More pages doesn't mean more revenue
Your product catalog could generate thousands of possible category pages. But building all of them wastes resources on pages that won't convert. The question isn't “what pages could we build?”; it's “which pages will drive the most revenue?”
No demand, no customers
Many product combinations look plausible but nobody searches for them. Building pages without validating real customer demand means investing in content that will never generate a sale.
Overlapping pages split your traffic
If a new page targets the same searches as an existing one, they compete for the same customers. Instead of one strong page, you get two weak ones that both underperform.
Poor product fit loses sales
A category page with only a few loosely-related products disappoints customers who clicked expecting a real selection. They leave without buying and don't come back.
How we identify pages worth building
Every potential page runs through five checks. Each one answers a clear business question. Pages that pass all five are worth creating. Pages that fail show you exactly why (see all rejection reasons), so you can focus your resources on the opportunities that matter.
Are customers searching for this?
We validate each topic against real search data: your Google Search Console plus third-party sources. No gut feelings. If customers aren't looking for it, we don't recommend building it.
Do you have enough products to satisfy them?
A page for "red leather sofas" is disappointing if you only stock one. We check that every recommended page has enough genuinely matching products to create a useful shopping experience.
Are you already winning this traffic?
If you already have a page performing well for this search, creating another one splits your success. We only recommend pages that fill a genuine gap in your coverage.
Would this compete with an existing page?
Two pages targeting the same customer need will undercut each other. We catch overlaps even when the wording is completely different, because customers and search engines treat them the same.
Will the products actually match?
False matches happen: a brand name mentioned but not sold, a material in the title but not the product. Pages with irrelevant listings disappoint customers. We catch those before you build.
From thousands of possibilities to the ones that matter
A lighting retailer with 8,000 products could build 12,000 different category pages. Here's how we narrow that to the 1,000 that will actually drive revenue.
From 12,000 possibilities, we recommend 1,000. Each one has customers looking for it, enough products to satisfy them, and no existing page to compete with. Those are the pages worth your investment.
Your rules, your decisions
The sieve makes recommendations. You make the final call. Every decision is transparent, configurable and reversible.
Override any decision
See a rejected topic you want to build anyway? Move it to approved with one click. Launching a new product line that changes the calculus? You decide what makes sense for your business.
The sieve shows you the data behind each decision, with a clear rejection reason for every topic it filters out. You add the business context it can't know.
Configure your thresholds
Set your own rules for what counts as “enough demand” or “enough products.” A luxury retailer values every qualified visitor differently than a mass-market store.
Adjust by category, by season, or by strategic priority. The sieve adapts to how you want to grow.
Start with full oversight. Review every recommendation until you trust the output. Then move to automatic once you're confident in the results.
Manual review vs automated sieve
Most teams review topic lists by hand in spreadsheets. The topic sieve replaces that entire process with automated, data-backed decisions.
Without the topic sieve
- ×Manual review of thousands of topics in spreadsheets (weeks of work)
- ×No systematic check for search demand; decisions based on gut feeling
- ×Duplicates slip through because you can't compare intent across thousands of topics
- ×Pages created for topics with too few products, leading to poor UX
- ×Already-ranking pages get cannibalised by new pages targeting the same intent
With Similar AI's topic sieve
- The topic sieve filters thousands of topics (results in hours, not weeks)
- Every topic validated against real search demand from Google Search Console
- Intent-level deduplication catches overlaps even when wording differs
- Minimum product thresholds ensure every page has enough relevant listings
- Ranking checks prevent cannibalisation of pages that already perform well
Why you can trust this
No black box. Every decision comes with clear reasoning you can inspect, question and override. You see exactly why each topic passed or failed, with the data to back it up. See the full list of rejection reasons.
Not enough customers looking
Fewer searches than your threshold. You decide what counts as "enough"; a luxury retailer values every visitor differently than a mass-market store.
Not enough products to show
Customers expect a real selection. A page with one or two items isn't a category; it's a disappointment waiting to happen.
Would compete with existing page
Another page already serves this need. We catch when two pages would fight for the same customers, even when the wording looks different.
You're already winning here
An existing page already performs well for these searches. Building a new one would split your success instead of adding to it.
Products don't truly fit
The products look like matches but aren't. A brand name mentioned in passing, a material in the title but not the product. We catch false matches before you build.
Your business rules
Add your own criteria. Block brand combinations, exclude seasonal topics outside their window, or set different thresholds by category. Your business logic, applied by the sieve.
Part of a complete revenue growth system
The topic sieve works alongside product feed enrichment and page creation. We identify the opportunities from your products, filter them to the ones worth building, then the New Pages Agent creates the pages. Every topic that doesn't pass is tagged with a specific rejection reason so you can review and override.
Once pages are live, performance tracking shows you exactly which pages drive revenue. Underperforming pages get flagged for review. Successful approaches get applied more broadly.
Every page exists because the data supports it. Every decision is traceable. Every result is measurable.
Frequently asked questions
Can we review recommendations before pages get created?
Yes. You can review every recommendation before any page is built, or let the topic sieve run on its own once you trust the output. Most teams start with full review and move to automatic once they're confident in the quality.
Can we override the system's decisions?
Yes. Every recommendation shows its reasoning and the data behind it. You can approve topics the system rejected or reject topics it approved. Your business context matters: launching a new product line, seasonal strategy, or strategic priorities that the data can't capture.
How is this different from keyword research?
Keyword research tells you which terms have search volume. The topic sieve tells you which of those terms you should actually build pages for. A keyword might have high volume, but if you already rank for it, don't have enough products, or would compete with an existing page, building it would waste resources.
Who controls the thresholds?
You do. Set your own rules for what counts as enough demand, enough products, or too much overlap. A luxury retailer values every qualified visitor differently than a mass-market store. Adjust by category, by season, or by strategic priority.
What happens to rejected recommendations?
Nothing is deleted. Every rejected topic is stored with its reasoning so you can review it later. You can see the full list of rejection reasons on our dedicated page. If conditions change (you add products, demand shifts, or an existing page is removed), the sieve re-evaluates previously rejected topics.
How many pages does a typical site end up building?
It varies by catalog size and how mature your category coverage already is. A useful rule of thumb: a site with 10,000 products might find around 1,000 new pages worth building. We'll show you the exact numbers for your catalog in a demo.
See which pages you should be building
Book a demo and we'll run your product catalog through the topic sieve. You'll see exactly which pages are worth building, why each one passes or fails, and the revenue opportunity you're missing.