Skip to content

Retail Cloud Connect

Why Search Conversion Drops After a Platform Migration

Search conversion usually drops after a migration because synonyms, redirects, merchandising rules, and filters didn't come across, product data changed shape, and AI models lost the shopper history they learned from. Some of the drop is often broken tracking rather than lost sales.

Portrait of Sebastian Hooker

Sebastian Hooker


Search conversion usually drops after a platform migration because the things that made search work didn't come across: synonyms, search redirects, merchandising rules, filter setups, and the shopper behavior history that AI ranking models learned from. Product data often changes shape during the move too, so tags and attributes your filters relied on no longer match. Part of the drop is frequently broken tracking, which makes things look worse than they are. The same happens when you switch Shopify search apps, not just platforms.

Search quality is built up over years in small pieces. A migration moves the catalog, but most of those pieces live in the old search tool, not in the product data.

What gets lostHow it shows upWhere to check
Synonyms"Sofa" no longer finds "couch" productsYour new search tool's no-results report
Search redirects"Returns" or a brand name lands on a results page instead of the right pageTop queries that used to have one destination
Boost and bury rulesSeasonal or high-margin products drop down the pageYour top 50 queries, checked by hand
Filter treesBrand or attribute filters missing or flattenedCollection and search pages
Behavior historyRankings feel generic, popular products don't riseSearch conversion by query
Old URLsBookmarked filtered pages and campaign links breakInbound links and 404s

Product Data Changed Shape

Migrations rarely move product data one to one. Tags get renamed, metafields get restructured, product types get consolidated, and variant options change. Any filter or rule built on the old structure breaks quietly.

The Bee's Knees British Imports is a good example of how much a store can depend on that structure. Their filter trees were built from Shopify tags with brand prefixes, so an "Emma Ball" product carries a tag like "Emma Ball/Animals" that generates a filter option.

When they moved from Boost to Retail Cloud Connect, Nimstrata imported their existing Boost configuration into the app settings so shoppers saw an identical experience. Retail Cloud Connect can import tags, tag prefixes such as Season:Spring, metafields, and product options into filters, which is how a tag-based filter tree carries across.

Filter limits also change between platforms. Shopify's Search & Discovery app allows a maximum of 25 filters, hides collection filters above 5,000 products, and hides search filters when a search returns more than 100,000 products. If your old platform showed more, shoppers on the new store simply can't narrow results the same way.

AI Models Start Cold

If your old search learned from shopper behavior, that learning stays behind. Google Cloud's AI Commerce Search (formerly Vertex AI Search for commerce) works the same way: every new Retail API customer starts in its first data quality tier, "Relevance," and moves up to "Relevance & Popularity" and then to fully personalized, revenue-optimized ranking as user events accumulate.

Large brands with millions of users move through those tiers quickly. Smaller stores should expect up to two to three months before the full benefit of user events shows up. That gap is where a cold-start dip right after launch comes from. The user events guide explains how events feed both search and recommendations.

Broken Tracking Makes the Drop Look Worse

Before you assume shoppers stopped buying, confirm you're still measuring correctly. After a migration, check for:

  1. Missing pixels. In Shopify, all app pixels are listed on the Customer events page in settings. Confirm your analytics and search app pixels are connected.

  2. Search tracking that counts differently. A new search tool may define a "search session" or attribute conversions differently from the old one, so the before and after numbers aren't comparable.

  3. Bot traffic. Crawlers that hit search and collection pages can inflate sessions, and sessions that never buy dilute your conversion rate. Filter them out before comparing.

  4. Wrong metric. Conversion rate alone can mislead. Revenue per visitor is a better measure for product discovery because search can lift conversion by steering shoppers to cheaper items.

Recovery Checklist

Work through this in order. Tracking comes first because every other fix depends on trusting the numbers.

  1. Fix tracking first so you can trust every number that follows.

  2. Export your top queries from the old platform or analytics, and test each one on the new store.

  3. Recreate redirects. Use Shopify's URL redirects under Content > Menus for old page URLs, with CSV import for bulk changes. Search redirects for queries live in your search tool.

  4. Rebuild synonyms for high-value queries using your no-results report as the list. The synonyms guide covers one-way and two-way rules.

