Manufacturer Pages — Reasoning & Recommendations

01TL;DR

• On the Tennant page, 252 machine tiles, plus the same 252 models again in the sidebar, come before any part. The first part sits 11,870px down on a desktop and 26,645px down on a phone (33 screens).
• Price is the only filter.
• Browsing stops at 10,000 products.
• Every one of the ~399 manufacturer pages uses the same template; Nilfisk-Advance renders 782 tiles.

Three directions, each one template for every brand:

• Step 1: native Category and in-stock filters as quick wins.
• Step 2: HawkSearch landing pages, once dashboard access arrives.

A and C scale with models and brands, so they can ship in weeks. B scales with the catalog. This is the same reasoning as PD-008.

Before and after, measured at 1440x900 and 375x812:

02Current state

Measured on the live site on 2026-09-29, at 1440x900 and 375x812.

The Tennant page today

The same template on every manufacturer page

What the rest of the site already does

The finder's lists and the category tree do not match (292 manufacturers against about 399 brand pages; 248 Tennant models against 252 tiles), so reconciling them belongs with the finder fix.

Fix first (any direction)

These are defects, not design choices. They go to the theme developer and the merch team now, whichever direction is picked.

1. The homepage finder builds its links from display names. Tennant Company plus T500E goes to /parts-and-supplies/Tennant-Company/T500E/, which returns 404.
   • The real page, /parts-and-supplies/tennant/t500e/, works. A near-empty "Tennant Company" category (1 model, 0 products) makes the brand-level link resolve while the model-level one fails.
   • Of 11 other manufacturer names checked, 8 return 404 the same way, including Floorcare.biz (promoted beside the finder on the homepage), Clarke (Alto), and APA and Kent (Nilfisk-Advance). 50 of the 292 finder names don't match a brand address.
   • Fix: link to the category the shopper picked (each option already carries its id), and retire the stray category.
2. Clean up the model list:
   • merge the duplicates, with 301 redirects: T1-B to T1B; V BP 10 and V-BP 10 to V-BP-10; V BP 6 to V-BP-6; V-WA 30 to V-WA-30
   • move the charger out of the list, and re-point or remove the off-brand "355" tile
3. Make the OEM flag agree with the "Aftermarket" names.
4. Fix the "Browse by & Price" label and the "Tennat" logo alt text.

Who lands here

From the Clarity data in the IA roadmap (March 2026):

• Tennant is the top non-search page (119 sessions).
• 100% of organic browse sessions go brand to model.
• 25.6% of sessions end in a quick back.
• 37% of traffic is mobile.
• Search is 32% of sessions.

A manufacturer page serves six visitors:

1. Knows the machine: gets to the model page in one step. This is the dominant job.
2. Knows the brand but not the model: recognizes the machine by type or photo, finds the model plate, or asks.
3. Has a part number: searches within the brand.
4. Restocks consumables across several machines: shops by part type, and chooses OEM or aftermarket.
5. Wants documents or service: diagrams, manuals, a technician. This is what the heading promises.
6. Arrives from Google on "Tennant parts": needs to see what the store has for the brand and where to go next.

03References

Three parts sellers with the same brand-page problem, checked live on 2026-09-29.

Parts Town, Hobart page (foodservice equipment parts). A filterable parts listing with brand content on top:

• a Model filter, and a searchable Part Category filter with counts
• a "How to find your Hobart model and serial number" link
• a brand-and-model "Partpredictor", and search by serial number in the header
• list rows showing the maker's part number, their own part number, "In Stock, Ships today", a quantity box and add to cart

The closest to B with a slice of C.

Repair Clinic, Amana page (appliance parts). Search first:

• one large field for "part number, model number or problem", and a "Help me find my model number" link
• above the results: "Multiple models matched. Select your model to confirm fit"
• 70,813 results with search within and category filters; cards say "Genuine OEM Part #"

Direction B with A's fit prompt.

Sweepscrub, Tennant page (floor-care parts, a direct competitor). A directory:

• an intro paragraph that names popular Tennant models by machine type ("Rider Sweepers like the S20 & S30")
• a grid of model cards with no filters; the page runs 7,460px tall
• its navigation labels models by type ("Tennant S20 Ride on Floor Sweepers")

Direction A without the finder.

The pattern across all three: each page names the model-number problem out loud, with a help link or type labels, and none makes the visitor scroll a model grid to reach a part.

USA Clean's own /parts-and-supplies/ A-Z brand index is the cleanest browse page on the site, and the pattern Direction A borrows for its older models.

04What BigCommerce Stencil gives us

• Category sort order. Every subcategory has a sort_order, set by drag order in the control panel or through the Catalog API. It puts popular and current models first with no code.
• Product Filtering (Pro and Enterprise plans).
  • Filter types: Category, Brand, Price, Rating, Product Options, Custom Fields, and "Other" (in stock, free shipping, featured).
  • Already switched on for this store, with Price as the only filter configured.
• The Contextual Filters API (/v3/settings/search/filters/contexts). A different filter set per category (category pages only, default channel). This is how the Tennant page gets "fits model" (the Category filter over its model subcategories) and an in-stock filter without touching other categories.
• Custom category templates (templates/pages/custom/category/*.html), assigned per category. One brand.html serves every manufacturer page.
• Category metafields, read through GraphQL in the template's front matter. They hold a model's machine_type and era, and a brand's popular-models list.
• Page Builder widgets in regions the theme defines. Whether the Supermarket theme defines a region on category pages is still to verify.
• The URL redirect manager, with CSV import, for the merged duplicate models.
• The theme's search within category, already on every brand and model page.

