Data Modeling

1. Introduction

Software testing occupies an odd position in the software industry — or so it has seemed for most of the field's history. It is, in principle, indispensable; nobody sets out to ship broken code, and Black's (2007) now-standard treatment of the discipline frames testing as the practice that stands between a working system and one that merely looks like it works. And yet, for decades, testers were treated as something closer to an afterthought: a checkpoint bolted onto the tail end of "real" development rather than a discipline with its own intellectual demands. That perception has been shifting, gradually and unevenly, as organisations have come to recognise — sometimes the hard way — that quality assurance shapes whether a product survives contact with the market. The rise of continuous delivery pipelines, in which testing is woven into every build rather than reserved for a final gate (Humble & Farley, 2010), and of agile methods that pull testers into design conversations earlier than before (Bordin & De Angeli, 2016), has arguably done more to change this image than any amount of advocacy could. Still, the change has been partial, and it plays out differently depending on where in the world one happens to look.

A typical software development life cycle moves through planning, requirements analysis, design, development, testing, deployment, and maintenance — and testing sits at the point where mistakes made earlier become visible, or, ideally, are caught before they become anyone else's problem. Requirements themselves rarely hold still long enough to make that job easy; Nurmuliani et al. (2004) documented just how much requirements volatility complicates downstream verification, which is one reason testing is harder in practice than the tidy life-cycle diagram suggests. Given the position testing occupies, it is a little strange that testers were long regarded as second-class members of development teams — Kassab et al. (2021) note that many engineers actively avoid testing as a career entry point, treating it instead as a stepping stone toward analyst, programmer, or architect roles. Part of this may be a visibility problem: Storey et al. (2005) argued, in a related context, that awareness of who is doing what on a development team depends heavily on how that work is represented and surfaced, and testing work has traditionally been represented poorly. Part of it may also be pedagogical. Mayer et al. (1986) and, later, Halpern (1998) both suggested that the kind of structured, transfer-oriented critical thinking testing actually requires is not something most curricula teach deliberately — it is assumed to emerge on its own, which it does not always do.

Where the literature is more encouraging — or at least more detailed — is on what testing work actually demands once someone is doing it. Florea and Stray (2018, 2019), examining postings from more than thirty countries, found employers expecting testers to bring not only test-planning and automation competence but a fairly demanding set of soft skills: communication, analytical reasoning, the capacity to hold up under pressure. That finding echoes a broader strand of research on soft competencies in technical roles. Holtkamp et al. (2015) traced soft-skill expectations across requirements engineering, design, implementation, and testing alike; Joseph et al. (2010) made a related case for "practical intelligence" as an underrated asset among IT professionals; and Sukhoo et al. (2005) argued that technical competence without interpersonal competence tends to break down under real project pressure. Cohen et al. (2004) put a finer point on this, showing that much of the day-to-day friction in testing work is conflict management wearing a technical disguise. Kumar and Hsiao (2007) went so far as to suggest engineers mostly learn such skills "the hard way," through experience rather than instruction — a claim with obvious implications for how testing careers get built. Rivera-Ibarra et al. (2010) tried to formalise some of this into a competency framework for software engineers generally, while Kuusinen et al. (2016) connected developer motivation and flow states to the sustained attentiveness testing seems to require. Garousi and Mäntylä (2016), reviewing the testing literature at a meta level, found the empirical base on tester roles thinner and more geographically lopsided than one might expect — a gap this study is, in its small way, trying to narrow.

That geographic lopsidedness is worth dwelling on. Masuda (2017), writing from a Japanese context shaped by a string of high-profile software failures in financial and airline systems, argued that the gap is not only about workplace practice but about education — only about half of IT and engineering curricula there meaningfully address testing at all. Capretz et al. (2020), surveying senior software students in Malaysia, found the students themselves ambivalent about the field, aware of both its merits and its drawbacks, which hints that whatever image problem testing has starts early, well before graduates ever reach a job posting. What has been studied far less, though — hardly at all, really — is how any of this plays out in South Asia, and in Bangladesh specifically. This matters for reasons beyond completeness. Bangladesh's IT and IT-enabled services sector has grown considerably over the past decade, and software testing there, as Javed (2012) observed in a broader discussion of quality assurance in developing economies, tends to face a distinct set of constraints — tight budgets, compressed timelines, inconsistent standards — that may or may not resemble those reported in more heavily studied markets. Whether they do is, frankly, an open question this study cannot fully settle, but it is one worth raising. Ocholla and Shongwe (2013), in an unrelated but methodologically instructive study of the library and information science job market in South Africa, showed that job-advertisement analysis can meaningfully characterise an emerging profession's requirements even with a modest sample, provided the approach stays transparent about its limits. That precedent, together with Kassab et al.'s (2021) tester-specific methodology in the United States, offers a reasonable template for adapting the same kind of analysis to Bangladesh.

