For BPO recruitment leads, shared-services HR and staffing agencies in the Philippines

Five ways to buy a Philippine hiring screen, and what each one is actually good at.

Choosing an assessment platform for Philippine hiring is the decision of what evidence you will hold about a candidate's English, judgement and integrity, and what it costs per applicant to hold it. Five options compete: dedicated spoken-English instruments, volume-hiring automation platforms, global enterprise vendors, subscription test marketplaces, and pay-per-use measurement. They differ most on what they measure, what they publish, and what happens at the cut score.

Run a pilot wave with 250 free credits +91 99000 11601

At a glance

Assessment Platforms Compared: the Philippines: the facts, with their units

Who publishes a price	Of six providers a Philippine employer commonly shortlists, checked at their own pages on 6 September 2026: TestGorilla publishes plan prices. SHL, Mercer Mettl, Pearson (Versant), Harver and Talkpush publish none — every one of them routes a buyer to a demo, a quote or a phone call. One in six is the market, not an accident.
What AssessAll costs, publicly	1 credit = US$0.50. A Philippine BPO screening set runs 15–26 credits, or US$7.50–13 per candidate, with volume pack discounts up to 13%. No platform fee, no seat licence, no annual minimum, 250 free credits to start.
Subscription vs pay-per-use, the crossover	TestGorilla's published Core plan is US$142 per month on annual commitment (US$1,704 a year). Divide that by an AssessAll screen and you get the volume at which the two cost the same: roughly 131 candidates a year at US$13 a screen, 227 at US$7.50. Below your crossover pay-per-use is cheaper; above it, the plan is.
The sentence this market repeats without checking	"Call-centre agents need at least B2." It is stated on employer-facing pages without a source, and it is not a specification. CEFR is a band, each vendor maps its own scale onto it, and the vendors deliberately avoid claiming equivalence. Specify a vendor, a scale, a number and subscale minimums — not a letter.
Why a letter is not a bar, in arithmetic	On AssessAll's published English weights (Sentence Mastery 0.30, Vocabulary 0.20, Fluency 0.25, Pronunciation 0.25) and its 10–90 reporting scale, a candidate scoring 59 on the first three subscales and 34 on pronunciation lands at 53 overall — the exact bottom of B2, with an A2 pronunciation subscale. That is the profile a voice queue should not hire, and a CEFR letter hides it.
What happens at the cut, stated in code	AssessAll does not treat a cut score as a bright line. `hireBand()` in lib/english/cefr.ts returns clear_hire at or above the cut and borderline within 4 scale points below it — so a 49 against a default voice cut of 53 arrives labelled borderline for a human to look at, not rejected. Ask every vendor what their equivalent of that four-point band is.
Language, stated plainly	AssessAll is English-first. A search of the codebase for Filipino, Tagalog and Cebuano returns nothing: there is no Filipino-language delivery, and none is being claimed. For a voice or chat account served in English that is the point of the screen; for a domestic Filipino-language process it is the wrong tool.
What AssessAll does not have here	No Philippine data-residency option is published, no Philippine norm group exists, and the English band boundaries are a documented launch calibration rather than empirically derived cut scores. If a client SOW names a recognised external English benchmark, a dedicated language instrument is the right purchase and this page says so in the row that sells us.
How to use this page	Not as a ranking. Every row names the case where that option is the right one, including the rows that are not ours, and every external fact carries the date it was read from the provider's own page.

SiddharthanFounder, AssessAll — Bodhih Training Solutions

Founder of AssessAll and of Bodhih Training Solutions, a corporate training company in Bangalore. Works on assessment design, scoring and reporting across hiring, L&D and certification programmes.

Last reviewed 6 September 2026

External facts in the comparison table were read from each provider's own published pages on 6 September 2026 and are re-verified monthly. Nothing here is an endorsement, a ranking, or a claim about any provider's validity or suitability for your roles.

