Why customers can't find what they want
Your site has thousands of products that should connect to each other. But manual linking breaks down quickly, and customers get lost.
Merchandising can't keep up
With thousands of products changing seasonally, your team can't manually curate which pages should link to which. New arrivals get isolated. Best sellers get over-promoted.
Obvious connections get missed
“Midi Skirts” and “A-Line Skirts” are related, but they don't share keywords. Traditional rules-based systems miss these connections that shoppers expect.
Links go stale fast
Products come and go. Categories evolve. Manual links break or become irrelevant. Customers click expecting one thing and find another, or nothing at all.
How the Linking Agent connects related categories
The Linking Agent uses deep semantic understanding to identify which category pages belong together. It builds vector embeddings of each page and measures cosine similarity to find genuine connections, not just keyword matches. Links stay fresh as your site changes.
Tell us which pages to connect
Select which category pages should show related links and which other pages they can link to. You control the scope: link within a department, across departments, or wherever makes sense for your shoppers.
The agent builds a semantic map
The Linking Agent creates vector embeddings for each category page, capturing a deep semantic understanding of what each page is about. It then measures cosine similarity to find genuinely related pages. 'Midi Skirts' connects to 'A-Line Skirts' because the pages share unmet needs, even though they don't share words.
You set the quality bar
Choose how closely categories must relate before showing a link. High threshold means fewer, more targeted suggestions. Lower threshold means broader discovery. Preview everything before it goes live.
Connections stay current automatically
New category pages get connected to relevant pages immediately. Retired pages disappear from suggestions. Seasonal trends shift connections naturally. No manual upkeep required.
What customers see
A shopper on your “Long-Sleeved Dresses” page should see related categories they're likely to browse next. Manual curation takes hours and misses obvious matches. Similar AI finds them instantly.
Manual approach
- •Merchandising picks a few related categories
- •Popular pages get featured, long-tail forgotten
- •New arrivals aren't connected for weeks
- •Links don't match how customers actually shop
Customers browse in circles, not finding what they want.
With the Linking Agent
- Midi Dresses with Sleeves (closely related)
- Maxi Dresses Long Sleeve (similar style)
- Casual Day Dresses (same occasion)
- Work Dresses (professional alternative)
- Long Sleeve Tops (complementary category)
Customers discover products they actually want to buy.
“The most important benefit is that we were able to deliver seven-figure annual revenue growth with amazing ROI. Using Similar AI pages also saves significant time and allows us to quickly respond to market trends and consumer behavior.”
Jennifer Skeen
VP of eCommerce, Visual Comfort & Co.
You control how closely products must relate
Different businesses need different approaches. A luxury brand wants precise, curated suggestions. A discovery-focused marketplace wants broader exploration.
Precise
For curated experiences
Show only the most closely related products. Best for luxury brands, specialized retailers, or pages where irrelevant suggestions would feel off-brand. Fewer links, but each one feels hand-picked.
Balanced
For most retailers
A mix of closely related and adjacent products. Customers see obvious matches plus some discovery options. The sweet spot for most e-commerce sites with 1,000+ products.
Discovery
For exploration-focused sites
Encourage browsing across your catalog. Good for marketplaces, large catalogs, or when you want to surface more of your inventory. More links, broader connections.
Works for any product catalog
The Linking Agent understands how shoppers think about products in your industry. Whether you sell fashion, furniture, or lighting, it finds the connections customers expect.
Fashion and apparel
Connect attributes like dress styles, sleeve lengths, occasions and materials. A shopper browsing 'Long-Sleeve Dresses' might see 'Midi Dresses with Sleeves', 'Maxi Dresses with Sleeves' and 'Casual Day Dresses': semantically similar categories.
Home and furniture
Link attributes such as room categories, materials and styles. For example, 'Washable Rugs' connects to 'Kitchen Rugs', 'Entryway Rugs' and 'Pet-Friendly Rugs': the practical groupings homeowners care about.
Lighting and electrical
Connect attributes like fixture types, rooms, styles and finishes. For example, 'Pendant Lights' links to 'Kitchen Island Lighting', 'Dining Room Chandeliers' and 'Modern Pendants': how designers and homeowners browse.
Part of a complete navigation system
Related Links is one way the Linking Agent helps customers navigate. It also provides Boost Links to prioritize your best-converting products, and Nearby Links to connect location-based pages.
When the New Pages Agent creates new category pages, the Linking Agent connects them to related products immediately. No manual work required.
The result: customers always find pathways to products they're likely to buy, across your entire catalog.
“Lyst used Similar.ai to launch a new internal linking module in a week to let search engine users find the pages they love. In a few weeks, we measured the impact, showing that we had significantly grown the ranking of these pages.”
Simon Dance
Chief Commercial Officer, Lyst
“The ability of Similar AI to quickly adapt ensures our customer experience remains seamless. This has been a standout feature.”
Jennifer Skeen
VP of eCommerce, Visual Comfort & Co.
Frequently asked questions
What ROI can I expect from related product connections?
Visual Comfort achieved $2.4M in new annual revenue with 29x ROI over 12 months. The exact impact depends on your catalog size, current navigation gaps, and how customers browse your site. Our growth calculator can give you a personalized estimate.
Does this require engineering work?
Most brands only take a couple of hours of dev time. After this, links across the site stay up to date without engineering effort.
How does the agent know which categories are related?
The Linking Agent builds vector embeddings of each category page and uses cosine similarity to find deep semantic connections. 'Midi Skirts' connects to 'A-Line Skirts' because both pages serve similar unmet needs, even though the words don't overlap. It's semantic understanding, not keyword matching or clickstream data.
What if I don't like the suggestions?
You preview everything before it goes live. Adjust the quality threshold to show more or fewer connections. Override specific suggestions if needed. You stay in control; the agent just does the heavy lifting.
What happens when I add new products?
New products automatically get connected to related pages on the next update cycle. Discontinued products disappear from suggestions. No manual work required: your navigation stays current as your catalog changes.
How quickly will I see results?
Traffic changes typically appear within 2-4 weeks as customers and search engines discover the improved navigation. Full impact usually visible within 60-90 days. We provide measurement dashboards so you can track progress.
See which products the Linking Agent would connect
Book a demo and we'll show you the product connections your site is missing. Real data from your catalog, no commitment.