This study, then, asks a fairly narrow but practically useful question: what do employers in Bangladesh actually say they want when they advertise for software testers? Drawing on job postings collected from BDJobs — the country's leading online job portal — over November and December 2022, we examine the education, training, experience, and skill requirements attached to testing roles, organised around the research questions detailed in the following section. The intent is not to offer a definitive account of the Bangladeshi testing labour market; two months of data from a single portal cannot support that, and we would be overstating our case to claim otherwise. What we hope to provide instead is an initial, evidence-based sketch — one that newcomers to the field, career counsellors, and curriculum designers might reasonably use as a starting point, and that future researchers, working with richer and longer-running data, can build upon or, just as usefully, revise.

2. Methodology

2.1 Study Design

This study used a descriptive, cross-sectional job advertisement analysis, a design that has been applied previously to characterise labour market demand for software testers in other national contexts (Florea & Stray, 2018, 2019; Kassab et al., 2021) and, more broadly, to describe emerging professions through job posting content (Ocholla & Shongwe, 2013). Rather than surveying testers or employers directly, we treated publicly advertised job postings as the unit of observation, on the reasoning that a posting represents an employer's own stated—and presumably deliberate—articulation of what they are looking for at the moment of hiring.

2.2 Data Source

Postings were collected from BDJobs (www.bdjobs.com), which is, by most accounts, the most widely used online job portal in Bangladesh and was therefore judged the most appropriate single source for an initial exploration of this labour market, in the same way Kassab et al. (2021) relied on Indeed.com as the dominant U.S. platform. We are conscious that a single-portal design is a limitation, not a strength, and we return to that point later.

2.3 Search Strategy

Postings were retrieved using three search terms, applied directly within BDJobs' search interface: “software tester,” “SQA engineer,” and “software quality assurance engineer.” These terms were selected because they represent the most common formal titles used for testing roles in the reviewed literature (Florea & Stray, 2019; Kassab et al., 2021) as well as in informal observation of the Bangladeshi market. The search was restricted to postings that were live, or had been live, during 1 November to 31 December 2022. To reduce the risk of missing relevant postings advertised under adjacent titles, results were supplemented—manually, not through an automated crawl—by cross-checking company career pages, LinkedIn job listings, and Facebook groups commonly used for IT recruitment in Bangladesh, searching 2.4 for the same core terms plus closely related variants (e.g., “QA engineer,” “test engineer”).

2.4 Eligibility Criteria

A posting was included if it (a) was published or active between November and December 2022, (b) targeted a Bangladesh-based position, and (c) described duties or requirements substantively related to software testing or quality assurance, regardless of seniority level (junior, mid-level, or senior). Postings were excluded if they were duplicates of an already-captured listing, if they referenced testing only tangentially (for example, a general software engineering role that mentioned “some testing” without further detail), or if the posting had been removed before its content could be fully recorded. Applying these criteria across the two-month window yielded a final analytic sample of 31 postings.

2.5 Data Extraction and Coding

Each eligible posting was read in full and manually coded into six categories, adapted from the framework used by Kassab et al. (2021): required education level, training or certification, years of experience, testing-specific skills, technical skills, and soft skills. Coding was performed by the first author directly from the original posting text; where a posting listed a range (for example, “2–5 years”), the full range was recorded and later grouped into the experience bands reported in the Results section. No automated text-mining or natural language processing tools were used; all coding was manual, consistent with the approach taken in comparable prior studies (Florea & Stray, 2018).

We acknowledge—and want to be upfront about—a limitation inherent to single-coder manual analysis: without a second independent coder, we cannot report an inter-rater reliability statistic (such as Cohen's kappa), and some degree of coder judgement is unavoidable, particularly for postings using ambiguous or informal skill descriptions. We discuss this further as a limitation and flag it as a priority for future replications.

2.6 Variables and Operational Definitions

Education level was coded as one of three categories: no stated requirement, bachelor's degree (BSc) in a computing-related field, or postgraduate/other degree. Training was coded as present if a posting explicitly named a certification or training programme (most commonly ISTQB or Certified Scrum Master); otherwise it was coded as not stated. Experience was grouped into three bands— 0–3 years, 3–7 years, and 7 or more years—based on the ranges most frequently observed in the sample. Testing skills, technical skills, and soft skills were each coded as present or absent at the posting level, with the specific skills named (e.g., manual testing, automation testing, SQL, Selenium, communication) recorded verbatim for later categorisation.

