Starting Simple
BrandKit began as a quick-turn project intended to let users add a logo and a few brand colors to their emails and landing pages. The initial scope was small and meant to fill a clear gap — users wanted their content to “look like them” without having to customize every template.
But early usability sessions quickly changed the conversation. Users didn’t just want logos and colors. They wanted the whole brand — fonts, imagery, and even tone — to be consistent across everything they made. That feedback shifted BrandKit from a simple utility into a potential cornerstone of the Constant Contact experience.
The Tension Between 'Fast' and 'Right'
Product management pushed for speed to capture early engagement, while design saw an opportunity to build something foundational and sticky. The compromise was to narrow scope but raise execution quality.
We removed the font-scraping functionality from the first release to simplify development and reduce risk. But we kept the focus on what mattered most: automatically pulling logos, colors, and imagery from a user’s website to help them build consistent, professional campaigns with minimal effort.
The Core Challenge: Mapping Colors Without Losing Users
The hardest part of the project wasn’t scanning the website — it was what came after. Once the system extracted colors, we needed to map them intelligently to different areas of templates. Done wrong, it could make users feel trapped in a mismatched design, forcing them to fix colors manually and breaking flow.
We tested multiple approaches to automatic color assignment. Early language models and mapping systems showed promise, but it was too early to train a model, and our current mapping structure was cumbersome for a light flow. Some overfit to brand palettes and failed when colors lacked contrast. Others worked technically but didn’t look good in real content.
Ultimately, we developed a hybrid approach:
- The system automatically mapped the most reliable colors to key template areas like backgrounds and accents.
- For ambiguous mappings, users received a simple interface they could use to adjust their palette manually afterwards.
This balance gave users a strong starting point without removing control, and it prevented the feature from feeling unpredictable.
Impact Was Measurable
BrandKit shipped as a focused but flexible experience. Users could scan their website, import logos and imagery, and see their branding applied instantly across email and page templates.
User Feedback
“It pulled most of my colors perfectly.”
“It felt like I had a designer helping me.”
“I actually feel like my emails represent my business.”
The lesson: automation is most successful when it feels personal.
What I Learned
This project taught me that speed and quality don’t have to compete — but they do have to be negotiated. Building BrandKit was an exercise in balancing immediate product goals with long-term design value.
We shipped on time, delivered real impact, and set the foundation for a feature that can keep growing. More importantly, it proved that automation succeeds when it gives users momentum without taking away control.
What Still Needs Work
There are still a few edge cases where the brand scanner doesn’t perform as expected. These are being tracked and prioritized based on frequency as support requests come in. One consistent issue involves SVG logos, which the scanner currently can’t process. When that happens, the system defaults to using the first image it successfully detects as the logo.
Engineering is developing a converter that automatically generates image files from SVGs to prevent users from having to update assets manually.