Yes. Selection Lab builds custom assessment bundles per job role, from one library instead of three vendors. You pick blocks from over 200 assessments, hold each role to a time budget, and combine a cognitive test, a personality measure and a role-specific skills test into one flow that fires from your ATS.
The harder question is not whether such a tool exists. It is how you assemble a bundle that measures the right thing without asking forty-five minutes of a candidate who has three other applications open. That is a design job, and it takes four steps.
The word bundle covers two very different things, and buyers get burned by the difference.
One is a vendor package. A pre-set combination sold under a role name, which changes when the vendor decides to change it. The other is a configuration you assemble yourself from separate blocks and own. Only the second one moves when your role moves, and roles move more often than vendor packages do.
Three things make a bundle role-specific rather than merely role-labelled.
Start from failure, not from a competency framework. Ask the hiring manager one question. Of the last five people who did not work out in this role, what went wrong?
The answers cluster faster than you would expect. They lost the overview once volume rose. They wrote sloppy client emails. They could do the work but clashed with the team. They needed six months longer to get up to speed than anyone had budgeted for.
Each of those points at something measurable. Losing the overview points at planning and prioritisation. Sloppy client email points at a language or writing test at working level. Team friction points at personality and culture fit. A slow ramp-up points at cognitive ability.
Do not assume the cognitive test is automatically the heaviest block. The 2022 recalculation by Sackett and colleagues, which corrected an overstatement carried in the 1998 Schmidt and Hunter estimates, puts a cognitive ability test at .31 against job performance. A test of knowledge for the job reaches .40 and a structured interview .42. So the cheapest gain in most funnels is not a longer test, it is pairing a shorter one with an interview guide on the same competencies. We work through that evidence in detail in our piece on what counts as evidence from an assessment vendor.
Write the answers down as a list of risks. That list, not the job description, is what the bundle has to cover.
Most teams skip this step, and it is the one that decides whether the bundle survives contact with real candidates.
Decide up front how many minutes you are willing to ask of a candidate at this point in the funnel. Then shop inside that budget. Not the other way around.
This arithmetic only works if the blocks come in different sizes, which is the practical argument for a modular library over a fixed package. Games run two to five minutes. A motives block or a culture test runs two. A coping test is five to seven. Most hard skills tests land around fifteen. A full Big Five personality assessment is thirty, and an AI avatar interview is around thirty as well.
Set the budget by where you are in the funnel rather than by seniority.
| Funnel stage | Time budget | What fits inside it |
|---|---|---|
| First screening step, volume hiring | Under 10 minutes | SmartChat for the knockout questions plus one or two games |
| First screening step, white collar | 10 to 15 minutes | Three games plus a two-minute motives block |
| Mid-funnel | 20 to 40 minutes | One cognitive block, one personality block, one role-specific skills test |
| Final stage, senior or scarce role | 40 to 60 minutes | A COTAN-reviewed personality assessment plus a skills test, with proctoring if the stakes justify it |
What is not real is a forty-five-minute bundle in the first step of a volume funnel, and you will discover that through a completion rate that falls through the floor.
Once you have a library in front of you the temptation is to measure everything. Resist it. Two tests measuring the same construct do not double your certainty. They double the length and hand you two scores to argue about in the debrief.
A rule that holds up in practice. One cognitive block. One personality or motivation block. One role-specific skills block. Add a fourth only when the role carries a genuine fourth risk, such as a compliance-heavy position that needs a GDPR and EU AI Act check, or an international role that needs a language test at working level.
Our assessment pages are built to make this checkable. Each one names the roles it was designed for and the assessments it works well beside. Read two of them next to each other and you will usually see straight away whether you are stacking duplicates.
A bundle that lives in a spreadsheet is not a bundle. It is an intention. The configuration has to sit in the tool, fire automatically when a candidate reaches a stage, and write the result back so the recruiter never has to leave the system they already work in.
Two things to settle here, and they are governance rather than technology. Who is allowed to change a bundle, and what happens to candidates who are halfway through when someone changes it. If either answer is vague you will end up comparing candidates who took different tests for the same job, which quietly destroys the reason you started measuring in the first place.
On our side that means a stated go-live of two to ten weeks and integrations with the ATS systems recruitment teams already run. Ask any vendor to demo the trigger and the write-back with your own ATS rather than a sandbox.
Account executive, mid-funnel, around thirty-five minutes. A cognitive game, a swipeable Big Five for the personality read, a CRM test on Salesforce and HubSpot, and a business numeracy block, because anyone who misreads a margin misprices a deal. Culture fit only if the team is small enough that one wrong hire changes it.