2.7 Data Analysis

Because the sample size was small and the data were categorical, analysis was limited to descriptive statistics: postings meeting each criterion were counted and expressed as a percentage of the total sample (N = 31). No inferential statistical tests were applied, and none would be appropriate given the sample size and non-random sampling frame; the percentages reported here should be read as descriptive summaries of this particular sample rather than population estimates. Results were visualised using bar charts (Figures 1–7) to make patterns easier to compare across categories.

2.8 Reproducibility

To support replication, we report the exact search terms, portal, date range, and eligibility criteria above. Researchers wishing to reproduce or extend this analysis on BDJobs or a comparable portal should note that job postings are typically time-limited and may be removed after the application deadline; we therefore recommend that future studies archive the full text of each posting at the time of collection (e.g., as a PDF or screenshot) to allow for later verification or re-coding, a step this study did not formally implement but recommends going forward.

3. Results and Discussion

3.1 Overview

Of the 31-software testing-related postings analysed, roles spanned junior, mid-level, and senior positions, reflecting employers across a range of company sizes and sectors advertising through BDJobs during November and December 2022 (Figure 1). The following subsections address each research question in turn, drawing comparisons with international findings were useful.

3.2 Educational Requirements

Under half of the postings we reviewed, 45.16%, required a bachelor's degree in Computer Science and Engineering specifically, while a nearly equal share, 41.93%, did not specify any particular educational qualification at all (Figure 2). Only 12.90% asked for a master's degree or another qualification. This is a somewhat more relaxed stance than what Florea and Stray (2019) reported in their sample of U.S., Canadian, and Norwegian postings, where 85% required a bachelor's degree outright. Whether that gap reflects a genuinely different hiring philosophy in Bangladesh, or simply a less mature, less standardised job market where formal requirements are stated less consistently, is difficult to say from this data alone—though it is probably some combination of both.

3.3 Training and Certification

Formal training requirements were rare: just 12.9% of postings mentioned any certification, most often ISTQB or Certified Scrum Master (CSM) (Figure 3). This low figure is broadly consistent with Florea and Stray's (2019) finding that only 14% of the postings they examined required testing certification, which is reassuring in a sense, because it suggests the pattern is not unique to Bangladesh but may instead reflect a more general tendency across markets to value demonstrated skill over formal credentialing in this particular field.

3.4 Experience Requirements

More than half of the postings, 54.84%, sought candidates with zero to three years of experience, 32.26% sought three to seven years, and only 12.9% required seven or more years (Figure 4). Read together with the education findings above, this points toward a labour market that is,

Figure 1. Conceptual coding framework for employer demand for software testers. The diagram maps how the overarching construct — demand for software testers, derived from the 31 BDJobs postings analysed — was decomposed into four requirement domains (education, training, experience, and skill), with skill further disaggregated into testing, technical, and soft-skill sub-categories. Arrows denote the hierarchical coding structure used to organise the data, not a causal or chronological sequence. This schematic underlies the category definitions applied throughout the Results section and does not itself report any frequencies.

Figure 2. Educational qualification requirements stated in the 31 software-tester job postings reviewed. Bars show the percentage of postings requiring a Bachelor's degree in Computer Science or Engineering, a Master's degree or other qualification, or specifying no formal educational requirement at all. A Bachelor's-level requirement was the most frequently stated criterion, closely followed by postings that left education unspecified, while advanced-degree requirements were comparatively rare. Percentages are calculated against the full sample (N = 31).

Figure 3. Proportion of postings specifying a formal training or certification requirement, such as ISTQB or Certified Scrum Master (CSM), versus postings that mentioned no training credential at all. Only a small minority of the 31 postings named a specific certification (12.90%), while the large majority set no explicit training expectation. This pattern indicates that Bangladeshi employers, at least in this sample, weigh demonstrated skill more heavily than formal credentialing.

Figure 4. Years of professional experience requested across the 31 postings analysed, grouped into three bands — 0–3 years, 3–7 years, and 7 or more years — and expressed as a percentage of the full sample. Entry- and early-career experience (0–3 years) was requested most often, mid-level experience (3–7 years) was moderately common, and senior-level experience (7+ years) was comparatively rare. The distribution points toward a labour market weighted toward junior and early-career tester recruitment.

at least at present, oriented toward entry- and early-career hiring rather than senior specialist recruitment—which makes some intuitive sense in a testing sector that is still establishing itself as a distinct, respected career path rather than a fallback option (Kassab et al., 2021).

