# aicodingpricing leaderboard — Content Fit Matrix

Date: 2026-05-28
Scope: existing-site inner-page optimization for `/llm-leaderboard`, `/best-llm-for-coding`, `/coding-agent-cost-calculator`.
Design source inputs:
- `/root/.hermes/reports/aicodingpricing-leaderboard-20260528/input-brief.md`
- `/root/.hermes/reports/aicodingpricing-leaderboard-20260528/seo-copy-freeze.md`
- `/root/.hermes/reports/aicodingpricing-leaderboard-20260528/data-contract.md`
- `/root/.hermes/reports/aicodingpricing-leaderboard-20260528/model-leaderboard-seed.json`

Design stance: Industrial editorial dashboard. Dense, source-led, decision-first. No generic SaaS hero. No fake universal score.

## Global rules

| Rule | Design implication | Block if violated |
|---|---|---|
| No fabricated benchmark/price/context/speed/task-cost data | Unknowns stay as visible badges/cell text | yes |
| Token price and task cost stay separate | Table shows unit price; calculator bridge estimates task cost | yes |
| Recommendations are workflow shortlists, not crowns | Cards say “candidate / best for workflow”; no “best overall” | yes |
| Caveats remain visible | Caveat is a row/card field, never hover-only | yes |
| Source/freshness remains visible | Every source-backed field needs source, date, confidence/status | yes |
| Existing site, not new site | Nav and CTA chain preserve aicodingpricing calculator intent | yes |

## /llm-leaderboard

| SEO/copy block | Source | Priority | Display location | Layout container | Frontend notes | Visual risk |
|---|---|---:|---|---|---|---|
| SEO Title / meta / OG | SEO Copy Freeze lines 33–54 | head-only | page metadata | n/a | Use exact copy; do not add universal ranking language | low |
| H1: AI Coding Model Leaderboard for Coding Agents and Developer Workflows | SEO Copy Freeze lines 55–59 | P0 | hero | asymmetric split hero | Stitch visual uses shorter display headline; implement H1 as actual semantic H1, display headline can be shorter if SEO H1 remains in DOM | medium |
| Above-fold answer block | lines 70–73, 129–133 | P0 | directly below hero | full-width statement strip | Must appear before first table; concise enough for mobile | low |
| Filterable leaderboard intro | lines 134–147 | P0 | filter rail intro/chips | horizontal chips / sticky toolbar | Filters are lenses, not rank types | medium |
| Leaderboard table helper copy | lines 149–158 | P0 | above or beside table | helper note + unknown legend | Unknown status copy should be in DOM, not image | medium |
| Leaderboard table rows | data contract + seed JSON | P0 | main content | desktop table, mobile cards | Use source-specific benchmark refs; no composite score | high |
| Recommender cards | brief lines 82–87 + contract lines 324–350 | P0 | above table | 4 workflow cards | Cards must show evidence used, confidence label, caveat, calculator CTA | high |
| Methodology caveat | lines 160–163 | P0 | near table | 2x2 methodology panel | Include “no fake universal score” | low |
| Source/freshness block | lines 165–168 | P0 | near table / row metadata | source chips + footer note | Per-row source date beats global freshness | high |
| Token price vs task cost | lines 170–173 | P0 | conversion bridge | split calculator preview | Never compute task cost without user assumptions | high |
| CTA bridge | lines 175–184 | P0 | after table + row CTA | row CTA + bottom module | Query params: `?model=<model>&workflow=<workflow>` | low |
| FAQ | lines 186–207 | P1 visible lower page | accordion | collapsible but crawlable | Only visible FAQs get FAQPage schema | medium |
| Schema | lines 209–222 | implementation | n/a | n/a | Dataset only if versioned/downloadable; ItemList only with methodology caveat | high |

## /best-llm-for-coding

| SEO/copy block | Source | Priority | Display location | Layout container | Frontend notes | Visual risk |
|---|---|---:|---|---|---|---|
| H1/subhead | SEO Copy Freeze lines 250–255 | P0 | hero | editorial guide hero | Avoid “best overall”; title says by workflow/evidence/cost | low |
| Short answer | lines 320–323 | P0 | above fold | answer strip | Should link user to scenario cards | low |
| Scenario cards | lines 325–363 | P0 | main content | 2+3 card grid desktop, stack mobile | Each card CTA filters leaderboard, not direct unsupported claim | medium |
| Evidence table intro | lines 365–368 | P0 | before evidence module | compact table intro | Can reuse leaderboard mini-table component | medium |
| Methodology copy | lines 370–373 | P0 | below cards | caveat block | State recommendation labels are editorial derived from visible fields | low |
| Calculator CTA bridge | lines 375–384 | P0 | conversion bridge | model/workflow prefill CTA | CTA goes to calculator with model/workflow when selected | low |
| FAQ/schema | lines 386–421 | P1 | lower page | accordion | No AggregateRating, no hidden FAQ schema | high |

## /coding-agent-cost-calculator

| SEO/copy block | Source | Priority | Display location | Layout container | Frontend notes | Visual risk |
|---|---|---:|---|---|---|---|
| H1/subhead | SEO Copy Freeze lines 449–453 | P0 | hero | compact calculator hero | Existing calculator intent must stay dominant | low |
| Calculator intro | lines 513–517 | P0 | above form | helper panel | Prefill only source-backed prices; unknown fields blank | high |
| Input helper copy | lines 518–532 | P0 | form helper | inline helper + “assumptions visible” block | Do not prefill task token defaults unless product approves | high |
| Formula copy | lines 534–541 | P0 | result explanation | formula panel | Show components separately | medium |
| Required caveat | lines 543–546 | P0 | near result | red/amber notice | “Estimate, not billing quote” must be visible | high |
| Sensitivity copy | lines 548–551 | P1 | result explanation | sensitivity strip | Highlight retry rate risk | medium |
| Sample scenarios | lines 553–566 | P1 | below form | preset cards | Presets can set workflow, not fake numeric defaults | medium |
| Leaderboard return CTA | lines 568–577 | P0 | after result | comparison CTA | Back to `/llm-leaderboard` | low |
| FAQ/schema | lines 579–615 | P1 | lower page | accordion | SoftwareApplication only if calculator is interactive | high |

## Downstream content decisions

- Homepage should not carry all leaderboard SEO copy. It should only add an entry point: “AI coding model leaderboard” → `/llm-leaderboard`.
- P1 pages `/coding-model-benchmark` and `/llm-api-pricing-comparison` should not receive prominent CTAs until unique content exists.
- Unknown-value legend must be reusable across all three P0 pages.
- Mobile cards must retain caveat/source/CTA; hiding them for space is a P0 design failure.
