Third of customer service applicants paste into English test: Armasourcing
MANILA, PHILIPPINES — Candidates who switched away from their customer service written assessment tab 10 or more times scored 68.7 out of 100 on average — nearly 16 points higher than the 52.9 averaged by candidates who never left the tab at all.
The finding comes from Armasourcing’s Customer Service Screening Report 2026, which captured browser telemetry across 699 timed written assessments and 1,848 voice recordings from 670 candidates over a seven-week window ending July 27, 2026.
Eli Gutilban, CEO and founder of Armasourcing, notes in the report that 88% of the data came from a single hiring push for a United States (U.S.) moving and storage company — a scope limitation Gutilban flags himself.
Browser telemetry reveals external-resource behavior in one in three assessments
Across all 699 assessments, 36.6% recorded at least one paste event — when a candidate pasted content into a written answer field during the test. A paste event is a browser-logged signal; it does not distinguish between text sourced from an artificial intelligence (AI) tool and notes the candidate had prepared and copied from their own prior writing.
Candidates who pasted at least once averaged 69.0 on the written test, against 48.1 for those who did not — and averaged 28.9 minutes on the assessment compared to 22.7 minutes for candidates who pasted nothing. In total, 42.9% of candidates switched tabs at least once during the assessment — with performance by tab-switch frequency rising steadily: 52.9 points for no switches, 53.5 for 1–4 switches, 61.6 for 5–9 switches, and 68.7 for 10 or more.
The performance gradient runs opposite to what a proctored-test assumption would predict: candidates showing the most external-resource behavior produced the highest written test scores, while candidates who stayed entirely on-screen produced the lowest — a pattern that reframes paste events and tab switches as ambiguous signals rather than evidence of misconduct.
“More than a third of the candidates we screened pasted into an unproctored written test, and the ones who did spent more time on it, not less. That rules out the lazy explanation. They were doing what anyone does at a desk with a browser open,” said Eli Gutilban, Founder, Armasourcing.
A third of assessed candidates worked on connections under 25 megabits per second
One in three candidates — 33.3% of 691 assessed connections — had download speeds under 25 megabits per second (Mbps), a threshold below which voice call quality is typically compromised.
The median assessed connection was 36.4 Mbps at 68 milliseconds of latency — a bandwidth profile that clears the minimum for voice work but leaves limited headroom under peak-demand conditions.
Candidates averaged 43.3 words per minute (wpm) across written assessments, with 36.2% clearing the 70-point written test threshold Armasourcing applied as a passing bar.
The connectivity data is operationally direct: a candidate who clears the written and voice screen but works on a sub-25 Mbps connection faces an infrastructure constraint that no hiring process can fix after an offer is extended.
“The lesson for BPOs is that an unproctored written test can tell you who to rule out, but it cannot tell you who to rank. If a score is going to decide who gets hired, it has to come from a step someone actually watched happen,” Gutilban said.
For BPO operators sourcing customer service talent through unproctored written assessments, the Armasourcing report raises a calibration question for standard screening designs: written scores in unproctored environments partly reflect tool access rather than underlying ability, tab-switch behavior has no predictable relationship with candidate quality, and connectivity gaps represent a pre-hire operational risk that self-reported readiness cannot capture.
Gutilban recommends capturing browser telemetry and connectivity data as standard inputs — whether used for active screening or held as post-hire reference data — before the assessment design is treated as a clean signal.
Armasourcing is currently subscribed as an Outsource Accelerator (OA) BPO Partner. Through OA’s Marketing, Sales, and Source Partner Hubs, the firm accesses a unified growth platform designed to convert high-intent prospects and accelerate deal flow in the outsourcing industry.
Related news
- Philippine BPO workers push back against AI displacement · 31 Jul
- Sourcefit survey finds 52% of its offshore workers use AI weekly · 04 Aug
- MicroSourcing, Beepo open Clark hub with 400 jobs · 28 Jul
Disclosure: Outsource Accelerator uses AI tools in the backend of its editorial workflow. Every article is reviewed and verified by a human editor before publication.
Stay ahead of the outsourcing industry. Join thousands of business leaders who rely on Outsource Accelerator for the news, trends, and expert insights that matter. Subscribe to our free newsletter and never miss an update.