3.5 Testing Skills

Testing-specific skills were specified in 93.55% of postings, making this the most consistently requested category across the entire sample (Figure 5). Manual testing dominated, with automation and performance testing also frequently mentioned. This emphasis on manual testing sits somewhat in contrast to Florea and Stray's (2019) finding that automation testing was requested more often than manual testing in their international sample—a difference that may say something about the relative maturity of automated testing infrastructure and tooling investment in Bangladeshi organisations, though confirming that would require data this study simply does not have.

3.6 Technical Skills

Technical skills appeared in 90.32% of postings, with database skills (MSSQL, MySQL), programming languages (Python, Java, C, C++, C#), and web development skills (HTML, CSS) most commonly named; Selenium was frequently requested specifically for automation testing (Figure 6). SQL's prominence here echoes Florea and Stray's (2019) finding that SQL ranked first among programming skills requested of testers internationally, a nice point of convergence given how different these two labour markets otherwise appear to be.

3.7 Soft Skills

Soft skills were requested in 54.84% of postings, encompassing communication (written and verbal), time management, project management, teamwork, and the ability to work under pressure (Figure 7). This is lower than the 64% reported by Florea and Stray (2018) across their 33-country sample, though the two studies used different coding approaches, so the comparison should be treated cautiously rather than as a firm finding of difference.

3.8 Expected Tasks

Beyond formal qualifications, the postings described a fairly consistent set of day-to-day responsibilities: manual testing, automation testing, performance testing, and user acceptance testing, alongside documentation duties such as writing test plans, test scenarios, test cases, and user manuals. Taken together, these expectations look less like a “second-class” role and more like a genuinely cross-functional one, requiring testers to move fluidly between hands-on technical execution and structured, written communication—a pattern that aligns with the broader argument, made across several of the studies reviewed here, that testing is gradually being recognised as a distinct professional discipline rather than an entry point to something else (Kassab et al., 2021; Masuda, 2017).

3.9 Synthesis

Taken as a whole, these findings suggest that Bangladesh's software testing job market shares some clear similarities with more heavily studied markets—SQL's prominence, a modest appetite for formal certification—while also showing some real differences, particularly a stronger tilt toward entry-level hiring and a continued reliance on manual testing over automation. Given the very small sample and short observation window, though, none of these patterns should be treated as settled; they are, at best, a first approximation worth testing against richer data.

4. Conclusion

This study set out to describe, in fairly plain terms, what Bangladeshi employers say they want from software testers, drawing on 31 job postings collected from BDJobs over November and December 2022. What emerged was a labour market that values practical testing and technical skill—manual testing and SQL above all—more consistently than formal education, certification, or extensive experience, and that appears to be building its testing workforce from the entry level up rather than recruiting heavily at the senior end. These patterns echo, and in places diverge from, findings reported in the United States, Norway, and elsewhere, suggesting that Bangladesh's testing profession is following a broadly similar trajectory while retaining some locally distinct features. Given the modest sample size, short data collection window, and reliance on a single portal, these findings should be read as an initial, exploratory account rather than a definitive one—a foundation, hopefully, for larger, longer, and more methodologically robust research to build upon.

Figure 5. Share of postings that explicitly named testing-specific skill requirements — such as manual testing, automation testing, or performance testing — relative to postings that did not mention any testing skill. Testing skills were specified in the large majority of the 31 postings reviewed (93.55%), making this the single most consistently requested requirement category identified in the study, with only a small residual share omitting any mention of testing competence.

Figure 6. Share of postings that specified technical skill requirements — including database knowledge (e.g., MSSQL, MySQL), programming languages (e.g., Python, Java, C/C++/C#), or web-development skills (e.g., HTML, CSS) — versus postings that named no technical skill at all. Technical requirements appeared in 90.32% of the 31 postings analysed, indicating that testing roles in this market are rarely framed as purely non-technical positions.

Figure 7. Share of postings that named at least one soft-skill requirement — such as communication, teamwork, time management, or the ability to work under pressure — relative to postings that specified none. Just over half of the 31 postings reviewed (54.84%) stated a soft-skill expectation, suggesting that interpersonal competencies, while common, were somewhat less consistently emphasised than testing-specific or technical skills in this sample.

References