The comparison pages in this market are written for candidates, not for buyers

Search for the tests a Philippine BPO uses and the results are practice-test sites, score-chart explainers and outsourcing agencies selling their own bench. Every one answers what a candidate should do about the test. Almost none answers what an employer should do with the score — where to set the cut, how wide the band is, what a wrong answer diagnoses. This page is written for the second reader.

A widely-coached screen measures practice as well as ability

The most commonly used spoken-English formats in Philippine recruitment are also the most heavily coached: an ordinary search returns paid practice packs, score charts and item-type drills aimed squarely at applicants. That is not a flaw in any instrument — it is what happens to any test with a stable public format and a large market. The remedy is a second, unrehearsed job-shaped sample, not a higher cut score.

A price you cannot see cannot be compared

Five of the six providers checked publish no figure at all, which makes a like-for-like table impossible until you are already in a sales process. The practical move is to convert every quote into one number — cost per applicant assessed, at the volume you will actually run this year — and to ask for it in writing at that volume, including the applicants who start and never finish.

Funnel problems and measurement problems get bought as the same product

Sourcing 35,000 applications a month, pre-screening them conversationally and scheduling the ones who reply is a funnel problem. Deciding which of the qualified ones will hold up on a US voice queue at 3 a.m. is a measurement problem. The platforms that are excellent at the first are not automatically the ones you want deciding the second, and a single demo rarely separates them.

What you get

Built for assessment platforms compared: the philippines

Convert every quote to cost per applicant assessed

Take the annual figure, divide by the number of applicants you will genuinely assess this year — not the number you will hire — and compare that one number across vendors. A plan that looks cheap at 4,000 applicants is expensive at 600, and a per-candidate price that looks high is cheap when the drive is two waves a year.

Ask what fraction of applicants never finish

In a walk-in or mobile-first Philippine drive, completion rate decides your real cost per assessed candidate more than the list price does. Ask every vendor for their median completion rate on mobile, on a weak connection, for an assessment of the length they are proposing — and whether an abandoned attempt is billed.

Ask for the coaching answer

Any test with a stable public format and a large candidate market gets coached; that is a fact about markets, not an accusation about instruments. Ask each vendor what changes between two candidates sitting the same assessment a week apart, and whether any part of the screen is composed fresh rather than drawn from a fixed public form.

Ask what happens the second time you measure

If any part of the budget is for training or nesting, the question is not what a report looks like but whether the same competencies can be re-scored months later and compared per person. Ask to see a before-and-after artefact. Screening-first platforms often have none, and it does not appear on a feature grid.

Ask for the failure case

Every honest vendor can name the situation in which their product is the wrong tool. A vendor who cannot has either not thought about it or is not going to tell you, and both make the reference call more important than the demo.

Run a pilot wave before signing anything

One real wave of real applicants tells you more than any comparison page, including this one. Run the same fifty candidates through your existing screen and a candidate one, then look at where the two disagree — the disagreements are the whole finding.

Before the first demo

The six questions that decide a Philippine shortlist, in the order to ask them

1
Is the bottleneck the funnel or the decision?
Write down how many applications arrive a month and how many of the people you hired last quarter you would hire again. If the first number is the pain, you are buying automation. If the second is, you are buying measurement. Most teams have both problems and only budget for one, so decide which one this purchase is for before anyone demos anything.
2
What is the cut, on whose scale, and how wide is the band?
Never write a CEFR letter into a requisition. Ask for the vendor's own scale, the number you would set on it, and what the instrument's measurement error is around that number. A vendor who cannot tell you how far apart two candidates must be before the difference is real is asking you to treat noise as a hiring decision.
3
Which subscales does the report break out, and can you set a floor on one?
For a voice queue, pronunciation and fluency matter differently from vocabulary. An overall band can hide a weak subscale entirely — on AssessAll's published weights a bottom-of-B2 overall score can contain an A2 pronunciation. Ask whether you can set a minimum on the subscale you actually care about, or only on the composite.
4
What does the platform do at the boundary?
Ask what happens to a candidate who lands just below the cut: rejected automatically, or flagged for a human? AssessAll returns a borderline band within four scale points below the cut. Whatever a vendor's answer is, it tells you whether you are buying a filter or a decision aid.
5
What does a proctoring flag actually do?
A flag that auto-rejects and a flag that lowers a confidence band are entirely different products with the same feature name. Ask which one it is, whether a human reviews it, what evidence that reviewer sees, and how long the imagery is kept — retention for a score and retention for a photograph should be two separate answers.
6
Where does the data sit, and under what basis does it leave?
Ask where attempt data and proctoring imagery are stored, on what basis either leaves that location, and how long each is retained. In the Philippines this sits alongside the Data Privacy Act and, for any recorded voice, the Anti-Wiretapping Act — both covered instrument by instrument on our Philippine compliance page. Get the answer in writing before the shortlist, not after the demo.

