The Discipline of Asking in Order. What EdTechs lose when they skip ahead?

Most EdTech start-ups can describe what they've built. Far fewer can describe, with confidence, why it works, who it’s for and how they know it works. The Brookings Centre for Universal Education’s global catalogue of EdTech companies featured 1,640 EdTech solutions from around the world and found that only 11% had been externally evaluated and 18% internally evaluated, leaving over 70% of the EdTech innovations in their catalogue that either did not make evaluation data publicly available or were not evaluated.

That evidence gap is more than a data problem; it's a sequencing problem. The questions that would have surfaced the evidence needed for effective evaluation were never asked, or were asked too late. 

Why order matters and what gets skipped? 

Based on this idea, Injini developed a roadmap that highlights the types of customer validation and impact-tracking questions that matter at different stages of business development (e.g., ideation, MVP, and growth). 

This roadmap is available here.

This roadmap reflects how confidence in a product's effectiveness is built iteratively, and how each stage's answers become the raw material the next stage depends on.

Two things happen as a result of treating evidence generation in this way: 

1. Evidence has to be proportionate to the stage. A pre-seed EdTech citing a randomised controlled trial is not further ahead; it is answering a question it does not yet need to ask, usually at the expense of foundational questions still sitting unanswered. Similarly, what constitutes “good” evidence, or “good enough” evidence, evolves with the business.  
2. Earlier questions don't retire once answered. They get revisited as the product, user base, and context change. A solution validated with one context can mean little within a new geography, language, or connectivity context. Cumulative, not sequential-and-discard, is the operative logic for successful product development.

EdTech start-ups will often skip the same handful of questions. Two common omissions are early desirability and early impact framing.

• Desirability (if users want to use a product) often gets collapsed into usability (if users can use a product). Teams confirm that users understand and can operate the product, and treat that as validation, without asking whether users want to keep using it once the novelty or the incentive wears off. The result shows up months later as a retention problem that looks like a marketing or onboarding issue when it actually is an unanswered question from the prototype stage. Another common mistake is that EdTech teams collect feedback only from their most engaged users, with few talking to those who dropped off early. This skews their understanding of whether people actually want the product.
• Build towards evidence generation from the start, but focus on the right things first: Some EdTech teams try to evaluate impact without first defining success, for example via a Theory of Change. Others move forward with impact measurement before they've confirmed the product itself is easy enough to use; however, you cannot credibly claim outcomes before you know the product works well enough for anyone to experience them consistently. Some start generating evidence too late, only getting serious about impact evidence when a funder or investor asks for it. But without early clarity on what success looks like, and early signs that they're on the right track, there is nothing solid to build rigorous evidence on.

What does this cost?

The consequences of this are inefficiencies both within individual EdTech companies and across the ecosystem more broadly.

• Money can be spent scaling a version of the product that was never confirmed to be wanted, meaning a costly pivot happens after the infrastructure, sales cycle, and team have already scaled around it.
• Money and time are spent on evaluations that don't provide answers or show evidence of outcomes because the appropriate groundwork wasn't laid. This can be the case even for very impactful solutions
• Equity gaps surface late and are expensive to fix. If no one asked early who is being reached and who is not, that question tends to resurface only after scale. At this point, excluded users are a structural feature of how the product is sold and built, not a fixable bug.
• Schools and system partners absorb the switching cost. They adopt on the strength of claims that later prove thin, and the trust cost of that lands on the sector's credibility more broadly, not just on the one EdTech.

Thus, at Injini we argue EdTech teams need to treat evidence generation as a discipline. Validation and impact questions should be built into the EdTech product roadmap from day one, answered at the depth appropriate to that stage, and revisited deliberately as the business changes.

‍