What is an attention to detail test?

It measures how accurately and quickly you notice small differences or errors in information. Typical formats are comparing two lists, finding typos in data, or checking that copied figures match the original. Some versions add a second task to test divided attention.

What is an error checking test?

An error checking test is a timed attention to detail test, usually with codes, numbers or addresses. You compare an original with a copy and mark where they differ. It is common for administrative, finance and operations roles, and it also appears in graduate assessments for banks and consultancies.

What is the SHL checking test?

"SHL checking test" is a common search term for clerical checking exercises of this kind. An SHL version was not observed first-hand for this guide, and its current format could not be confirmed from SHL's own pages. The core task of any checking test is the same as in the drills above: compare items for differences under time pressure. Check your invitation for the exact format and timing.

What is the cut-e (Aon) multitasking test?

The term refers to multitasking tasks in Aon's (formerly cut-e) catalogue, in which you handle two or more simple jobs at once. That version was not observed first-hand here. The cyquest Multitaskingtest observed for this guide combines address checking with message routing. The skills the drills train, a fixed checking order and clean switching between tasks, carry over to any dual-task format.

How can I improve my attention to detail for the test?

Practise with self-made comparison lists. Use a fixed field order and read long words in syllables. Most people have one blind spot, often dropped letters in long words, and a few days of targeted practice fixes most of it.

Is it better to be fast or accurate?

Both count. cyquest, for example, reports concentration and processing speed separately. Aim for steady speed with very few errors. A pile of fast but false "identical" answers usually scores worse than a slightly slower, clean run.

Attention to Detail Test: Error Checking

An attention to detail test asks you to compare information quickly and spot differences: two versions of an address, an invoice copied from one system to another, a code typed into a form. Multitasking versions add a second job at the same time, such as routing incoming messages to the right department. These error checking tests are short (often 5 to 15 minutes) and fast (about 1 to 2 seconds per comparison). They are scored on both speed and accuracy. This guide shows what the tests look like, which differences are hardest to spot, and gives you original drills with answers.

Key facts

Publishers: cyquest, SHL, Saville, Aon (cut-e) and others
Format: 6 address pairs per screen, each identical ✓ or not ✗
Second task: route a chat message to 1 of 4 departments
Time: about 6 min, one countdown, no item counter
Pace: about 10 s per screen of 6 pairs plus 1 message
Scoring: speed and accuracy
Adaptive: no

What it looks like

In the cyquest version seen in a 2026 application round, the section is called Multitaskingtest. A short introduction comes first, with two example tasks of two address rows each. You can start the examples or skip them. The scored part then runs as a series of screens:

Main task: a table with six rows. Each row shows an address from "Datenbank 1" on the left and one from "Datenbank 2" on the right. Next to each row are two buttons: ✓ (identical) and ✗ (not identical). Addresses have a street, a house number and a postcode. The order of postcode and street can change between rows, but within a pair it is the same.
Second task: a chat message pop-up in the bottom-left corner. It shows a customer message and four buttons for departments: costs, contracts, customer support and services. You forward the message by clicking one of them. The examples used two other categories, so expect the labels to change.
Side panel: a countdown and an exit icon. There is no item counter, so you cannot see how many screens are left.

The instruction amounts to: check whether the addresses are identical and click the matching button, and at the same time forward incoming chat messages.

One layout quirk matters. When the page is scrolled down, the chat panel overlaps the bottom of the address list, and the fourth routing button can disappear below the fold. If you see only three departments, scroll.

cyquest also sells these tasks as separate products. The Datenbankvergleich module compares two sets of entries to measure concentration with numbers and letters. The MultiTasker (about 10 to 15 minutes) builds several simultaneous tasks around a realistic concentration text and reports multitasking ability, concentration and processing speed. Other vendors' "checking tests" use the same core comparison, sometimes with codes, names or numbers instead of addresses.

Which differences are hard to spot

In the observed test, the differences came in four families:

Family	Example (invented)	Difficulty
Changed, dropped or swapped digit	240 vs 24; 9LP vs 9PL; 30729 vs 30792	easy: numbers are short, and the eye checks them digit by digit
One letter changed, dropped or added in a long word	Merriweather vs Merriwether; Crescent vs Cresent	hard: the brain auto-corrects familiar word shapes
Street type swapped	Road vs Street; Lane vs Way	medium: the word stays plausible
Spacing, punctuation, abbreviation	Old Mill vs Oldmill; St Aldhelm's vs St Aldhelms; 4a vs 4 a	medium to hard: easy to dismiss as "the same"

The key point: number differences are the easy part. One-letter differences inside long street names are where the misses happen. The typical error is judging a pair "identical" when it is not. Flagging an identical pair as different is much rarer. So when in doubt, check the word letter by letter before you click ✓.

