Building Grade 6-10 Financial Literacy Textbooks for a Private Bank

A listed private bank asked us to build a complete set of financial literacy textbooks for grades 6 through 10, five books in total, running to roughly a lakh words. The brief drew on an existing content bank the bank already owned. The books needed to work as actual classroom material, not as branded content with a bank's name attached to it.

What Made It Difficult

This was the first initiative of its kind by a private bank in India, and the client wanted it to set a standard, not just fill a gap. The client was reasonable about timelines and prioritized quality over speed, which meant the pressure wasn't about racing a clock. It was about getting something right that had no precedent to check against.

The writing itself sat outside our usual register. This wasn't B2B content built to persuade a reader; it was educational content built to teach a child a financial concept correctly, at the reading level of that grade, with none of the marketing instincts we'd normally reach for. We were effectively being asked to write at the level of an established schoolbook publisher, without having published a schoolbook before. There was no existing framework to adapt. The client treated us as the subject matter experts on how to build this, which meant most of the structural decisions, not just the writing, were ours to make.

What We Tried First That Didn't Work

Our first approach was to take the existing content bank and synthesize it down to grade-appropriate versions, book by book. On paper this looked efficient: the raw material already existed, and financial concepts don't change by grade, just the depth. In practice, it read like an edited brochure, not a textbook. It lacked the pacing, the built-in checks for understanding, and the visual and structural rhythm that make something feel like it belongs in a classroom rather than in a bank's marketing folder. We caught this early, before it went far, and dropped the synthesis approach in favor of writing each book from the ground up, using the content bank as a source of facts rather than as a starting draft.

How We Organised the Work

The full set, five books and close to a lakh words, went from brief to delivery in about a month and a half. That pace was possible because the workflow was fixed from the start rather than figured out as we went:

• A writer produced the first draft of each section
• A second writer ran quality control on it
• An editor reviewed it next
• Only then did it move to delivery

On the client side, sign-off came from a designated point of contact, typically across two rounds, with the bank's CMO stepping in on specific sections when something needed a final call. Having that chain fixed upfront, rather than improvised per book, is what let five books move through review without the turnaround stretching out.

What Review Caught

The client's instruction on tone was to "keep it general." Taken at face value, that's a low bar; it would have been easy to write something plain and move on. Instead, we treated it as a floor, not a ceiling, and built the books against a more specific standard, drawing on:

• The Rights of Persons with Disabilities Act 2016
• The National Education Policy 2020
• NCERT's guidelines on inclusive, barrier-free content

In practice, this meant applying Universal Design for Learning principles, so information was presented in more than one way for different learners, and being deliberate about how disability was represented in examples and illustrations, avoiding the pity-based or token framing that shows up by default in generic content.

None of this was explicitly asked for. It came out of recognizing that a bank building a first-of-its-kind national schoolbook initiative couldn't afford to fall short on accessibility even if nobody on the client side had flagged it as a requirement. That gap, between what "general" meant on paper and what it needed to mean in a document going into classrooms, was the most consequential thing we caught before delivery.

Separately, wherever AI tools were used to speed up drafting, human review had to catch a specific, recurring failure mode: repeated phrasing across sections that should have read as distinct, and invented figures that sounded plausible but weren't sourced from anything real. Both are the kind of error that survives a fast read and only surfaces when someone checks each claim against what the content bank actually supported.

What This Means for You

If you're evaluating a content partner for work like this, the useful question isn't whether they can follow a brief. Most can. It's whether they'll notice what the brief left out and act on it without being asked, whether that's an accessibility standard, a regulatory detail, or a gap between what was requested and what the situation actually required. That's usually where the real risk in a project like this sits, not in the parts everyone already knows to check.

FAQs

Why not adapt an existing content bank into grade-level textbooks?

That was the first approach and it failed. Synthesising the existing material down by grade looked efficient because the raw content already existed, but it read like an edited brochure rather than a textbook. It lacked pacing, built-in checks for understanding, and the structural rhythm that makes material belong in a classroom. Each book was rewritten from the ground up, using the content bank as a source of facts rather than as a starting draft.

How do you write educational material without having published a schoolbook before?

By treating the structural decisions as part of the brief rather than assuming them. There was no existing framework to adapt, so the workflow was fixed before drafting: a writer produced the first draft of each section, a second writer ran quality control, an editor reviewed it, and only then did it move to delivery.

What did review catch that the brief had not asked for?

The client's instruction on tone was to keep it general. Taken at face value that is a low bar. The books were instead built against the Rights of Persons with Disabilities Act 2016, the National Education Policy 2020 and NCERT's guidelines on inclusive, barrier-free content, applying Universal Design for Learning principles and being deliberate about how disability appeared in examples and illustrations. None of it was requested.

What has to be checked when AI is used to speed up drafting?

Two failure modes recurred: repeated phrasing across sections that should have read as distinct, and invented figures that sounded plausible but were not supported by the content bank. Both survive a fast read and only surface when each claim is checked against the source.