Compare the approaches

The five options, and when each one is the right answer

Approach-level rows with named examples. Every external fact below was read from that provider's own published page on 6 September 2026, and a price is stated only where the provider publishes one. This is a comparison, not a ranking: the 'when to choose this instead' line is the point of each row.

External facts verified at source on 6 September 2026.

Approach	What it measures	What it costs	When to choose it instead
A dedicated spoken-English instrument (for example Versant by Pearson)	Language, precisely and repeatably. Pearson publishes six Versant tests running 17 to 60 minutes; its own page states that "Versant by Pearson Tests scores are 100% AI-based", that results arrive within minutes, and that scores are reported against the Global Scale of English and the CEFR. A long-established scale with a name a client will recognise is a real asset a newer platform cannot manufacture.	Not published. Checked 6 September 2026: the Versant tests page states no price and routes a buyer to contact Pearson.	Choose it when a recognised external English benchmark is itself the requirement — a client statement of work that names one, an account transition where your buyer wants a scale they already know, or a regulator or accreditor who does. Choose it also when you need one stable number to compare across several suppliers' funnels. If English is the whole decision, buy the instrument built for English.
High-volume hiring automation platforms (for example Harver, Talkpush)	The funnel. These platforms are built around application volume: sourcing, conversational pre-screening, document validation, scheduling and progression rules. Harver states that it serves "leading BPOs, contact centers, and retail organizations" and reports lowering 90-day attrition by 25%; Talkpush's own pricing page cites 35,000 monthly applications managed and 2.2 million candidates processed annually, and names McDonald's Philippines as a customer. Those are the providers' own published claims, not ours.	Not published by either. Checked 6 September 2026: Harver has no pricing page and its call to action is "Request a Demo"; Talkpush's pricing page carries no figures and asks a buyer to "Get a quote" or "Book a demo". Every third-party figure circulating for these platforms comes from procurement aggregators rather than from the vendor.	Choose them when the bottleneck is genuinely the funnel — tens of thousands of applications a month, walk-in and social sourcing, no-shows, scheduling, chat-first candidates on mobile. At that scale the automation is worth more than any instrument on this page. Ask separately what the assessment layer inside it actually measures and how the cut was set, because that is a different question from how many applications it can move.
Global enterprise assessment vendors (for example SHL, Mercer Mettl)	Large validated instrument libraries with published norm data, contact-centre simulations, assessor services and enterprise integrations. SHL's catalogue lists Contact Center Hiring and Call Center Simulations among its products. Decades of accumulated validity evidence is the thing a buyer is paying for.	Not published by either. Checked 6 September 2026: SHL's product catalogue asks a buyer to "Speak With Our Team" or book a demo; Mercer Mettl's pricing page offers a "Customised plan as per your organisation" and a phone number, with no tiers and no figures.	Choose them when you need norm data and validity evidence for a job family genuinely like yours and you are prepared to read the technical manual; when a simulation or an assessment centre is part of the design; or when procurement wants one master agreement across several countries. If a selection decision is going to be challenged, published norms are what you want to be holding.
Subscription skills-test marketplaces (for example TestGorilla)	A broad library of short role and skill tests assembled into a screen, with the cost carried by a plan rather than by the candidate.	Published, and unusually so. Checked 6 September 2026: a Free plan at US$0 a month with 10 credits a month; Core at US$142 per month on annual commitment, billed US$1,704 annually; Plus from US$400 per month, annual commitment from US$4,800; Enterprise on request.	Choose a subscription when hiring runs at a steady volume all year, because a plan amortises and pay-per-use does not. Against an AssessAll Philippine screen the published Core plan becomes the cheaper shape somewhere between roughly 131 and 227 candidates a year, depending on which screen you run. Do that division with your own two numbers before assuming either model wins.
An in-house panel English interview plus a typing test, marked on a form	Exactly what your team asked, which can be the most job-relevant thing on this page — and nothing about consistency between interviewers, or about what a score means next month, unless you build that too. It remains the most common Philippine screen by volume.	No licence, and a real cost in recruiter and team-leader hours per applicant, plus the problem that two panels rarely mean the same thing by "good English".	Choose it when volume is genuinely low, when the account is specific enough that nothing off the shelf measures it, and when the same two people can interview everyone. Below a few dozen hires a year the overhead of any platform is hard to justify. If you keep it, write the scoring guide and the anchors before the first candidate, not after the first disagreement — that single document is most of what a platform sells you.
Pay-per-use measurement with a re-measure loop (AssessAll)	One proctored sitting on a share link or QR code, with no candidate account: AI-scored spoken English on a 10–90 scale with four reported subscales, listening, typing speed and accuracy, and customer-scenario judgement, plus AI-composed items graded against your own competency framework. Proctoring carries an identity-baseline photo and an integrity band that a human reads. For training teams, the same cohort can be baselined, developed and re-scored so a gain is a measured delta rather than a feedback form.	Published: 1 credit = US$0.50, a Philippine BPO screening set 15–26 credits (US$7.50–13) per candidate, volume pack discounts up to 13%, 250 free credits, no seats and no annual minimum.	It is the wrong choice when a client SOW names a recognised external English benchmark — AssessAll publishes no Philippine norm group, and its band boundaries are documented as a launch calibration rather than empirically derived cut scores. It is the wrong choice for a Filipino-language process, since there is no Filipino-language delivery. It is the wrong choice when the bottleneck is sourcing rather than selection, and the wrong choice when in-country data residency is a procurement requirement, because none is published.

