× Varley
Varley × Searchflex

Protecting and growing search through the Liquid migration

A replatform puts existing search performance at risk, and it is the one point where URL and template decisions can still be changed cheaply. This sets out where we can help before, during and after the move.

In short

Organic search brought 933,693 clicks to the three storefronts over the last 12 months. That is what the migration puts at risk. Google Search Console, August 2025 to July 2026

Three things need deciding before templates are written: collection taxonomy and URL structure, pagination, and the redirect map.

Most of what follows is optional, and most of it is after launch. What we’re asking for now is a short session on the migration timeline.

Prepared for Gaby v1 · August 2026 Private link · not indexed
Why this document exists

Varley is moving from Hydrogen back to Liquid. That decision creates a window where a number of things become cheap to fix — and, if they’re missed, expensive to retrofit once the new templates are live.

This is a plain list of what we can take on, where we’d add most value, and what we’d recommend happens in what order. It’s meant as a reference rather than a proposal.

What the migration puts at risk

Twelve months of organic search, across three storefronts

Organic clicks over the 12 months to July 2026. This is the traffic the new build has to carry across intact.

577,044
varley.comUS and rest of world — 82% United States
264,675
uk.varley.comUnited Kingdom storefront
91,974
eu.varley.comEuropean Union storefront

The three storefronts don’t move as one. Each has its own URL patterns and its own hreflang relationships with the other two, and Liquid hands that routing to Shopify Markets rather than handling it in code. All three have to survive the move together.

Source: Google Search Console · August 2025 to July 2026

Where we’d fit around the migration

Three windows: before, during and after the move

Before — pre-build

Set the target structure before templates are written

Anything unresolved when the build starts gets carried into the new templates and has to be fixed twice. Most of the URL change is set by the platform rather than chosen — Liquid serves collections, products, pages and articles at fixed paths. The choice is whether that change is designed once, now, or discovered during build and redirected twice.
  • Agree the collection taxonomy and URL structure for the new build
  • Settle the pagination strategy before it’s coded, not after
  • Identify which filtered and faceted URLs currently earn traffic — Liquid’s native filters canonicalise to the parent collection, so any that matter have to be rebuilt as real collections
  • Confirm whether the three storefronts stay separate or consolidate under Markets — that decision sets the URL structure for all three
  • Establish the performance baseline we’ll measure the migration against
During — build & launch

Make sure nothing is lost in the move

This is the part that decides whether organic performance holds. We run this as a governed checklist rather than a review at the end, because most migration losses are only visible weeks later.
  • Export and re-home the existing redirect rules. They currently live in the Hydrogen app and stop existing at cutover; anything not rebuilt in Shopify’s redirect table is lost silently
  • Old → new URL mapping, resolved to single hops where the platform allows, with the exceptions agreed up front rather than found in the crawl
  • Locale routing and hreflang rebuilt for Liquid. Hydrogen handles market routing in code; Liquid hands it to Shopify Markets, so the URL pattern and hreflang cluster for each storefront need rebuilding and verifying before cutover
  • Render testing on every template, so crawlers see the content
  • Structured data preserved and re-validated across product, collection and article
  • Analytics, dataLayer and checkout tracking rebuilt and reconciled against real orders
  • Go / no-go checklist before launch, then daily monitoring through the first fortnight
After — growth

Use the new platform for the things the old one blocked

A cleaner template layer makes a lot of previously awkward work straightforward.
  • Keyword-led collection and landing pages for demand you don’t currently cover
  • Template-level CRO and structured A/B testing
  • Content programme built around the categories worth owning
  • International and market expansion across the UK, EU and US storefronts
How we run a migration

Every stage has a gate that has to be green

The three windows above are where we sit relative to your build. These five phases are how we run our side of it — 01 and 02 sit in Before, 03 and 04 in During, 05 in After. Nothing progresses to the next stage until the previous one is signed off. At launch we give a clear go or no-go against the checklist. The call is yours — our job is to make sure it’s made with the risks written down rather than discovered afterwards.

