How AI Redirects Combine Pattern-Based Wildcards and Machine Learning

A wildcard rule catches a whole family of URLs sharing a structure in one entry, while machine learning matching finds the best individual replacement for a 404 that does not fit any known pattern. Neither approach alone covers everything a real store's broken links look like, which is why they work best combined.

This guide is for merchants who want to understand how pattern-based wildcards and machine learning matching complement each other, and when each approach is actually doing the work behind a given redirect.

Quick Answer

Yes, pattern-based wildcards and machine learning matching serve different but complementary roles within AI Redirects. SC Easy Redirects uses redirect patterns to automatically catch entire families of URLs sharing a predictable structure, while machine learning evaluates individual 404s that do not fit a known pattern, finding the best replacement based on content and structural similarity. Together they cover both the systematic and the one-off cases.

What This Involves

Combining wildcards and machine learning means using redirect patterns for systematic, predictable URL families, an old category prefix, a discontinued handle structure, while machine learning handles individual 404s that do not follow a recognizable pattern, evaluating each on its own textual and structural merits to find the best replacement.

Who Needs This

• Merchants with both systematic URL changes and one-off broken links
• Stores that have restructured URL formats at some point
• Any business wanting comprehensive 404 coverage, not just individual fixes
• Teams wanting to understand which mechanism is actually catching which 404s
• Businesses wanting efficient handling of large predictable URL families

Why It Matters for Your Business

• Wildcards efficiently catch entire URL families in one rule
• Machine learning covers the individual cases wildcards cannot anticipate
• Relying on only one approach leaves genuine coverage gaps
• Combined coverage handles both the systematic and the unpredictable
• Understanding which mechanism applies helps you audit and trust the results
• This layered approach scales efficiently as a catalog grows and changes

How AI Redirects Combine Pattern-Based Wildcards and Machine Learning on Shopify

Step 1

Start by identifying your systematic versus individual 404 patterns.

• Review your 404 list for URLs sharing an obvious structural pattern
• Separate these from 404s that appear genuinely one-off
• Note any historical URL format changes that created predictable patterns

Step 2: Install and Configure SC Easy Redirects

Install SC Easy Redirects and set up both mechanisms.

• Create redirect patterns for the systematic URL families you identified
• Let machine learning matching handle the remaining individual 404s
• Confirm patterns are not overly broad, avoiding unintended matches

Step 3

Verify the two mechanisms are not conflicting.

• Check that a pattern rule does not inadvertently override a better individual match
• Confirm pattern-caught URLs are not also generating separate machine learning suggestions
• Review edge cases where a URL could plausibly fit either mechanism

Step 4

Test both mechanisms against real 404 examples.

• Test a pattern rule against several URLs it should catch
• Test an individual 404 against the machine learning matching directly
• Confirm both produce sensible, correct destinations

Step 5

Maintain the combined system as your store evolves.

• Add new patterns as new systematic URL changes occur
• Let machine learning continue handling the long tail of individual cases
• Periodically review whether new patterns should be extracted from repeated individual matches

Examples & Use Cases

Store With a Historical URL Format Change (Consumer goods)
Problem: Thousands of 404s from an old category URL structure needed handling alongside a smaller number of unrelated individual broken links
Setup: Set up a redirect pattern catching the old category structure in SC Easy Redirects, while machine learning handled the remaining individual 404s
Result: The bulk of 404s resolved through the efficient pattern rule, with the machine learning layer covering everything else

Multi-Brand Retailer (Retail)
Problem: Relying only on individual machine learning matching missed the efficiency of catching an entire discontinued brand's URL family in one rule
Setup: Added a wildcard pattern for the discontinued brand's URL prefix, freeing machine learning to focus on genuinely individual cases
Result: Coverage became more efficient and comprehensive across both systematic and one-off broken links

See more case studies.

Best Practices

• Identify which 404s are systematic versus genuinely individual
• Use patterns for predictable URL families sharing a structure
• Let machine learning matching handle the individual, unpredictable cases
• Keep patterns scoped tightly to avoid unintended matches
• Check for conflicts between pattern and individual matching
• Test both mechanisms against real examples before relying on them
• Extract new patterns when individual matches reveal a repeated structure

Summary

Pattern-based wildcards and machine learning matching cover different kinds of 404s, systematic URL families versus individual unpredictable cases, and combining them gives comprehensive coverage neither approach achieves alone. The core steps are identifying which of your 404s are systematic versus individual, setting up patterns for the former, and letting machine learning handle the rest.

If your 404s include both predictable families and one-off cases, SC Easy Redirects can handle both with patterns and machine learning working together.