Edtech content development economics — the hidden line item

How to size content production as a permanent line item

Compute it as a % of revenue, not a fixed budget:

• Years 1-2 (catalogue build): 30-50% of revenue allocated to content. This is high because revenue is low.
• Years 3-5 (catalogue maintenance + selective expansion): 15-25% of revenue.
• Year 5+ (steady state): 8-15% of revenue. The catalogue is mature; ongoing work is refresh + selective expansion + remediation.

If you can't keep content costs below 30% of revenue by Year 3, the business has unit economics that don't compound.

The lever that changes the math: user-generated content

For some edtech categories (skills development, professional learning), high-quality user-generated content can scale at near-zero marginal cost. Examples: Stack Overflow, Khan Academy's community translations, Duolingo's community challenge sets.

This works when:

• The category has natural experts willing to share.
• Quality control is feasible at scale (peer review, voting, editorial layer).
• The product design captures the contribution effortlessly.

When it works, content cost as % of revenue can drop below 5%. When it doesn't (because the category requires curated authoritative content), trying to force UGC produces low-quality content that damages retention.

The other lever: AI-assisted content production

Production cost per finished hour drops materially with AI assistance:

• AI-generated first drafts for written content: 60-80% cost reduction with editor-in-the-loop.
• AI-generated exercises and questions: 50-70% reduction in question-bank development.
• AI-generated video (Synthesia-style avatars): still rough but improving; 40-60% cost reduction for instructor-led content where production value matters less than scale.

The trap: optimising too far on AI cost-reduction degrades quality below what learners accept. The right shape is AI-augmented human production, not AI-replacement production.

Common mistakes

• Treating content as a one-time spend in the financial model. It's perpetual.
• Underestimating the editing-to-production ratio. Each hour of finished video typically takes 5-15 hours of production work. Recording is 20% of the cost; planning, scripting, editing, QA, and updates are 80%.
• Building too much content at launch. A small, high-quality catalogue retains better than a large, mediocre catalogue. Ship 20 hours great rather than 200 hours okay.
• Not budgeting for accessibility from day one. Captioning, transcripts, alt-text for diagrams. Retrofitting these is 3-5x more expensive than building them in.

What to do today

1. Compute your current content cost as % of revenue. If you don't know, you don't have a sustainable model yet.
2. Plan content production as an ongoing line item (15-25% of revenue at Year 3+).
3. Identify whether UGC or AI-assistance levers apply to your category. If yes, invest in the workflow now.
4. Set quality targets per content type and budget against them. Generic budgets ("we'll spend $X on content") miss the per-piece economics.