Cutting hours of unnecessary construction admin
Selecting a subcontractor typically involves manually restructuring multiple quotes into a like-for-like format. It's often hours of copy-paste from PDFs into Excel before analysis even begins.
With BidLevel, users upload quotes, generate a structured breakdown, and produce a first-pass price comparison in minutes, with AI.
Enabling full editing controls to fix extraction errors in-product
After extraction, users valued the ability to review pricing alongside highlighted source documents. However, when errors appeared, they had to export to Excel to correct them, losing the benefit of side-by-side review.
To make edits, we assumed excel-native users expected familiar behaviour. After prototyping the necessary interactions, we rebuilt the UI and underlying data schema to support seamless, keyboard-driven editing.
Introducing credits to drive feedback and validate value
Our early user base wasn't providing enough consistent qualitative feedback, despite multiple survey attempts. We also needed to test whether users valued the product enough to make a form of payment during free beta.
We introduced a simple credit system, with feedback as the primary top-up mechanism. Feedback response rates increased from 2% to 30%, giving us significantly stronger signals on product quality and perceived value.
Designing and iterating on the marketing site weekly
We needed a web presence to test go-to-market assumptions. In a single sprint, we shipped a live landing page capable of capturing leads.
From there, I made iterations each sprint based on paid acquisition learnings, refining messaging, hierarchy, and layout to improve conversion.
Creating a fast, flexible visual identity
As BidLevel needed to attract a broader market than ProcurePro's existing ICP, I created a distinct but lightweight identity that worked across product and marketing.
The goal was speed and consistency, a system simple enough to evolve rapidly, but structured enough to maintain cohesion.