differs, and walking into abstract series questions expecting operator flowcharts costs you the opening minute of a six-minute section. One more pairing to expect: Saville frequently sells Swift Analysis Aptitude alongside Work Strengths or the Saville Wave personality questionnaire. Those are separate instruments with no right answers, and no amount of aptitude practice prepares you for them — if your invitation includes one, prepare for it separately. A two-week preparation plan Preparation for Swift Analysis Aptitude is about building speed on formats you already understand. Nobody fails because they cannot calculate a percentage. They fail because they spend forty seconds finding the right row. Two weeks at 45 minutes a day is enough for most candidates. Front-load numerical, because that is where the average score is lowest and where improvement comes fastest. DaysFocusWhat you doTarget 1–2DiagnosticOne full 18-minute timed test, untimed review of every errorKnow your weakest section 3–5NumericalPercentage change, ratios, table navigation7 of 8 correct, untimed 6–7VerbalTrue/False/Cannot Say drills on causation and scope trapsUnder 40 sec per statement 8–9DiagrammaticOperators in all three directionsDecode any operator in 2 paths 10–11Mixed timingFull sections back to back, strict 6-minute limitsFinish each section 12–13Full simulationsTwo complete 18-minute tests, review in fullStable score across both 14Light reviewRe-read your error log only. No new materialRest What actually moves your score Keep an error log. For every wrong answer write one line: what the question asked, what you did, what you should have done. Patterns appear within forty questions and they are usually the same three mistakes. Practise untimed before timed. Speed built on a shaky method just produces faster errors. Get accurate, then compress. Drill the format, not the questions. Because the test draws from an item bank, familiarity with question structure is the only thing that transfers. Train on screen. You will sit the real test on a screen with an on-screen clock. Practising on paper trains the wrong motor habits. Set up your desk the way you will sit the test. Physical calculator, scrap paper, pen, no phone. Rehearse the environment, not just the content. Test day The test is unsupervised and online, so you choose your conditions. That is an advantage most candidates waste. Before you start Sit it at the time of day your head works best, not the evening you receive the link. Use a laptop or desktop with a wired connection if you have one. The test runs on tablets, but a larger screen makes dense numerical tables far easier to scan. Physical calculator, blank paper and a pen on the desk. Phone in another room. Close every other tab. A notification during a six-minute section is expensive. Do two or three warm-up questions first so the real test is not your cold start. During the test Commit to roughly 45 seconds per question and move. A question you have not cracked in a minute is a question you are donating time to. Answer every question. There is no negative marking, so a considered guess always beats a blank. When you guess, eliminate first. Two eliminated options turn a 25% chance into 50%. Do not look back. Testlet time limits mean a finished block is finished. Treat a difficulty jump as normal. Items are arranged to get harder, so the last block feeling brutal is the design, not a verdict. The four most expensive mistakes Hunting for perfection on question one. The opening block is the easy one. Banking it quickly and calmly sets your pace for the rest. Rushing because pace is measured. It is measured separately from accuracy. Speed without correct answers improves nothing. Taking the test on a phone between other things. Eighteen minutes of full attention decides whether your application survives. Using answer dumps. The item bank makes them useless, and a supervised retest later in the process will expose any score you cannot reproduce. Frequently asked questions How long is the Saville Swift Analysis Aptitude test? 18 minutes in total: three sections of six minutes each, covering verbal, numerical and diagrammatic analysis. How many questions are on the test? 24 questions — eight per section. What is a good score on Swift Analysis Aptitude? Scores are reported on a 1–10 sten scale against a norm group. Sten 5–6 is average. Sten 7 places you above roughly 75% of the comparison group and clears most graduate thresholds. Competitive schemes often look for 8 or higher. What is the pass mark? There isn't one. Employers set their own cut-off against the norm group, and they rarely publish it. Can I use a calculator? Yes, and you should. A physical calculator is faster than an on-screen one. Is the test negatively marked? No. Answer every question — an eliminated-options guess is always better than a blank. Does everyone get the same questions? No. The test draws testlets from a large item bank and equates scores using Item Response Theory, so candidates sit different but comparable forms. This is why "answer PDFs" do not work. Can I go back and change an answer? No. Questions come in testlets of four with their own time limits, so once a block closes it stays closed. What is the difference between Swift Analysis and Swift Executive? The first two sections are identical. Swift Analysis finishes with diagrammatic reasoning; Swift Executive finishes with abstract reasoning and targets senior and director-level roles. Will I have to retake it in person? Possibly. Saville provides supervised follow-up versions for verification, so some employers retest shortlisted candidates. How much preparation do I need? Two weeks at 45 minutes a day is enough for most candidates. The gains come from format familiarity and timing, not from new knowledge. Which section should I prioritise? Numerical. Across 95,337 test takers it has the lowest average score at 53.6% correct. Sources Analysis Aptitude Range Technical Summary — Saville Assessment / Willis Towers Watson, 2022. Example Swift Analysis Aptitude Pace Rx Report — Saville Assessment. Saville Assessment aptitude range — publisher overview and client case studies. Saville Assessment test provider guide — Psychometric Success. Saville test types explained — Practice Aptitude Tests. All sample questions in this guide are original and written for illustration. They will not appear on your assessment.

Day: November 7, 2026

Saville Swift Analysis Aptitude Test: 2026 Guide & Practice

Quick Links

Resources

Support