Imagine a tile online store. The catalog has more than a thousand items, and the filters are done properly: format, surface, color, country, stock. A visitor who is already on the site finds matte porcelain tile 60 by 120 in half a minute and is pleased. Another person types that exact phrase into search. They land on the main catalog page, where they are invited to start from the beginning: find the filter panel, choose the format and the surface, and hope the result matches what they were looking for.
The filter did its job perfectly, only its job was a different one. It helps the person who is already inside. The person who is only searching needs a page with an address, a heading and content that a search engine can find and show. A sign hangs in the tile shop window, and a consultant works on the floor. The consultant will not replace the sign, and the other way around. What follows is how to combine these two roles in one catalog.
Two jobs of one product selection
Selection by attributes solves two jobs that are easy to mix up. The first is navigation. A person is already on the site, they need to narrow the choice, and the more flexible the filters, the more convenient it is for them. There can be as many combinations as you like, and each of them makes sense only for the person who chose it right now.
The second job is an answer to an external query. Here you need a stable page that exists independently of the visitor’s actions: it has its own address, its own heading, a description and a product selection. It can be found in search, linked from an article or an ad, and returned to a month later.
A filter works on the first scenario, a landing page on the second. Sometimes they show the same products, but they are responsible for different things. When this difference is not stated at the start, the catalog usually tilts to one side: either it is convenient to look around in and impossible to find something through search, or search finds hundreds of nearly identical pages.
What happens to search when there are many filters
The most common way filters enter search results is parameters in the page address. In its guide to faceted navigation, Google describes the problem directly: there can be a great many combinations of criteria, each one produces a new address, and the crawler cannot tell in advance whether it is useful. In the end it spends time on useless addresses and finds new needed pages more slowly. The page was updated on 5 March 2026, but the problem itself has been known for a long time.
Google offers the solution in two variants. If filtered addresses are not needed in search, it is better to limit their crawling. If they are needed, addresses should be built carefully: separate parameters with a standard ampersand, always list filters in the path in one order, and return a 404 for combinations with no results. A canonical address and nofollow are named there as less effective methods over a long distance.
For addresses with parameters that do not change the page content, Yandex has the Clean-param directive. Yandex recommends it instead of a ban through Disallow, because that way the accumulated signals are passed to the main address. The help also explains the cost of a mistake: extra parameters create duplicates, the crawler goes through the site more slowly, and important changes take longer to reach search. Before adding a parameter to the directive, it is worth checking that it really does not change the page. Otherwise the addresses you need will leave search together with the duplicates.
When a combination of attributes needs its own page
A separate landing page by attributes is justified when there is demand for it. If people search for “matte porcelain tile 60 by 120” and nobody types “green tile from Spain in stock for a bathroom”, the decision is obvious. It is better to check this against search engines’ query data. Going through every filter combination will not help here, and intuition often fails too.
The second condition is that the page must have its own content. A product selection, a heading, a couple of paragraphs about how this format differs and where it is used. If a page differs from the next one only by a couple of cards, a search engine will very likely see a duplicate in it. I think a useful check sounds like this: can I write at least a short text for this page that fits no other page.
The third is data. A landing page by an attribute rests on that attribute being filled in the same way for every product. If half the tile has the surface “matte” and the other half “mat”, the filter will show an incomplete list, and so will the page. Order in the data is needed by both scenarios. A page for search simply reveals the disorder sooner.
Catalog structure before the first thousand products
Sections, attributes and pages for search are easier to lay down before filling the catalog than to rebuild on a live catalog that already has traffic. If you are planning an online store or reassembling an old one, I start with the data structure: which properties the filters will need, which combinations deserve their own page and how to keep that up.
How this works in a Tilda catalog
Product filters on Tilda are set up in the “Catalog” module. According to the help on filters, selection can be made by variant properties, attributes or other parameters, for example by stock or category, and search and sorting by name, price and time added can be turned on as well. Attributes are set on the product card, and filters are built from them. So order in the attributes directly decides how useful the filters turn out.
For search structure, Tilda has catalog sections and subsections. According to the help on sections, a section can have its own page with products, a name and a description, which the ST340 block can insert dynamically. Subsection nesting goes up to ten levels, and a section can be hidden in the filters or in the navigation. Tilda’s blog separately announced addresses for sections, subsections and product links, including for indexing. I did not find the announcement date on the page.
So a section is closer to a landing page, and a filter is closer to navigation. The combination “porcelain tile 60 by 120”, with clear demand, is logical to make a section or a separate page. “Matte, 60 by 120, from Italy, in stock” can stay a filter. If addresses with filter parameters start showing up in Yandex Webmaster reports as duplicates, Tilda’s SEO settings include control of the Clean-param directive. Which exact parameters appear in the address when a filter is chosen is better checked on your own published catalog.
Catalog search or filters: what to give the visitor
Internal search and filters also serve different people. Search is for the person who knows the name, the SKU or the collection and wants to get to the product at once. Filters are for the person who chooses by parameters and does not yet know what the tile they need is called. A good catalog usually has both, and the main tool becomes the one your buyers use more often.
If selection is more complex than ordinary selection by properties, for example if compatibility has to be taken into account or consumption calculated, that is already custom work. It helps to remember the boundary from the first part. However smart the selection logic is, it works for the person who is already on the site, and it does not replace search structure.
Back to the tile store. It does not need to choose between filters and landing pages. It needs to decide which combinations deserve a sign in the window and which remain the consultant’s work on the floor. After that, a person searching for matte porcelain tile 60 by 120 lands straight on the right page. The filter stays for those who come into the store to look.