Legal trainee, final stage, around fifty minutes. Legal Knowledge Basics, Legal English if the practice is international, and a COTAN-reviewed personality assessment. Add Legal Research with AI Tools when juniors are expected to work with AI from week one, and AI Critical Editing and Verification when their output leaves the building under a partner's name.
Warehouse operator, volume hiring, under ten minutes. SmartChat for the knockout questions, a two-minute personality game, and a cognitive ability test that does not penalise for language background, which matters when a large share of your applicants speak Dutch as a second language. Nothing else. At this volume every additional minute costs you candidates you already paid to attract.
Those three share exactly one block type, the personality read. That is the whole argument for building per role. One bundle stretched across all three would measure the wrong thing twice and the right thing never.
Ask a talent team which assessment suppliers they pay and you almost always get three names.
A psychometric publisher for cognition and personality, because those instruments need norms and an independent review. A skills-testing tool for the hard skills, because the publisher does not cover Excel or Salesforce. And a screening or chatbot tool at the front of the funnel, because neither of the other two speaks to candidates before the test goes out.
Three contracts. Three admin panels. Three consent flows. Three reports a recruiter lines up by hand on a Monday morning. That fragmentation is also why bundles fall apart in practice. A bundle spread over three systems cannot fire in one fixed order, cannot be held to one time budget, and cannot arrive in the ATS as one score a hiring manager reads in thirty seconds.
Our catalogue covers all three of those layers. Cognition and personality with instruments that carry a COTAN review, the Dutch quality judgement for psychological tests, scored against published criteria rather than a vendor's own claim. Over thirty hard skills tests across finance, data, IT, legal, marketing, creative and technical work, plus language tests in eight languages, culture fit, situational judgement, coping under pressure and optional proctoring. SmartChat for the screening conversation at the front. One platform, one contract, and one report at the end. You can browse the full library and filter it by type, format and length.
The gain is mostly not licence cost, and we would rather say that plainly than pretend otherwise. The gain is coordination. One person owns the configuration, one consent flow covers the whole assessment, and the candidate walks through one uninterrupted flow instead of three invitations from three senders on three different days.
Bundles multiply. Twelve roles turns into thirty-one bundles inside a year, because every hiring manager wants a variant. Cap the library at role families rather than job titles.
Nobody owns the library. Tests get added and never removed. Review once a year and delete what nobody reads.
The report does not follow the bundle. If a recruiter opens four tabs to compare two candidates, the bundle has failed even when every test inside it is valid.
The time budget creeps. Someone adds fifteen minutes for a good reason, four times over. Track total minutes per bundle as a number somebody is accountable for.
Two gaps worth knowing before you count on us for everything.
We do not run a live coding environment. Our Programming, Python, SQL and Database Management tests are fifteen-minute knowledge tests, not a repository a candidate builds and runs code in. An engineering team that wants a real work sample needs a specialist platform beside us. We also do not do background or pre-employment screening. If both of those sit in the middle of your process, we take two of your three vendors rather than all three.
And bundling on its own does not improve your hires. It removes one specific failure, which is measuring the wrong thing for the role or measuring so much that candidates walk away. Whether your hires get better is something you establish afterwards, by keeping the scores and comparing them with performance data twelve months on.
Yes. Selection Lab is one. You assemble a bundle per role from over 200 separate assessments, from two-minute games to thirty-minute personality instruments, and the bundle fires automatically from your ATS. What makes a tool suitable for this is whether its assessments are sold as separate blocks rather than as fixed role packages.
For most teams yes. Those three layers are what a typical assessment stack is made of, and a catalogue that carries all three lets one bundle fire in one order under one consent flow and land as one report. Two exceptions. Live coding work samples and background screening still need a specialist, so a tech-heavy process keeps one extra supplier.
Three is usually right. One cognitive block, one personality or motivation block, one role-specific skills block. Add a fourth only for a genuine fourth risk, such as compliance knowledge or a language at working level. Two tests measuring the same construct add length, not certainty.
Set the budget by funnel stage rather than by role. Under ten minutes for a first screening step in volume hiring, twenty to forty minutes mid-funnel, forty to sixty for a final stage on a senior or scarce role. Then pick blocks that fit inside it.
Yes, and that is the normal setup. Each bundle attaches to a role or role family and fires when a candidate reaches the stage you linked it to. The condition is that one person owns the configuration, otherwise candidates for the same job end up taking different tests.
Deciding between one integrated platform and a stack of separate tools? Read our guide on what one assessment platform has to cover, see the full assessment library, or book a demo and we will build a bundle for one of your open roles on the call.
