How We Pick Gear: The SteelHead Addiction Curation Methodology

The affiliate disclosure (the important part)

SHA is an Amazon Associates affiliate. When you click a product link from a SHA page and complete a purchase on Amazon, SHA earns a small commission. This commission is what funds the site\'s operation -- server costs, USGS gauge polling infrastructure, editorial time, development.

Several things are not the case:

• No brand pays SHA for placement on Best-For pages.
• No brand pays SHA for inclusion in the Made-in-USA filter.
• No brand pays SHA for higher ranking in search results.
• No content has been altered or removed at a brand\'s request in exchange for compensation.

The commercial relationship is between SHA and Amazon (via the Associates program), not between SHA and the brands whose products appear on Amazon. The picks reflect what the editorial + algorithmic process selected, not what a brand sponsorship paid for. Every affiliate link on the site carries the FTC-required disclosure (rel="nofollow sponsored") and a footer disclosure appears on every page.

The three-layer curation process

Layer 1: Taxonomy classification

Every product in the SHA catalog (currently ~16,000 products) is classified by:

• gear_type -- the primary category (fly_rod, fly_reel, fly_line, spinning_reel, centerpin_reel, waders, wading_boots, flies, lure, etc.)
• technique -- the techniques the product supports (fly_fishing, centerpinning, spey, nymphing, indicator, etc.)
• target_species -- the species the product is designed for (steelhead, smallmouth_bass, walleye, etc.)
• price_tier -- budget / mid / premium
• brand -- normalized brand name

Classification is done by a combination of rule-based parsing (title + description keyword matching) and AI-assisted enrichment (for products where the title doesn\'t clearly indicate category). The classification pipeline is documented in the enrichment admin tools and uses Sonnet-level model accuracy where AI assistance is required.

Roughly 95% of catalog items get accurate taxonomy classification from rule-based parsing alone. The remaining 5% use AI assistance with confidence scoring. Items below confidence threshold get manual review.

Layer 2: Per-page editorial curation

Each Best-For page (e.g., Best Fly Rods for Steelhead) is built from a hand-written configuration that includes:

• Hand-written intro + buyer guide + FAQs. The editorial prose explaining why the picks are picks. Not auto-generated.
• Filter rules. Which taxonomy fields + keyword matches define the page\'s product universe.
• Exclude keywords. Patterns that explicitly filter out non-fishing items that share fishing-adjacent keywords (the famous "rod" matches for curtain rods and plumbing).
• Price gates. Some pages have minimum price thresholds to exclude obviously inappropriate items.
• Related pages. Cross-links to adjacent categories.

The combination of filter rules + editorial framing produces a result set that, in theory, only contains products an angler shopping that specific category would actually consider. In practice, the filters fail at the edges -- this is why SHA spends ongoing effort on filter refinement and why the Best-For pages occasionally show items that shouldn\'t be there.

Layer 3: Quality signal ranking

Within the filtered result set, products are ordered by a composite quality score:

• Amazon review rating (weight: 20 points per star, max 100)
• Amazon review count (weight: cap at 50 points)
• Editorial enrichment presence (weight: 10 points if present)
• Editorial featured flag (weight: 15 points if editor-flagged)

The product with the highest composite score appears first. Subsequent products in declining order. Featured-tier products (where the editor has explicitly flagged a pick as the recommended choice) get the top slot regardless of rating.

The Made-in-USA filter

The Made-in-USA filter is built on a curated brand list (config/made-in-usa-brands.php) that includes only brands with clear, durable US-manufacturing reputations for their flagship products. The conservative bias is intentional: false positives (claiming a brand is US-made when it isn\'t) undermine the filter\'s credibility more than false negatives (missing a brand that should be on the list).

Brands with mixed sourcing (e.g., Orvis -- Helios is US-made but Clearwater is imported) are not on the list. Brands with US-distribution but offshore manufacturing (e.g., Costa, KastKing) are not on the list. The full editorial reasoning is in the Made-in-USA Fishing Gear companion article.