Prices and published-pricing statements were read from each provider's own pages on 6 September 2026 and are re-verified monthly; a row that cannot be re-verified is deleted rather than left to go stale. Claims attributed to a provider are that provider's own published claims and are not independently verified here. Nothing on this page is a judgement about the quality, validity or suitability of any provider for your roles. AssessAll holds no accreditation of any kind.

Sources: Pearson — Versant tests · Harver · Talkpush pricing · SHL product catalogue · Mercer Mettl pricing · TestGorilla pricing

Common use cases

Choosing a screening platform for a new voice or chat account
Comparing per-applicant cost against a subscription before a large drive
Writing the assessment section of an RFP for a Philippine BPO or shared-services centre
Deciding whether a dedicated external English benchmark is a hard requirement
Separating a funnel-automation purchase from a measurement purchase
Staffing agencies choosing the screen they will run on behalf of clients

Pricing for philippine platform comparison

AssessAll publishes its prices because a buyer cannot compare what they cannot see. Philippine organisations are quoted and charged in US dollars: 1 credit = US$0.50, a BPO screening set 15–26 credits (US$7.50–13) per candidate, with volume pack discounts up to 13%. There is no platform fee, no seat licence and no annual minimum, and new organisations get 250 free credits — enough to run a pilot wave before deciding anything on this page.

See full pricing solutions@bodhih.com

Frequently asked questions

Which assessment platform is best for BPO hiring in the Philippines?+

There is no single best one, and a page that names one is usually selling it. The choice is decided by four questions in this order: is your bottleneck the funnel or the decision; does a client statement of work name a recognised external English benchmark; is your volume steady enough that a subscription beats pay-per-use; and do you need to measure the same people again after training. Answer those four and the shortlist is usually two providers, not ten.