Rules that held in the observed test, and that you should assume unless told otherwise:

Any difference counts, including a single space, an apostrophe or an abbreviation (Str. vs Straße, St vs Street).
A pair can have two differences. Mark it ✗ once. You do not count differences.
The chat routing is driven by keywords, but not always. Most messages contain the department's name. Some use the name of another department as a lure, while the request clearly belongs elsewhere.

Drill 1: address comparison (12 pairs)

Time yourself: 40 seconds. For each row, write ✓ (identical) or ✗ (not identical).

#	Database 1	Database 2
1	27 Bramblewood Close, Norwich NR4 6TJ	27 Bramblewood Close, Norwich NR4 6TJ
2	118 Kestrel Avenue, Bristol BS7 9LP	118 Kestrel Avenue, Bristol BS7 9PL
3	5 Ashcombe Terrace, York YO24 1HD	5 Ashcomb Terrace, York YO24 1HD
4	63 Wellington Road, Derby DE21 4QX	63 Wellington Street, Derby DE21 4QX
5	9 Fairhaven Gardens, Bath BA2 5RW	9 Fairhaven Gardens, Bath BA2 5RW
6	240 Thornbury Way, Leeds LS6 2EB	24 Thornbury Way, Leeds LS6 2EB
7	31 St Aldhelm's Place, Exeter EX4 7NS	31 St Aldhelms Place, Exeter EX4 7NS
8	77 Merriweather Drive, Reading RG30 2UF	77 Merriwether Drive, Reading RG30 2UF
9	4 Cobblestone Row, Durham DH1 3LE	4 Cobblestone Row, Durham DH1 3LE
10	156 Larkspur Crescent, Leicester LE3 8GZ	156 Larkspur Cresent, Leicester LE3 8GZ
11	12 Old Mill Lane, Chester CH2 1BY	12 Oldmill Lane, Chester CH2 1BY
12	88 Quarry Hill, Stockport SK4 4HW	88 Quarry Hill, Stockport SK4 4HW

Answer: ✓ ✗ ✗ ✗ ✓ ✗ ✗ ✗ ✓ ✗ ✗ ✓ (4 identical, 8 different).

2: postcode letters swapped (9LP / 9PL).
3: "e" dropped (Ashcombe / Ashcomb).
4: street type swapped (Road / Street).
6: digit dropped (240 / 24).
7: apostrophe missing (Aldhelm's / Aldhelms).
8: "a" dropped (Merriweather / Merriwether). This is the hardest row: the word still looks familiar.
10: "c" dropped (Crescent / Cresent).
11: space removed (Old Mill / Oldmill).

If you missed 8 or 10, you have the typical blind spot. Read long words in syllables (Mer-ri-wea-ther), not as a whole shape.

Drill 2: dual task (2 screens + 6 messages)

This drill simulates the multitasking format. Departments: Invoices / Contracts / Customer service / Product information. Handle each screen and its two messages in 20 seconds.

Screen A

#	Database 1	Database 2
1	14 Hollins Road, Oldham OL8 3AE	14 Hollins Road, Oldham OL8 3AE
2	9 Penderel Street, Walsall WS3 1QJ	9 Penderell Street, Walsall WS3 1QJ
3	201 Garth Lane, Neath SA11 2DN	210 Garth Lane, Neath SA11 2DN
4	6 Cotterill Way, Crewe CW1 4RT	6 Cotterill Way, Crewe CW1 4RT
5	33 Lindale Mews, Kendal LA9 6HB	33 Lindale Mews, Kendal LA9 6BH
6	2 Ruskin Parade, Harrow HA1 4NE	2 Ruskin Parade, Harrow HA1 4NE

Messages: - A1: "My last invoice shows the same delivery charge twice." - A2: "What would your services cost for a team of 20 people?"

Screen B

#	Database 1	Database 2
1	57 Belmont Drive, Preston PR2 8JX	57 Belmont Drive, Preston PR2 8JX
2	4a Sefton Court, Luton LU2 7WE	4 a Sefton Court, Luton LU2 7WE
3	18 Tranmere Grove, Hull HU5 3PS	18 Tranmere Grove, Hull HU5 3PS
4	72 Aldridge Close, Slough SL1 9TU	72 Aldrige Close, Slough SL1 9TU
5	3 Highfield Rise, Truro TR1 2LG	3 Highfield Rise, Truro TR1 2LG
6	140 Wexham Street, Dudley DY2 0QA	1400 Wexham Street, Dudley DY2 0QA

Messages: - B1: "Can you send me a copy of my signed agreement?" - B2: "Nobody has called me back since Monday, although I was promised a call."

