Why EdTech Companies Are Hiring More Data Analysts in 2026

Typical learner events table (one row per action)

event_id | learner_id | course_id | event_type      | event_time
---------|------------|-----------|-----------------|---------------------
1        | L101       | C12       | video_started   | 2026-11-01 19:05:10
2        | L101       | C12       | video_completed | 2026-11-01 19:21:44
3        | L101       | C12       | quiz_submitted  | 2026-11-01 19:30:02
4        | L205       | C12       | video_started   | 2026-11-01 20:10:15

Almost every edtech metric is built from tables like this.

-- Where do learners stop watching within a course?
SELECT
  lesson_no,
  COUNT(DISTINCT CASE WHEN event_type = 'video_started'   THEN learner_id END) AS started,
  COUNT(DISTINCT CASE WHEN event_type = 'video_completed' THEN learner_id END) AS completed,
  ROUND(100.0 *
    COUNT(DISTINCT CASE WHEN event_type = 'video_completed' THEN learner_id END) /
    NULLIF(COUNT(DISTINCT CASE WHEN event_type = 'video_started' THEN learner_id END), 0), 1)
    AS completion_pct
FROM lesson_events
WHERE course_id = 'C12'
GROUP BY lesson_no
ORDER BY lesson_no;

-- Weekly retention by signup month (PostgreSQL)
WITH firsts AS (
  SELECT learner_id, DATE_TRUNC('month', MIN(event_time)) AS cohort
  FROM lesson_events
  GROUP BY learner_id
),
activity AS (
  SELECT DISTINCT e.learner_id, f.cohort,
         FLOOR(EXTRACT(EPOCH FROM (e.event_time - f.cohort)) / 604800) AS week_no
  FROM lesson_events e
  JOIN firsts f USING (learner_id)
)
SELECT cohort, week_no, COUNT(*) AS active_learners
FROM activity
WHERE week_no BETWEEN 0 AND 8
GROUP BY cohort, week_no
ORDER BY cohort, week_no;

import pandas as pd

funnel = pd.DataFrame({
    "channel": ["Search", "Social", "Referral"],
    "leads": [1200, 2500, 400],
    "demos": [300, 350, 160],
    "enrolments": [90, 70, 64],
})
funnel["lead_to_demo_%"] = (100 * funnel.demos / funnel.leads).round(1)
funnel["demo_to_enrol_%"] = (100 * funnel.enrolments / funnel.demos).round(1)
funnel["lead_to_enrol_%"] = (100 * funnel.enrolments / funnel.leads).round(1)
print(funnel.sort_values("lead_to_enrol_%", ascending=False))
# (illustrative numbers for practice)

# Did a new reminder message improve next-day return? (illustrative data)
from statsmodels.stats.proportion import proportions_ztest

returned = [412, 468]      # control, new reminder
sent     = [2000, 2000]

stat, p = proportions_ztest(returned, sent)
print(f"control {returned[0]/sent[0]:.1%}, new {returned[1]/sent[1]:.1%}, p = {p:.3f}")
# If p is small and the lift matters in practice,
# recommend the new message - and keep monitoring.

Course health dashboard (for academic teams)

Top KPIs:
- Active learners this week
- Course completion rate
- Average quiz score

Charts:
1. Lesson-by-lesson completion (find drop-off lessons)
2. Quiz score distribution per module
3. Weekly active learners trend

Filters: course, batch, language, device
Insight box: 'Lesson 6 has the steepest drop-off'

Project: Improving course completion for an online course

Data: public or synthetic learner events
(learner_id, lesson, event_type, timestamp, score)

Questions:
1. Which lessons have the biggest drop-off?
2. Do quiz scores predict completion?
3. Which signup cohorts retain best?

Deliverables:
- SQL queries
- One dashboard
- One-page recommendation memo

import numpy as np
import pandas as pd

rng = np.random.default_rng(42)
rows = []
for learner in range(1, 501):
    stop_at = rng.integers(1, 11)          # lesson where learner stops
    for lesson in range(1, stop_at + 1):
        rows.append({"learner_id": learner, "lesson": lesson,
                     "score": int(np.clip(rng.normal(70, 12), 0, 100))})

events = pd.DataFrame(rows)
events.to_csv("synthetic_lessons.csv", index=False)
print(events.groupby("lesson").learner_id.nunique())

To: Head of Academics
Subject: Lesson 6 drop-off

Finding: Completion falls most sharply at lesson 6.
Evidence: Lesson-wise completion chart, quiz scores.
Likely cause: Lesson length and a difficult concept
introduced without practice.
Recommendation: Split lesson 6 into two parts and add
a practice quiz; test with one batch first.
Measure: Lesson 6 completion and module quiz scores.