05Stencil limitations

• No model-search box. The finder is custom, and it's already built on the homepage.
• No grouping of subcategories by an attribute. Subcategories render as one flat list in sort order; type sections need the custom template.
• No searchable filter for 250 values. The native Category filter lists values with "show more"; a model search that sets the filter is custom.
• Custom-field filters may need Enterprise. Older BigCommerce documentation limits them to that plan; confirm USA Clean's plan.
• Pagination stops at 10,000 products (measured), and BigCommerce advises staying under 10,000 per category. Filtering, or HawkSearch landing pages, is the way past it.
• Page Builder works only where the theme defines regions.
• No FAQ block or FAQ markup natively. The content goes in the description; the markup goes in the template.

06App marketplace shortlist

07Caster impact

All new CSS stays inside each prototype page (rule 7). The patterns below are marked SHARED CANDIDATE in the source and move into Caster when the client picks a direction, as the equipment set did.

• Shared by all three pages:
  • .model-tile and the machine directory (.md), in js/brand-page-ui.js
  • the model-search box .machine-combobox, in js/machine-combobox.js
  • .product-card__label, the OEM / aftermarket / other-brand label on part cards
• Second use of existing patterns:
  • Direction B reuses the search page's filter group, removable filter chip and filter drawer.
  • Direction C reuses the model-landing .doc-card, .service-block and .section-head.
  • A second use is the rule-7 trigger for promotion. .section-head carries its fix: the row stacks below 640px.
• Page-specific:
  • A: .brand-head, .finder, .popular
  • B: .b-head, .brand-search, .ptype, .b-toolbar
  • C: .brand-hero, .door, .faq, .c-row

No token changes and no cache-bust.

08Phased recommendation

A · Machine-first

Graduate to near-term when, over 30 days after the quick wins, the finder or a model tile is the first click on most brand-page visits, and the brand-page quick-back rate is still above the site-wide 25.6%.

Graduate to future when "Where's my model number?" is opened on 5% or more of visits, or the AI Parts Finder's half-day evaluation passes.

B · Parts-first

Graduate to near-term when all three hold:

• search within the brand or filter use reaches at least one in ten brand-page visits
• Hawk dashboard access is granted
• a part-type value exists on 80% or more of the brand's parts

Graduate to future when the fleet phase enters scope.

C · Brand hub

Graduate to near-term when organic search is the largest entry source for brand pages (the GA4 landing-page report).

Graduate to future when a policy for discontinued and superseded parts is agreed and signed-in visits are a material share of brand-page traffic.

The recommended path

1. Quick wins, all brands, in weeks:
   • the fix-first list
   • category sort order
   • the native Category and in-stock filters
   • search within the brand at the top
   • the live finder pre-set to the brand
   • document and service links in the description
2. Near-term, top 20 brands first:
   • one brand template on Direction A, with C's documentation and service strip
   • HawkSearch landing pages for filtering, when access lands
3. Future: the Direction B and C future items, as their triggers arrive.

09Implementation notes

• Tagging job. A machine_type and era per model category: about 245 rows for Tennant and about 780 for Nilfisk-Advance. It is spreadsheet work that finishes, unlike part-level tagging.
• Redirects. The four duplicate groups above. Their slugs are recorded in the prototype data (merged), ready for a CSV import.
• Finder fix. Link to the chosen category's own address; each option already carries an id.
• Analytics. Instrument from day one:
  • finder selection
  • type-chip clicks
  • model-tile clicks
  • search within the brand
  • filter applied
  • door clicks (C)
  • FAQ opens (C)
  • quick-back rate on brand pages
• Accessibility:
  • The model search follows the ARIA combobox pattern: arrow keys, Home, End, Enter, Escape, and the active option announced.
  • Chips are toggle buttons with aria-pressed.
  • The phone filter drawer traps focus, closes on Escape and returns focus to its button.
• Performance. Never render 782 tiles. Tagged groups show images; the models without a machine type yet are a plain list; images load lazily.
• Search engines and AI answers. The machine list is rendered by the server in the template, never only by script. The FAQ is written to be quotable, which is Stage 1 of the GEO/AEO plan.
• Prototype data.
  • Live values: model names, addresses and images; part names, SKUs, prices, images and flags, captured 2026-09-29.
  • Illustrative: machine types, eras, popular flags and part types.
  • Real fitment: each sampled part's "fits" list is real category membership among the four sampled models (T500E, T7, T3, S20).

10Catalyst / replatform candidacy

Directions A and C ship entirely in Stencil. They need native sort order and filters, a custom category template, and category metafields. Nothing in them is hard for Stencil.

Direction B's near-term also ships in Stencil, through HawkSearch landing pages.

B's future adds to the case without making it. Personalized ordering and instant filtering across 10,000+ parts is the same headless-toolkit case as search Phase 3.

Class: accumulating evidence. Not a reason to replatform on its own; the verdict is recorded in Platform strategy.

11Sources