Yes. Selection Lab builds custom assessment bundles per job role, from one library instead of three vendors. You pick blocks from over 200 assessments, hold each role to a time budget, and combine a cognitive test, a personality measure and a role-specific skills test into one flow that fires from your ATS.
The harder question is not whether such a tool exists. It is how you assemble a bundle that measures the right thing without asking forty-five minutes of a candidate who has three other applications open. That is a design job, and it takes four steps.
The word bundle covers two very different things, and buyers get burned by the difference.
One is a vendor package. A pre-set combination sold under a role name, which changes when the vendor decides to change it. The other is a configuration you assemble yourself from separate blocks and own. Only the second one moves when your role moves, and roles move more often than vendor packages do.
Three things make a bundle role-specific rather than merely role-labelled.
Start from failure, not from a competency framework. Ask the hiring manager one question. Of the last five people who did not work out in this role, what went wrong?
The answers cluster faster than you would expect. They lost the overview once volume rose. They wrote sloppy client emails. They could do the work but clashed with the team. They needed six months longer to get up to speed than anyone had budgeted for.
Each of those points at something measurable. Losing the overview points at planning and prioritisation. Sloppy client email points at a language or writing test at working level. Team friction points at personality and culture fit. A slow ramp-up points at cognitive ability.
Do not assume the cognitive test is automatically the heaviest block. The 2022 recalculation by Sackett and colleagues, which corrected an overstatement carried in the 1998 Schmidt and Hunter estimates, puts a cognitive ability test at .31 against job performance. A test of knowledge for the job reaches .40 and a structured interview .42. So the cheapest gain in most funnels is not a longer test, it is pairing a shorter one with an interview guide on the same competencies. We work through that evidence in detail in our piece on what counts as evidence from an assessment vendor.
Write the answers down as a list of risks. That list, not the job description, is what the bundle has to cover.
Most teams skip this step, and it is the one that decides whether the bundle survives contact with real candidates.
Decide up front how many minutes you are willing to ask of a candidate at this point in the funnel. Then shop inside that budget. Not the other way around.
This arithmetic only works if the blocks come in different sizes, which is the practical argument for a modular library over a fixed package. Games run two to five minutes. A motives block or a culture test runs two. A coping test is five to seven. Most hard skills tests land around fifteen. A full Big Five personality assessment is thirty, and an AI avatar interview is around thirty as well.
Set the budget by where you are in the funnel rather than by seniority.
| Funnel stage | Time budget | What fits inside it |
|---|---|---|
| First screening step, volume hiring | Under 10 minutes | SmartChat for the knockout questions plus one or two games |
| First screening step, white collar | 10 to 15 minutes | Three games plus a two-minute motives block |
| Mid-funnel | 20 to 40 minutes | One cognitive block, one personality block, one role-specific skills test |
| Final stage, senior or scarce role | 40 to 60 minutes | A COTAN-reviewed personality assessment plus a skills test, with proctoring if the stakes justify it |
What is not real is a forty-five-minute bundle in the first step of a volume funnel, and you will discover that through a completion rate that falls through the floor.
Once you have a library in front of you the temptation is to measure everything. Resist it. Two tests measuring the same construct do not double your certainty. They double the length and hand you two scores to argue about in the debrief.
A rule that holds up in practice. One cognitive block. One personality or motivation block. One role-specific skills block. Add a fourth only when the role carries a genuine fourth risk, such as a compliance-heavy position that needs a GDPR and EU AI Act check, or an international role that needs a language test at working level.
Our assessment pages are built to make this checkable. Each one names the roles it was designed for and the assessments it works well beside. Read two of them next to each other and you will usually see straight away whether you are stacking duplicates.
A bundle that lives in a spreadsheet is not a bundle. It is an intention. The configuration has to sit in the tool, fire automatically when a candidate reaches a stage, and write the result back so the recruiter never has to leave the system they already work in.
Two things to settle here, and they are governance rather than technology. Who is allowed to change a bundle, and what happens to candidates who are halfway through when someone changes it. If either answer is vague you will end up comparing candidates who took different tests for the same job, which quietly destroys the reason you started measuring in the first place.
On our side that means a stated go-live of two to ten weeks and integrations with the ATS systems recruitment teams already run. Ask any vendor to demo the trigger and the write-back with your own ATS rather than a sandbox.
Account executive, mid-funnel, around thirty-five minutes. A cognitive game, a swipeable Big Five for the personality read, a CRM test on Salesforce and HubSpot, and a business numeracy block, because anyone who misreads a margin misprices a deal. Culture fit only if the team is small enough that one wrong hire changes it.