Two extra messages, to route without a screen: - C1: "Does the premium plan include weekend delivery?" - C2: "I want to end my agreement at the end of the notice period."

Answers

Screen A: ✓ ✗ ✗ ✓ ✗ ✓. Row 2 has an added "l" (Penderel / Penderell). Row 3 has swapped digits (201 / 210). Row 5 has swapped postcode letters (6HB / 6BH).
Screen B: ✓ ✗ ✓ ✗ ✓ ✗. Row 2 has an added space (4a / 4 a). Row 4 has a dropped "d" (Aldridge / Aldrige). Row 6 has an added digit (140 / 1400).
A1 → Invoices. A billing error on an existing invoice.
A2 → Product information. "Cost" is a lure. The customer asks for a price before buying, which is not an invoice question.
B1 → Contracts. "Agreement" is a synonym for contract.
B2 → Customer service. A service failure, with no product or contract question.
C1 → Product information. A question about what a plan includes.
C2 → Contracts. Ending an agreement is a contract matter, even though no department name appears.

Route by the underlying request, not the first keyword you see. In the observed test most messages were unambiguous, and one was built so that a department name pointed the wrong way.

Drill 3: numerical checking (8 rows)

The original entry is on the left, the copy on the right. Mark every row where the copy differs. Time: 25 seconds.

#	Original ref	Original amount	Copy ref	Copy amount
1	INV-40718	£1,284.50	INV-40718	£1,284.50
2	INV-40731	£906.20	INV-40713	£906.20
3	INV-40744	£3,517.00	INV-40744	£3,571.00
4	INV-40752	£48.95	INV-40752	£48.95
5	INV-40769	£2,260.10	INV-40769	£2,260.01
6	INV-40770	£715.80	INV-40770	£715.80
7	INV-40786	£9,032.40	INV-40786	£9,032.40
8	INV-40791	£164.00	INV-40719	£164.00

Answer: rows 2, 3, 5 and 8 differ. Row 2: the reference has swapped digits (40731 / 40713). Row 3: the amount has swapped digits (3,517 / 3,571). Row 5: the decimals are swapped (.10 / .01). Row 8: the reference has swapped digits (40791 / 40719). Rows 2 and 8 show why you must check every column. The amounts match there, so a reader who checks only the money misses them.

Traps and how to avoid them

Reading words as shapes. Long, familiar words hide dropped letters. Read them in syllables, or check the first three and the last three letters, then the middle.
Stopping at the first match. If the number and postcode match, keep going: the street name may still differ. Check every field, in the same order every time (number → street → town → postcode).
Ignoring spaces and punctuation. In these tests, "Old Mill" ≠ "Oldmill" and "St" ≠ "Street". Treat any visible difference as a difference.
Letting the second task break your rhythm. Finish the current row, route the message, then go back to the next row. Do not jump halfway through a comparison.
Keyword routing on autopilot. Read the whole message. Price questions are not invoice questions, and a complaint about a missing call is a service issue even if it mentions a product.
Hidden buttons. If a routing option seems missing, scroll. Panels can overlap.
Speeding without checking. The scoring rewards both speed and accuracy. Racing through with a pile of false "identical" answers is not a winning trade.

How to prepare for an attention to detail test

Five-day plan (20 minutes a day):

Day	Focus
1	Drill 1 untimed, then timed. Note which difference family you miss
2	Make your own drill: copy 20 addresses from any directory and introduce 10 one-letter changes (or have a friend do it)
3	Drill 2 with a timer. Practise the rhythm: row, row, row, message
4	Drill 3 plus a similar set with account numbers. Practise the fixed field order
5	Full simulation: 15 screens of 6 pairs in 3 minutes, with a message every screen

Targets: about 1.5 seconds per pair and fewer than 1 error in 30 pairs. In the observed test, the pace was about 10 seconds per screen of six pairs plus one message.

Setup: use a mouse rather than a trackpad, because the task involves a lot of clicking. Use a full-size screen, so the address table and chat panel do not overlap. Close notifications and other tabs, because a real interruption on top of the simulated one costs accuracy.

Sources

執筆：

Adrian、専門家

心理学修士、7年の受験歴。

最終更新：Oct 1, 2026