Parcel Perform

A shopper in Berlin just asked an AI where to buy boots. Parcel Perform's data decided the answer.

Somewhere this week, someone typed "best winter boots that actually arrive on time" into ChatGPT or Gemini, and the assistant answered with a brand name. The shopper assumed that was a hunch. It wasn't. Behind that recommendation sits a scoring system most retailers have never heard of, built from logistics data most retailers never thought to clean up - and Parcel Perform built it.

This is the company in its present tense: a Singapore-built platform processing more than 100 billion parcel updates a year, threading data across 1,100-plus carriers in over 160 countries, used by roughly 3,000 e-commerce brands from Shopify to TikTok to Nespresso. It spent its first decade fixing the unglamorous plumbing of online shopping - tracking pages, delivery dates, return portals. Then, in February 2026, it launched something new: a public index ranking which retailers AI shopping assistants actually recommend, and why.

Every carrier speaks a different language. Nobody had bothered to translate.

Here's the part most shoppers never see: when a parcel moves from a warehouse in Shenzhen to a doorstep in Stuttgart, it usually passes through three or four different carriers, each logging its own version of events, in its own format, in its own time zone, in its own language. One says "out for delivery." Another says "in transit - last leg." A third says nothing useful at all until the parcel has already arrived. multiply by 1,100 carriers

That's not a delivery problem. It's a translation problem wearing a delivery costume. And it explains the now-famous acronym every customer service team dreads: WISMO, "Where Is My Order?" - the single most common, most expensive, most preventable support ticket in e-commerce.

The tension only sharpened with the arrival of AI shopping agents. Those systems don't read marketing copy and they don't care about brand voice. They read operational signals - delivery speed, return friction, customer sentiment - and decide, often silently, who gets recommended and who gets skipped. A retailer with excellent products and chaotic logistics data is, to an AI assistant, functionally invisible.

Two people who'd already lived the chaos decided to standardize it first, and worry about selling it second.

Parcel Perform was founded in Singapore in 2016 by Dr. Arne Jeroschewski and Dana von der Heide, who had both spent years inside the e-commerce machinery they would later try to fix. Von der Heide's background includes time at Zalora, the Southeast Asian e-commerce platform - a vantage point that exposed exactly how much operational noise sits between "order placed" and "order received," and how little of it customers ever needed to see.

Their bet was unfashionable at the time: don't build a flashier tracking widget. Build the unglamorous data layer underneath it - the one that makes sense of 1,100 carriers saying 1,100 different things - and let the product follow from there. It's a slower path to a demo, and a much harder one to copy.

One data foundation, six ways to use it

The platform reads as modular - checkout, tracking, returns, logistics - but every module draws from the same underlying spine: a standardized, cleaned, harmonized read on what's actually happening to a parcel, in 155-plus event types and 36-plus languages.

The numbers a logistics company is allowed to be smug about

Plenty of SaaS companies talk growth. Fewer can point to a customer roster that includes Shopify, TikTok, Nespresso, and Puma, or claim profitability before their first big funding round - which is roughly the position Parcel Perform was in heading into its 2021 Series A.

Puma reports the platform lets it "provide highly personalized customer responses." Sleep brand Emma credits it with giving the company "visibility, control, and data to optimize" post-purchase operations. None of these are headline numbers built for a pitch deck - they read more like quiet operational relief, which is usually the more believable kind.

Found by AI. Chosen by shoppers. In that order, now.

Parcel Perform's own framing of its mission has shifted, slightly but tellingly, in the last year: from helping brands satisfy customers to helping brands satisfy customers and the AI systems now standing between brands and customers. The company describes the goal as building delivery experiences that get a brand "found by AI, chosen by shoppers" - which sounds like a slogan until you notice it's also an org chart of who decides what gets bought.

The gatekeepers of shopping just changed. Somebody had to build the proof layer underneath them.

It's easy to treat "AI shopping assistant" as a novelty feature bolted onto a search bar. It's harder, and more accurate, to treat it as a new layer of commerce infrastructure - one that will reward brands who can show their work and quietly bury brands who can't. Parcel Perform's wager is that this layer needs the same thing e-commerce logistics needed a decade ago: a standardized, trustworthy data foundation, built before everyone else realizes they need one.

Whether that wager pays off depends on something that hasn't fully settled yet - how much weight AI platforms ultimately give to operational trust signals over raw popularity or price. But the company that already solved the unglamorous translation problem behind a billion parcels a month has a head start most competitors don't: it isn't starting from zero data. It's starting from a decade of it.