Improve Google Drive

First, I’d clarify the product boundary. Google Drive is the storage and sharing layer for files and learning materials. Tools such as Docs, Slides, Classroom, and Meet support related workflows, but they aren’t all features of Drive.

For this case, I’ll assume we can improve the education workflow around materials stored in Drive, including connections with those tools. That matters because improving live teaching would require more than a change to file storage.

1. Clarify the objective and scope

I’d ask:

• Are we improving adoption, engagement, teacher efficiency, or learning outcomes?

• Are we focusing on schools, colleges, or another education segment?

• Is the experience for physical, remote, or hybrid classes?

• Can we work across connected education tools?

• What are the geography, timeline, and budget constraints?

For this case, we’re targeting existing users in schools and colleges in metropolitan India, with a one-year timeline and no fixed budget constraint. I’ll assume we’re working within Google and can use appropriately permissioned product analytics.

I’d focus on engagement because we’ve chosen institutions already using the tools. I wouldn’t assume that engagement has declined or that every institution is at the same stage of adoption without checking the data.

2. Define what engagement means

Different stakeholders would judge success differently:

I’d focus on students completing meaningful learning activities, supported by timely teacher intervention.

Attendance, recording views, and assignment completion are useful signals. They don’t independently prove learning. Similarly, a score of 85% or 95% isn’t a universal measure of engagement.

3. Choose the primary user

I’d segment the main users into administrators, teachers, and students, with parents as supporting stakeholders in school settings.

I’d choose teachers, because helping one teacher identify and address a confusing explanation could benefit an entire class.

My initial persona would be a teacher in a metropolitan Indian institution who is comfortable delivering lessons in English and moderately familiar with Google’s tools. Those are pilot assumptions, not characteristics I’d assign to every teacher.

Their work has two parts:

• During teaching: present material, answer questions, and adjust explanations.

• Outside teaching: review assignments, maintain attendance, prepare materials, and communicate with families.

I’d focus on the first part: helping teachers respond while learning is happening, rather than discovering every difficulty after an assignment is submitted.

4. Understand the journey and needs

I’d map the journey as:

Prepare materials → Share them → Teach a topic → Check understanding → Adjust the explanation → Assign practice → Review results

I see four needs:

• Content relevance: Is the explanation appropriate for the students’ current understanding?

• Outcome visibility: Which concepts are students mastering, and where might they need further practice?

• Readiness and understanding during class: Do students need clarification, a different pace, or a short break?

• Operational efficiency: How much teaching time is lost to access, connectivity, audio, and administrative problems?

I’d also investigate whether questions raised during class are captured against the relevant material. Repeated questions could help teachers improve the next lesson.

I wouldn’t assume technical issues consume 20–30% of every class. I’d measure that through observations and teacher feedback.

5. Prioritise the problem

Using basic needs, satisfiers, and delighters, I’d treat access to the lesson and opportunities to communicate confusion as basic needs. Better topic-level feedback would improve the experience further. Predicting distant outcomes, such as an examination result, would be a later and much harder problem.

I’d prioritise helping teachers recognise misunderstanding early enough to respond.

I’d describe this as understanding a student’s response to a particular lesson, rather than measuring a fixed “capacity to learn.” A student may understand one topic and struggle with the next.

6. Explore the solutions

I’d consider the three ideas in this case, with some changes to how they work.

A lesson-readiness and understanding check

The proposed “Mind Sync Score” aims to help teachers decide whether to continue, explain differently, or take a break. That is a useful decision to support.

However, I wouldn’t use facial recognition or camera-derived expressions to assign a cognitive-readiness score. A facial signal would need evidence before it could justify conclusions about understanding, and an arbitrary threshold such as 70% wouldn’t establish when a student needs intervention.

I’d instead collect direct signals: a short comprehension question, an optional “I need another example” response, or a pace check. I’d show the responses and participation level rather than compress them into a seemingly precise mental-state score.

Focus Area Alerts

Link questions, practice errors, and requests for clarification to a topic or section of the teaching material.

For example, if many students struggle with a rotational-dynamics question, the teacher could see the misconception and return to the relevant slide. I’d use a calm, teacher-facing prompt rather than turn the screen red or publicly label students.

Short learning games

For younger students, simple activities could provide practice or introduce a break. I’d measure performance on the activity itself.

I wouldn’t call them psychometric assessments or infer a student’s general concentration span without appropriate validation. Any animated characters would also need suitable usage rights.

7. Prioritise and define the first version

I’d compare learning value, implementation effort, evidence requirements, and teacher workload.

My first version would be Lesson Check-ins, connected to materials stored in Drive.

A teacher could add a question to a section of a lesson, share it during class, and see an aggregate response summary. Students could also flag a confusing section. The teacher could then revisit it, offer another example, or continue.

For instance, a wrong-answer pattern might show that students are confusing mass with weight. That is more actionable than knowing that an engagement score has fallen.

8. Evaluate trade-offs and rollout

Too many checks could interrupt teaching. I’d make them optional and quick to reuse, with teachers choosing when to run them.

Non-response must remain separate from an incorrect answer. A student may have a connection problem or no individual device. Shared-device and teacher-entered class responses would need consideration.

I’d minimise collection, restrict access appropriately, and avoid public rankings. Teachers should see the evidence behind a suggestion and retain control over the response.

Across the year, I’d research the workflow, pilot a small set of subjects, evaluate results, and expand gradually. Existing institution relationships and teacher training could support rollout. I’d avoid requiring teachers to rebuild all their materials.

9. Measure success

My primary metric would be the percentage of participating students completing a lesson-linked activity and demonstrating its intended learning objective.

I’d report participation separately so the metric doesn’t improve simply because struggling students stop responding. I’d also track class completion, but treat a 90% target as something to assess against a baseline rather than assume it is appropriate.

Supporting metrics would include:

• Teacher adoption and repeat use.

• Time required to prepare and run a check.

• Percentage of flagged concepts receiving a follow-up explanation.

• Improvement on a comparable follow-up task.

• Completion of related assignments and use of revision materials.

Guardrails would include added teacher workload, lesson interruptions, student discomfort, access-related non-response, and cases where positive signals are followed by poor task performance or drop-off.

I’d compare learning and participation outcomes with a suitable control group. A correlation between a high score and class completion alone would not establish that the feature works.

My recommendation is to connect teaching materials with quick, explicit feedback about understanding. This preserves the goal behind the original idea while giving teachers clearer evidence for what to do next.