Inside the journey: How we brought AI-powered analytics to Tradeshift users
By Anca Andone
Senior Product Manager
Tradeshift
About the Author
Anca Andone is the Senior Product Manager for Data and Analytics at Tradeshift. She is focusing on supporting the strategic growth in the AI and Innovation areas and on integrating AI tools for operational excellence.
Five lessons from migrating a legacy analytics tool to an embedded new app powered by Amazon QuickSight and Amazon Q
👉 Discover how Tradeshift is revolutionizing AI analytics with Amazon QuickSight and Amazon Q.
Lesson 1: Rebuilding data trust from the ground up
Our legacy analytics tool, Insight Center, had its quirks. It restricted scheduled reports to 25MB, and required manual work to keep data pipelines healthy. But it was familiar, and familiarity breeds trust.
So when we introduced the new Reporting & Analytics app with embedded dashboards and reports built in Amazon QuickSight, even with better performance and design, users were quick to point out differences in numbers. This wasn’t a bug—it was a trust gap.
What we did:
- Created a “metrics mapping and side-by-side comparison” to highlight how key metrics are reported in the old vs. the new analytics solution.
- Documented data definitions and aligned filters, logic, and joins.
- Launched internal “AskProduct” sessions with stakeholders to walk through the reports and show how metrics are calculated.
Lesson: Even accurate data will be questioned if it “feels” different. Bring users along early and validate together.
✨ For a complete breakdown of features, download the Reporting & Analytics datasheet.
Lesson 2: Untangling years of tracking debt
Our original analytics app evolved over time, with tracking implemented by different teams using inconsistent naming and parameters. Some events were no longer in use, others were duplicated with slight variations.
What we did:
- Conducted a complete audit of tracked events and dashboards.
- Rebuilt a clean tracking plan in the new system focused on high-signal events.
- Took the opportunity to simplify metrics and remove outdated KPIs.
Lesson: Migration is the perfect moment to clean house, but do it with a scalpel, not a hammer. Preserve what’s valuable and shed what’s not.
Lesson 3: Managing historical data expectations
Insight Center provides historical data for many years in the past, while Amazon QuickSight, the technology behind our Reporting & Analytics app, provides only 2 years of historical data.
What we did:
- Allow Insight Center to run in parallel with the new Reporting & Analytics app.
Lesson: Users often want “just one number”, but making that number feel consistent over time requires data strategy, not just tooling.
Lesson 4: Rethinking the embedded experience
Embedding Amazon QuickSight and Amazon Q inside the Tradeshift platform wasn’t a 1:1 translation of old functionality. We had to rethink access controls, navigation, and how analytics showed up in the user journey.
What we did:
- Designed an Analytics App Store experience that let users launch dashboards from a familiar interface.
- Used role-based views to show buyers and sellers only what mattered to them.
- Embedded Amazon Q’s natural language capabilities to lower the barrier to data access through conversational AI.
Lesson: Embedded analytics is about context, not just convenience. Put insights where users already work.
✨ AI has transformed from a mere upgrade into a fundamental necessity. The promise of truly autonomous finance lies in its ability to condense processes that once consumed weeks into a matter of seconds.
Lesson 5: Driving change across a global organization
What we did:
- Offered early access ahead of launch for early feedback.
- Delivered live Q&A sessions and async onboarding materials tailored to both technical and non-technical users.
Lesson: Tooling change is easy. Behavior change is hard. Build momentum with community, not mandates.
Final Thoughts: Migration is a Product, Not a Project
Looking back, we didn’t just replace one analytics tool with another, we actually shipped a better product experience for data.
That means:
- Thinking about user journeys, not just dashboards.
- Treating trust as a feature, not a given.
- Planning for iteration after launch, because analytics use cases will keep evolving.
We’re still early in our new Reporting & Analytics app’s journey, but the foundation is strong. And the payoff is real: improved performance, smarter self-service, and a dramatically better experience for both our buyers and sellers. We have switched from the question “What tool should we use?” to “What experience do we want to deliver?”.
You may also be interested in
AI-powered Reporting & Analytics app, gives buyers unprecedented control over AP data
This is Tradeshift’s new, advanced business intelligence app, available now, and powered by Amazon QuickSight. It transforms how buyers interact with big data data, processing millions of data points into visual insights within seconds. The app features 11 pre-built dashboards and reports (with more to come), covering a wide range of AP data.
Payment Predictor for Sellers: Your new favourite report
Sellers on the Tradeshift platform all want to know when they’ll get paid. Now you can get answers anytime with predictions from this new ML-powered report, embedded in the Tradeshift platform. This ML-powered Payment Predictor for Sellers report forecasts when invoices are likely to be paid. The report categorizes invoices into time frames (e.g., to be paid in 15 days, 30 days, etc.) and displays the confidence level of each prediction.
From OCR to Autonomous Finance: Building an AI-Native AP Platform
When it comes to AI, we’re way past the theoretical. I remember just a few years ago, in 2021, when we were discussing with businesses about Autonomous AI on the Tradeshift platform, it felt like something out of a distant future. The market just wasn’t ready for it. As a forward-thinking company, our clients often share our mindset. We had some trailblazers who decided to use Ada, our AI coding layer for AP automation, trusting her with up to 99.5% of their processes, even in those early days.
AI-powered insights: The new Reporting & Analytics App at a glance
Artificial intelligence is rapidly transforming how finance teams operate. From accounts payable to procurement to supplier collaboration, business leaders are turning to AI to simplify complex processes and improve decision-making across the board. At Tradeshift, we’ve been on this journey since 2017, steadily embedding AI into the fabric of our platform to support a more autonomous, efficient, and data-driven future. This year we’ve taken another step forward with the Reporting & Analytics app: a smart, embedded analytics solution that brings AI-powered reporting directly to users across our global business network.