  5. Rebuild filters from the attributes shoppers actually use, and confirm the new tags or metafields are populated.

  6. Remove the old search app's code. Shopify notes that uninstalling an app doesn't automatically remove its code from your theme.

  7. Restart behavior data collection and leave it running. Every day of user events shortens the cold start.

How to Avoid the Drop Next Time

Most of this is preventable if you plan for it before launch.

  • Build on a preview theme. Set up the new search on an unpublished theme while your current search stays live, test your top queries, then switch.

  • Collect events before you switch. Retail Cloud Connect's Core Components App Embed can track Shopify's native search results and collection pages before the store fully switches, so AI Commerce Search learns from existing shopper behavior.

  • Import history. AI Commerce Search can import historical user events into a Google Cloud project from Google Analytics 4, Google Analytics 360, BigQuery, Cloud Storage, or inline API calls. Cloud Storage and BigQuery imports use the Retail user event schema, while GA4 and GA360 imports read BigQuery-linked Analytics data without manual schema mapping.

  • Send events from your old search app. If the existing tool can send user events to Nimstrata's Storefront API, those events accelerate model training before you switch.

  • Carry rules across. Retail Cloud Connect's Search Redirects page imports rules from CSV, up to 1,000 rules.

  • Keep old filter URLs working. URL key mappings translate a previous search provider's query parameter names into Retail Cloud Connect filter keys, so saved collection links and email campaigns still apply the right filters.

  • A/B test the switch. The Bee's Knees ran a five-week Shoplift test against Boost before switching, and AI Commerce Search delivered a 20.58% conversion lift at 99% confidence.

Switching Search Apps, Not Platforms

Everything above applies when you move between Shopify search apps, with two extra risks. The old app's code can linger in your theme, and the old and new result sections can both stay enabled. Retail Cloud Connect's dashboard shows a Duplicate Sections Detected alert when the theme's native results or a second results block is still enabled alongside its own. Rules also change names between tools: boost and bury rules become serving controls in AI Commerce Search, configured in the Google Cloud console, and query redirects move to the Search Redirects page.

A planned switch can pay off quickly. Cotopaxi moved from Fast Simon to Retail Cloud Connect and saw 44.78% higher conversion from search, with its custom theme integration ready for testing within 72 hours. PPD Performance replaced Boost across seven Shopify stores and more than 100,000 SKUs.

If your search conversion dropped after a migration, request a free Search Relevance Report to see which queries and filters are losing sales, or compare plans on the pricing page.

FAQ

Common Questions

Why did my search conversion rate drop after moving to Shopify?

The most common causes are synonyms, redirects, and merchandising rules that weren't recreated, product tags and attributes that changed during the data import, filters that were rebuilt differently, and search models that lost the shopper behavior history they used to rank results. Broken conversion tracking can also make the drop look larger than it is.

How long does search take to recover after a migration?

Rules, redirects, and filters recover as soon as you recreate them. Behavior-trained ranking takes longer. Large stores with millions of users see AI Commerce Search improve quickly, while smaller stores should expect up to two to three months of user events before the full benefit shows up.

Can I import my old shopper behavior data into Google Cloud's AI Commerce Search?

Google Cloud's AI Commerce Search supports importing historical user events into a Google Cloud project from Google Analytics 4, Google Analytics 360, BigQuery, Cloud Storage, or inline API calls. With Retail Cloud Connect, you can also give the models a head start by tracking Shopify's native search and collection pages before you switch, or by sending events from your current search tool to Nimstrata's Storefront API.

Does switching Shopify search apps cause the same problems as a platform migration?

Yes. Moving from one search app to another can lose synonyms, redirects, boost and bury rules, filter configurations, and old filter URLs, and the old app's code can stay in your theme after it is uninstalled.

Will my old filter URLs keep working after switching to Retail Cloud Connect?

They can. Retail Cloud Connect's URL key mappings translate query parameter names from a previous search provider into its own filter keys, so saved collection links and email campaign URLs keep applying the right filters.

Get Started

See Google’s AI Commerce Search on your store

Install the Shopify app, or book a demo to see it on your catalog.