Legal trainee, final stage, around fifty minutes. Legal Knowledge Basics, Legal English if the practice is international, and a COTAN-reviewed personality assessment. Add Legal Research with AI Tools when juniors are expected to work with AI from week one, and AI Critical Editing and Verification when their output leaves the building under a partner's name.
Warehouse operator, volume hiring, under ten minutes. SmartChat for the knockout questions, a two-minute personality game, and a cognitive ability test that does not penalise for language background, which matters when a large share of your applicants speak Dutch as a second language. Nothing else. At this volume every additional minute costs you candidates you already paid to attract.
Those three share exactly one block type, the personality read. That is the whole argument for building per role. One bundle stretched across all three would measure the wrong thing twice and the right thing never.
Ask a talent team which assessment suppliers they pay and you almost always get three names.
A psychometric publisher for cognition and personality, because those instruments need norms and an independent review. A skills-testing tool for the hard skills, because the publisher does not cover Excel or Salesforce. And a screening or chatbot tool at the front of the funnel, because neither of the other two speaks to candidates before the test goes out.
Three contracts. Three admin panels. Three consent flows. Three reports a recruiter lines up by hand on a Monday morning. That fragmentation is also why bundles fall apart in practice. A bundle spread over three systems cannot fire in one fixed order, cannot be held to one time budget, and cannot arrive in the ATS as one score a hiring manager reads in thirty seconds.
Our catalogue covers all three of those layers. Cognition and personality with instruments that carry a COTAN review, the Dutch quality judgement for psychological tests, scored against published criteria rather than a vendor's own claim. Over thirty hard skills tests across finance, data, IT, legal, marketing, creative and technical work, plus language tests in eight languages, culture fit, situational judgement, coping under pressure and optional proctoring. SmartChat for the screening conversation at the front. One platform, one contract, and one report at the end. You can browse the full library and filter it by type, format and length.
The gain is mostly not licence cost, and we would rather say that plainly than pretend otherwise. The gain is coordination. One person owns the configuration, one consent flow covers the whole assessment, and the candidate walks through one uninterrupted flow instead of three invitations from three senders on three different days.
Bundles multiply. Twelve roles turns into thirty-one bundles inside a year, because every hiring manager wants a variant. Cap the library at role families rather than job titles.
Nobody owns the library. Tests get added and never removed. Review once a year and delete what nobody reads.
The report does not follow the bundle. If a recruiter opens four tabs to compare two candidates, the bundle has failed even when every test inside it is valid.
The time budget creeps. Someone adds fifteen minutes for a good reason, four times over. Track total minutes per bundle as a number somebody is accountable for.
Two gaps worth knowing before you count on us for everything.
We do not run a live coding environment. Our Programming, Python, SQL and Database Management tests are fifteen-minute knowledge tests, not a repository a candidate builds and runs code in. An engineering team that wants a real work sample needs a specialist platform beside us. We also do not do background or pre-employment screening. If both of those sit in the middle of your process, we take two of your three vendors rather than all three.
And bundling on its own does not improve your hires. It removes one specific failure, which is measuring the wrong thing for the role or measuring so much that candidates walk away. Whether your hires get better is something you establish afterwards, by keeping the scores and comparing them with performance data twelve months on.
Yes. Selection Lab is one. You assemble a bundle per role from over 200 separate assessments, from two-minute games to thirty-minute personality instruments, and the bundle fires automatically from your ATS. What makes a tool suitable for this is whether its assessments are sold as separate blocks rather than as fixed role packages.
For most teams yes. Those three layers are what a typical assessment stack is made of, and a catalogue that carries all three lets one bundle fire in one order under one consent flow and land as one report. Two exceptions. Live coding work samples and background screening still need a specialist, so a tech-heavy process keeps one extra supplier.
Three is usually right. One cognitive block, one personality or motivation block, one role-specific skills block. Add a fourth only for a genuine fourth risk, such as compliance knowledge or a language at working level. Two tests measuring the same construct add length, not certainty.
Set the budget by funnel stage rather than by role. Under ten minutes for a first screening step in volume hiring, twenty to forty minutes mid-funnel, forty to sixty for a final stage on a senior or scarce role. Then pick blocks that fit inside it.
Yes, and that is the normal setup. Each bundle attaches to a role or role family and fires when a candidate reaches the stage you linked it to. The condition is that one person owns the configuration, otherwise candidates for the same job end up taking different tests.
Deciding between one integrated platform and a stack of separate tools? Read our guide on what one assessment platform has to cover, see the full assessment library, or book a demo and we will build a bundle for one of your open roles on the call.
