21phi builds and runs the analytics layer for Shopify Plus and Hydrogen storefronts, from pixels and event taxonomy to attribution, cohorts and a weekly review, and stays embedded as the team that answers the next question.
Shopify data analytics is the tracking, reporting and analysis layer that tells a store which products, pages, channels and customers drive orders. 21phi builds and runs that layer for DTC brands on Shopify Plus and Shopify Hydrogen, for teams who want numbers they can trust and a senior person who explains what changed each week.
Most stores have analytics. Few trust it. Tags fire twice or not at all, checkout events never reach the ad platforms, and the dashboard disagrees with Shopify. On a headless storefront the gap widens: Liquid theme tooling does not carry over.
We fix that at the source. GA4 and GTM are Google Analytics 4 and Google Tag Manager; 21phi prefers pixels in code over GTM-fired tags on Hydrogen. Shopify pixels, Shopify's sandboxed tracking layer for checkout and storefront events, cover the checkout. Amplitude, a product analytics platform (events, funnels, retention, session replay) 21phi uses for storefront analysis and A/B test reads, is where the team reads the results.
The same clean layer is what makes AI ecommerce analytics work: store data connected to the brand's own AI assistant, so a founder can ask a question in plain language and get an answer from their own numbers. Our article AI for Ecommerce Analytics explains how. If you only need a GA4 Shopify setup checked once, we will say so on the first call.
An event taxonomy is the named set of tracked storefront events and their properties. We write it before any tag ships: product views, color and size selection, add to cart, checkout steps, account actions and filters, each with the properties that make it useful by segment. The names are plain, so marketing, customer care and operations recognize what they later see in reports.
On a Hydrogen storefront the events live in code and are tested in staging like every other change. An event reaches production once it is in the document.
Attribution is matching orders to the ad click or channel that drove them; it depends on pixels, click IDs and server-side events surviving from storefront to checkout. A headless build adds joints to that path, and each can drop data.
We own the path end to end. Pixels live in storefront code rather than a tag-manager container, event flow is verified against the attribution tool, and when a click ID stops short of the purchase event we trace where it drops and fix it, in code or server side. Test orders in each market confirm the fix, alongside the client's ads team.
We do not claim perfect attribution. Ad platforms, GA4 and Shopify count differently, and some discrepancy is expected. We explain the differences so the team can plan against them.
A weekly analytics review, run by us. Each week we walk the team through what changed in the funnel, product views, add to cart, checkout and repeat purchase, by segment, and what to test next. Sergey Gridin, co-founder of 21phi, leads the review and the engineering behind it, so the person presenting the numbers also owns the tracking.
RFM segmentation, grouping customers by Recency, Frequency and Monetary value to target and analyse retention, is built from storefront, order and product-analytics data. Cohort analysis, following a group of customers acquired in the same period to measure repeat purchase and retention over time, shows whether customers won during a promotion come back. A/B test results are read the same way, by segment, never from an average.
The clean layer above is the precondition for AI. Once events, orders and customers agree, we connect them to the brand's own AI assistant through MCP, an open standard for connecting assistants to business tools, so a founder or marketer can ask why last week was slow and get an answer grounded in the store's own numbers. This is Shopify AI analytics as a working tool, not a feature in a dashboard.
Metric definitions are agreed in advance and kept in writing, so revenue, conversion and customer value mean the same thing however the question is phrased. When the data is incomplete, the answer says so.
Analytics is not a setup project. New campaigns, apps and markets each break something, and the team that owns the tracking has to be reachable when they do. 21phi stays embedded after launch as the team behind the store. We sit in the client's own channels, and a request from any department, a broken pixel, an integration that stopped sending events, a landing page that needs tracking before a drop, an urgent fix, lands with the same senior people who built the storefront.
This costs more than a body shop, and it should. The alternative is a chain of vendors and a founder who reconciles them. One embedded team that owns the storefront, the tracking and the weekly read removes that work, and the numbers stay trustworthy because the people changing the storefront maintain the events.
Brands come to us with a new Hydrogen or Shopify Plus build that needs analytics designed in, a live store whose numbers nobody trusts, or data nobody reads. Each engagement is scoped individually, starting from the questions the team cannot answer today. The analytics layer of a Shopify Plus apparel storefront we run shows the whole stack in production.
21phi is a Shopify Hydrogen agency based in Toronto, working with brands across North America. We are a Shopify Partner with Shopify Plus experience, and you will know the names of the people on your account. Shopify data analytics is usually one part of a wider engagement with conversion rate optimization (CRO) and A/B testing and ongoing support and maintenance, because the read and the change belong together.
A Shopify Hydrogen storefront for a men's performance apparel brand, built around product systems and run by 21phi as the embedded storefront, analytics and growth team.
Shopify Plus · Shopify Hydrogen on Remix / React Router · Oxygen edge hosting · Storefront API and Customer Account API · Shopify Markets · Shopify web pixels and ad pixels, implemented in code · Amplitude for funnels, retention and session replay · Mux for video
A complete Shopify data analytics setup has four layers: GA4, Shopify pixels and ad pixels implemented in storefront code; a documented event taxonomy naming every tracked interaction; attribution verified from storefront to Shopify checkout; and a product analytics tool such as Amplitude for funnels, retention and session replay. 21phi adds a weekly review so the data is acted on.
On a Liquid theme, GA4 is usually fired by Shopify's Google integration or a tag manager. A Shopify Hydrogen storefront is custom code, so 21phi implements GA4 and ad pixels directly in the storefront rather than through GTM-fired tags, and uses Shopify pixels for checkout events. Events are tested in staging and documented before they reach production.
Ad platforms, GA4 and Shopify disagree because each system counts differently. Ad platforms attribute orders to clicks within their own windows, GA4 models sessions and consent gaps, and Shopify counts every order regardless of source. Some gap is expected. Technical losses can be fixed: click IDs that drop between storefront and checkout, missing purchase events, duplicate tags. 21phi fixes those in code or server side and explains the rest.
AI ecommerce analytics means connecting a store's own data to an AI assistant so the team can ask questions in plain language and get answers grounded in those numbers. It is reliable only when the tracking and metric definitions underneath are clean, which is why 21phi builds the event layer first and agrees definitions in writing. When the data behind a question is incomplete, the answer says so.
A weekly analytics review covers what changed in the funnel, from product views through add to cart, checkout and repeat purchase, by segment such as new versus returning customers, market or device. It names what to test or fix next and who owns it. At 21phi the review is led by Sergey Gridin, co-founder, as part of the ongoing support engagement.
21phi audits an existing Shopify analytics setup without touching the storefront build: the GA4, tag manager and Shopify pixel configuration is reviewed, events are compared against orders, attribution is checked from storefront to checkout, and anything missing or double counted is documented. The output is a prioritized fix list and an event taxonomy the team can keep.
21phi turns funnel data, UX friction and merchandising gaps into a disciplined A/B testing program on Shopify, read by segment, and stays embedded with your team to act on what it finds.
21phi designs, builds and keeps running headless storefronts on Hydrogen, Oxygen and the Storefront API for Shopify Plus brands, and stays embedded with your team as the people who own the store after launch.
The real work starts after launch.