Phase 01

Discovery & baseline

  • Full crawl, URL inventory and the baseline every later phase is measured against
Gate — baseline agreed
Phase 02

Build & parity

  • On-page, schema and content parity, template by template
Gate — parity mapped
Phase 03

Staging & QA

  • Schema, tracking and Core Web Vitals validation
  • Shopify has no conventional staging environment, so we agree the QA method up front rather than compressing it into launch week
Gate — staging signed off
Phase 04

Launch / cutover

  • Redirects live with the deploy, smoke test run, rollback ready
Gate — go / no-go
Phase 05

Stabilise & grow

  • Index coverage, 404 rate and rankings tracked by template, reported at 30 / 60 / 90 days
Gate — measured against baseline at 30 days
Running now, on another site

We’re part-way through a replatform for another ecommerce client on exactly this process. It runs to 120+ tracked tasks across 15 workstreams, each with an owner, a priority and a launch dependency. The workstreams cover redirects, on-page parity, structured data, content migration, image hosting, rendering, performance, apps and integrations, tracking and consent.

That project runs in the opposite direction to yours — into headless rather than out of it. The platforms differ; the places a migration loses traffic don’t. We can share the redacted task register and a signed-off gate document if that’s useful.

Source: Searchflex migration programme, live engagement 2026 · client anonymised

How this works alongside your build

We specify, they implement, we verify

We’re not looking to duplicate whatever your development team is already doing. Their job is to build the platform; ours is to make sure the build doesn’t cost visibility and that the commercial opportunities get designed in rather than bolted on.

In practice that means we specify, they implement, and we verify. Where it’s more useful for us to build a piece ourselves, we can do that instead.

Design and build

We can specify the work or do it

Your build team owns the platform. This section only applies where you tell us there’s a gap. A replatform is a design and front-end project as much as a technical one, and these are the pieces of it we can take on. Recent examples of each are available on request.

01 · lightest

Audit & recommendations

A UX, SEO and CRO review of the current templates with a prioritised list of what to fix and why.

Output
Findings and a prioritised backlog
02 · structure

Wireframes

Low-fidelity, greyscale layouts of the key templates. Hierarchy, content order and components resolved before any visual design.

Output
Annotated wireframes per template
03 · visual

High-fidelity designs

Full, on-brand visual designs in Figma. Responsive across breakpoints, real content, every state considered.

Output
Figma designs, all breakpoints and states
04 · validate

Interactive prototype

A clickable prototype of a key flow — checkout, a product finder, faceted browsing — that stakeholders can sign off and that can be user-tested before we build.

Output
Shareable clickable prototype
05 · build

Front-end build

We build the templates and sections in Liquid, either handing your developers a production-ready front end or pairing with them to move faster.

Output
Production-ready coded front end
06 · scale

Design system

A reusable component library and design tokens, documented so your team can extend the site consistently long after launch.

Output
Component library, tokens and docs
What we can help with

The full list

Not all of it is needed, and not all of it is needed now.

Technical SEO 01

Protecting what the site already earns.

  • Migration planning and governance
  • Redirect mapping and QA
  • Crawl, index and render testing
  • Site speed and Core Web Vitals
  • Structured data and rich results
  • Log file and crawl budget analysis

Structure & content 02

Covering demand the current structure can’t reach.

  • Collection and URL taxonomy
  • Keyword and demand research
  • New collection and landing pages
  • Product and category copy
  • Editorial content programme
  • Internal linking strategy

CRO, design & dev 03

Turning the traffic into revenue — and building it if you want us to.

  • Template-level conversion work
  • A/B testing programme
  • Checkout and cart optimisation
  • Front-end implementation support
  • Accessibility review

Data & measurement 04

Knowing what actually happened, before and after launch.

  • GA4 and GTM implementation
  • Ecommerce event and dataLayer build
  • Server-side tracking
  • Consent mode and compliance
  • Reporting dashboards
  • Attribution and forecasting
Suggested next steps

Four things, in order

  1. We propose three times for a 30-minute session on the migration timeline, so we know when the structural decisions get made
  2. You confirm what’s worth clearing before the build starts
  3. We send the migration checklist and URL mapping approach for review
  4. We agree which growth workstreams to pick up after launch