Do assessment vendors publish their prices?+

Mostly not. Checked at each provider's own pages on 6 September 2026: SHL's product catalogue asks a buyer to speak with its team or book a demo; Mercer Mettl offers a customised plan and a phone number; Pearson's Versant tests page states no price; Harver has no pricing page and asks for a demo; Talkpush's pricing page carries no figures and asks for a quote. TestGorilla is the exception and publishes plan prices, including a free plan with 10 credits a month and Core at US$142 per month on annual commitment. AssessAll publishes a per-credit price of US$0.50 and per-assessment prices derived from it.

Is B2 the minimum English level for a Philippine call-centre agent?+

"B2 minimum" is repeated across employer-facing pages in this market, generally without a source, and it is not a specification. CEFR is a band rather than a point, every vendor maps its own scale onto it, and vendors deliberately avoid claiming equivalence between their scales — AssessAll reports English on a 10–90 scale specifically to avoid implying equivalence with any other test's numbers. Two candidates can both be "B2" and be thirteen points apart, and two candidates one point apart can sit in different letters. Specify the vendor, the scale, the number and a minimum on the subscale you care about.

Can a candidate pass an English screen with weak pronunciation?+

Yes, and the arithmetic is public. AssessAll weights its English composite Sentence Mastery 0.30, Vocabulary 0.20, Fluency 0.25 and Pronunciation 0.25, and reports on a 10–90 scale where B2 runs 53 to 66. A candidate scoring 59 on the first three subscales and 34 on pronunciation lands at 53 overall — the exact bottom of B2, with an A2 pronunciation subscale, which is the wrong profile for a voice queue. This is why a composite band is a starting point and a subscale floor is the actual bar. Ask every vendor whether their report breaks the score out and whether you can set a floor on one part of it.

What happens to a candidate who scores just below the cut?+

It depends entirely on the platform, and it is worth asking before you buy. AssessAll does not treat the cut as a bright line: its hireBand function returns clear_hire at or above the cut and borderline within four scale points below it, so a 49 against a default voice-process cut of 53 reaches a recruiter labelled borderline rather than rejected. The default cuts themselves are published — 53 for a voice process, 45 for chat, 67 for a client-facing team lead — and are editable per product. Ask every other vendor what their equivalent band is; a vendor with no answer is selling you a filter and calling it an assessment.

Are widely used English tests coached in the Philippines?+

Any test with a stable public item format and a large applicant market attracts coaching, and a search for the common spoken-English formats used in Philippine recruitment returns paid practice packs, score charts and item-type drills aimed at candidates. That is a fact about markets rather than a criticism of any instrument. The practical consequence for an employer is that raising the cut score does not fix it — a rehearsed candidate clears a higher bar too. The remedy is a second, unrehearsed, job-shaped sample in the same sitting: a scenario the candidate has not seen, scored on the behaviours the account actually needs.

Does AssessAll deliver assessments in Filipino or Tagalog?+

No. The platform and catalogue are English-first, and a search of the codebase for Filipino, Tagalog and Cebuano returns nothing — there is no Filipino-language delivery and none is claimed. For a voice, chat or back-office account served in English that is the point of the screen, since workplace English is usually one of the capabilities being measured. For a domestic process delivered in Filipino it is the wrong tool, and a local or bilingual provider belongs on the shortlist instead.

Where is assessment data stored, and does that matter in the Philippines?+

It matters whenever a client's procurement or data-protection review asks, which is increasingly often. AssessAll publishes no Philippine data-residency option. Treat it as a question for every vendor, in writing, in three parts: where attempt data is stored, where proctoring imagery is stored, and on what basis either leaves that location. It sits alongside two Philippine instruments worth reading before you design the form — the Data Privacy Act, including the National Privacy Commission's position on consent inside an employment relationship, and the Anti-Wiretapping Act on recorded voice.

How often is this comparison re-checked?+

Every external fact on this page was read from the provider's own published pages on 6 September 2026, and the page is re-verified monthly. A row that cannot be re-verified is deleted rather than carried forward with an old date. If something here has changed, it is worth telling us — a comparison page that is wrong is worse than no comparison page, and this one is dated and sourced so you can check it against the same pages yourself.