How EduSpark Academy transformed their testing workflow with Primates

Background

EduSpark Academy is an education technology company based in Portland, Oregon, that provides a comprehensive learning management system (LMS) to K-12 school districts across the western United States. The platform serves over 150,000 students and 8,000 teachers across 45 school districts, offering digital curriculum delivery, interactive assessments, automated grading, progress tracking, and parent communication tools. Founded in 2018 by former educator Jennifer Walsh, the company has grown to 80 employees with an engineering team of just 12 developers.

EduSpark's platform is built on a modern stack: a Next.js frontend, a Python/Django backend, PostgreSQL for transactional data, and Redis for caching and real-time features like live quiz participation. The system handles over 2 million assessment submissions per month during the school year and generates report cards, progress reports, and transcript data that directly affect students' academic records.

The Challenge

Education technology carries a unique burden of responsibility. Unlike consumer applications where a bug might cause minor inconvenience, errors in an LMS can have real consequences for students' academic records, teacher evaluations, and district compliance reporting. EduSpark's small engineering team was acutely aware of this responsibility, but their testing capacity was fundamentally insufficient for the complexity of the platform they maintained.

Several factors compounded the challenge:

• Grading logic complexity: EduSpark supported 14 different grading scales (letter grades, percentage-based, standards-based, pass/fail, and custom district-defined scales), each with its own rounding rules, weighting algorithms, and grade boundary definitions. A single change to the grading engine required validation across all 14 scales, with multiple edge cases per scale (e.g., what happens when a student has zero graded assignments in a category that carries 30% weight?). The team had documented 847 distinct grading scenarios, of which they could manually test approximately 120 per release cycle.
• Content rendering diversity: Teachers created lessons using a rich content editor that supported text, images, embedded videos (YouTube and Vimeo), interactive widgets (drag-and-drop, hotspot, and timeline activities), mathematical notation (LaTeX and MathML), and code snippets with syntax highlighting. This content needed to render correctly across Chrome, Safari, Firefox, and Edge on both desktop and Chromebook devices (the dominant hardware in K-12 classrooms). Rendering bugs were reported an average of 45 times per month.
• Accessibility requirements: Many of EduSpark's school district clients were legally required to provide accessible digital content under Section 508 and IDEA. The platform needed to work with screen readers (JAWS, NVDA, VoiceOver), support keyboard-only navigation, and meet WCAG 2.1 AA standards. Accessibility testing was almost entirely manual and was typically performed only before major releases.
• Assessment integrity: The platform's assessment engine needed to ensure that questions were presented in the correct order (or properly randomized), time limits were enforced accurately, auto-save worked reliably, and submitted answers could never be lost or corrupted. A bug in the assessment engine during the 2024 spring testing season caused 340 student submissions to be recorded with incorrect timestamps, triggering a formal investigation by one school district.

The engineering team's manual testing process consumed approximately 40% of each monthly release cycle. Despite this investment, an average of 8 bugs per release reached production, with 2-3 of those classified as high-severity issues affecting grading accuracy or assessment integrity. Client trust was eroding: annual retention had dropped from 95% to 89% over two years, with departing districts citing "platform reliability concerns" as a primary factor.

The Solution

Jennifer Walsh, EduSpark's founder and CTO, made the decision to invest in automated testing despite the company's limited budget, recognizing that the cost of continued quality issues -- both in lost clients and in potential harm to students -- far exceeded the investment required. After a focused evaluation, the team selected Primates for its ability to deliver enterprise-grade testing capabilities at a price point appropriate for a company of EduSpark's size.

Grading Logic Validation Engine

Primates' data-driven testing framework allowed EduSpark's team to express all 847 grading scenarios as parameterized test specifications in a structured YAML format. Each specification defined the inputs (assignment scores, category weights, grading scale, and district-specific rules) and expected outputs (calculated grades, GPA values, and honor roll eligibility). The platform executed all 847 scenarios in under 90 seconds, compared to the 3 days required for manual testing of the subset the team could previously cover. When a developer modified the grading engine, they received comprehensive feedback within minutes rather than discovering issues days later during QA review.

Cross-Browser Content Rendering Tests

Primates' visual testing module was configured to render representative samples of each content type across 6 browser/device combinations. The platform maintained a library of 280 baseline screenshots representing correct rendering of every content element -- from simple formatted text to complex LaTeX equations embedded within interactive drag-and-drop activities. Each pull request that touched the content rendering pipeline triggered a targeted visual comparison, and the platform generated side-by-side diff images for any detected discrepancies. Content rendering bug reports from teachers dropped by 89% within the first two months.

Automated Accessibility Auditing

Primates integrated with axe-core and custom ARIA validation rules to perform automated accessibility testing on every page and interactive component. The platform tested keyboard navigation paths, screen reader announcement sequences, color contrast ratios, and focus management behavior. While automated tools cannot catch every accessibility issue, the system identified and prevented approximately 85% of the accessibility regressions that had previously required manual auditing. The remaining 15% -- primarily issues related to cognitive accessibility and content clarity -- continued to be evaluated through periodic manual reviews, but with a dramatically reduced scope.

Assessment Integrity Monitoring

Primates' end-to-end testing framework simulated complete assessment workflows: students starting timed quizzes, answering questions across multiple pages, using auto-save, experiencing simulated network interruptions, and submitting final answers. The platform validated that every answer was persisted correctly, timestamps were accurate, time limits were enforced within a 1-second tolerance, and the anti-cheating randomization logic produced statistically valid question orderings. These tests ran against a realistic dataset of 10,000 simulated student interactions, providing confidence that the assessment engine could handle real-world conditions.

The Results

The transformation in EduSpark's quality metrics was remarkable, particularly given the company's small team size:

Regression test automation reached 96%, with 813 of 847 grading scenarios and over 2,400 additional functional test cases running automatically on every code change. The remaining 4% consisted of edge cases that required human judgment (such as evaluating whether a teacher's custom rubric was rendered in an intuitive layout) and were covered during periodic manual review sessions.

Grading accuracy improved to 99.97%, with only 3 grading calculation errors reported across the entire 2024-2025 school year, compared to 34 in the previous year. Each of the 3 remaining errors was traced to ambiguous district-specific grading rules that had been incompletely specified -- not to software defects.

The deployment cadence accelerated from monthly releases to weekly feature drops, with the ability to push critical hotfixes within hours when needed. The team shipped 48 releases in the 2024-2025 school year, compared to 11 the previous year, giving teachers and administrators access to new capabilities faster than ever before.

Most importantly, annual client retention climbed from 89% to 98%. Three school districts that had been evaluating competing platforms renewed their contracts specifically citing the improvement in platform reliability. Two new large districts signed on after reference calls with existing clients who praised EduSpark's quality turnaround.

"As a former teacher, I founded EduSpark because I believed technology could make education better -- not more frustrating. Primates helped us live up to that promise. Our teachers trust the platform again, and I can sleep at night knowing that a grading bug won't derail a student's academic journey."

What's Next

EduSpark is working with Primates to develop specialized testing capabilities for their upcoming AI-powered adaptive learning module, which will dynamically adjust lesson difficulty based on student performance. The team is also planning to leverage Primates' load testing features to ensure the platform can handle the concentrated usage patterns typical of school environments -- where thousands of students in a district might simultaneously log in at 8:00 AM and begin taking assessments at exactly the same moment.