Programmatic style guide

Overview

Knowing how to describe a brand’s tone, style and persona, and applying that consistently to writing can get tricky. Our team built a comprehensive tone analyzer tool to capture, analyze, and apply brand voice globally across content generation that included:

A smart system to automatically create and update style guides

Ability to analyze and save different writing styles

A fun analyzer that matches your brand voice to a “trading card”

I designed the frontend, prioritized what features would go into the MVP and wrote the feature specs.

Key features

1️⃣

A sample of a brand archetype, with an editable, saveable description

Style analysis and analyzer: I used the IKEA effect to help users feel more connected to their brand voice—when a user partially creates something, they value it more.

Paste or import document feature for canonical style analysis

Automatic style extraction and saving

1️⃣

Ability to save canonical style preferences

Programmatic style guide: Reduces time spent on manual enforcement and updates. The ability to saved enforced writing rules, which were integrated with the data processing pipeline and generation workflow, like:

Topic and word avoidance lists

Importing brand dictionary terms, like names and trademarks

1️⃣

Flexible tone switching: Users can quickly change between different writing styles for different channels and personas.

Results

People spent 50% less time managing their style guides

The content produced was 35% more consistent in its brand voice

Takeaways

It was interesting thinking about about how to create levels of abstraction for reusable components of a style guide.

From a user perspective, companies have a wide range of formalization of their style guides and perferences (some people really care about the oxford comma).

It was a balance between surfacing enough different customization options to prompt the user to formalize their guidelines into rules, and not overwhelming them with too many options.

From a backend perspective, we aimed to build flexible and extensible components that could be reused with minor changes to add on new customization options.

Future improvements

More options for programmatic enforcement: punctuation, capitalization rules, international style guides

Style guide importing and automatic conversion to saved rules

Implement AI-driven assistant corrections in line with style guide

Create a collaborative feature for team-based style guide management