If a brand on the SHA Made-in-USA filter is no longer manufacturing in the US (or never was), that\'s a bug. Contact the editor at the address in the footer with corrections.

The BuellEA Catalog v2 relationship

SHA does not maintain its own product database from scratch. The catalog of ~16,000 items is consumed from BuellEA Catalog v2 via authenticated API. The same catalog powers multiple partner sites that share the same Amazon Associates infrastructure.

The technical relationship:

• BuellEA Catalog v2 maintains the product master, taxonomy, enrichments, and variant relationships
• SHA pulls a cached subset relevant to fishing/outdoor gear (~16K items)
• Daily sync via php artisan content:sync keeps the SHA cache fresh
• The Amazon Associates tracking ID (steelhead-addiction-20) is appended at link-render time

This is documented technical infrastructure, not a hidden commercial relationship. The benefit is that SHA can focus on editorial curation and fishing-specific filtering rather than rebuilding catalog management from scratch.

What the algorithm cannot capture

Despite the documented methodology, several things the curation system cannot do:

1. "The editor would actually fish this." The composite quality score reflects market signals, not personal preference. Sometimes the editor would pick a different rod than the algorithm ranks first.
2. Recent product quality changes. A brand whose 2024 model was excellent and whose 2026 model is poor will not necessarily show up in the rankings until enough reviews accumulate.
3. Personal fit + ergonomics. Which 9-foot 7-weight feels right in your hand is an empirical question that no algorithmic system can answer.
4. Edge-case technical fit. Some fishing situations have niche gear requirements that the broad taxonomy doesn\'t capture (specific tippet ratings, exact rod-line balance specifications, etc.).
5. Inventory volatility. Amazon inventory changes constantly. A product that was the editor\'s pick last week may be discontinued this week.

The honest editorial position

SHA\'s gear-curation goal is to produce a Best-For page that, for a typical angler in the SHA region, returns 24 products where:

• At least 20 are legitimately appropriate for the page\'s stated category
• The top 3 picks are products the editor would consider buying
• The methodology is transparent enough that an angler with different preferences can recognize where the algorithm + their preferences diverge

This is a meaningful bar. It does not, by design, produce "the single best fly rod for steelhead" as a definitive answer. The picks reflect a process; the process has documented limitations.

How to use SHA gear pages

1. Read the Best-For intro + buyer guide to understand the editorial framing
2. Look at the top picks -- the editorial-flagged + algorithmically-top items
3. Skim the rest grid for adjacent options
4. Check the Amazon page for current price, availability, recent reviews
5. If the pick fits your preferences + budget, buy. If not, the algorithm gave you a starting point, not an ending point

Honest gaps

• Some Best-For pages have weaker product pools than others. Categories like steelhead spey rods have rich inventory; categories like centerpin reels have fewer options. The page quality reflects the underlying inventory.
• The quality-signal ranking favors products with many reviews. Brand-new products with strong editorial signals but few reviews can be under-ranked.
• The taxonomy classification, while ~95% accurate, occasionally produces edge cases where a kit or assortment shows up in a niche category. Ongoing filter refinement addresses these as they\'re identified.

Full source list

• Tier 1 -- BuellEA Catalog v2 API (/api/catalog/items) -- product master + taxonomy.
• Tier 1 -- Amazon Product Advertising API + Amazon Associates Program -- pricing + review signals.
• Tier 1 -- SHA editorial review + filter refinement (ongoing).
• Tier 2 -- Direct angler reports + brand verification for the Made-in-USA filter.
• Tier 2 -- IFTD trade show reporting on US-manufacturing claims.

Continue learning

• Made-in-USA Fishing Gear -- the editorial reasoning behind the brand list.
• Source Discipline + Tier Hierarchy -- SHA's source-validation methodology.
• How We Score Rivers -- companion methodology piece on the river-fishability scoring.
• Best-For Index -- the live Best-For landing pages.
• Gear-O-Rama -- the full filtered catalog browser.
• Before You Go -- the angler-facing gear primer.