AI Product Comparison Generator
The highest-converting affiliate pages and SaaS comparison articles are all the same shape: a side-by-side table where each row is one decision criterion and each cell is a concrete value a buyer can actually compare. Generic AI writes a prose paragraph. scrb's comparison endpoint returns the table - as JSON, 6-10 rows per comparison, concrete cell values (numbers, units, yes/no), plus an honest one-paragraph summary naming the actual trade-offs. Works for 2-5 products at once.
What scrb knows about product comparisons that a chat model doesn't
A chat model will write you a paragraph. A review author wants a row-by-row structured table they can render as HTML or paste into Notion or embed in a WordPress article via a comparison block.
- Structured JSON:
{products: [...], rows: [{feature, cells: [{product, value}]}], summary}. Each row is one decision criterion. Each cell is a concrete scalar value, not a sentence. - 6-10 rows per comparison: chosen to cover the features a buyer actually cares about for that category - not a max-out of 25 fields padded with filler. For headphones: driver size, battery, ANC, codec, weight, warranty. For SaaS: starting price, seat limit, integrations, trial length, SSO, data export.
- Concrete cells: "45mm dynamic", "40 hours", "$9.99/mo", "5 seats", "yes", "no" - never vague adjectives like "good" or "premium".
- Honest summary: one paragraph that names the actual trade-offs ("product A wins on battery and price, product B wins on ANC and build quality") - not a neutral "each has its merits" weasel.
- 2-5 products per call: more than 5 and the rows get too wide to read on mobile; less than 2 isn't a comparison. Billing debits N generations (one per product) atomically before calling the LLM, so 5/5 quota + 3-product request returns 429 cleanly without a burnt API call.
- Category-aware: the generator reads the category and features you provide and picks comparison criteria relevant to that category. Automotive products get different rows than wireless earbuds.
Before / after: real examples
Generic AI output:
The Sony WF-1000XM5 is a premium choice for audiophiles seeking industry-leading noise cancellation. The AirPods Pro 2 integrate seamlessly with Apple devices and feature Spatial Audio. The Sennheiser Momentum 4 TWS offers excellent sound quality. Each has its own strengths - choose based on your needs.scrb output:
rows:Why this wins: every cell is a specific value you can check. The summary names which product to pick in which scenario, not a "depends on your needs" shrug.
• Codec: Sony = LDAC | Apple = AAC only | Sennheiser = aptX Adaptive
• Battery (with case): Sony = 24hr | Apple = 30hr | Sennheiser = 28hr
• ANC: Sony = yes | Apple = yes | Sennheiser = yes
• Water resistance: Sony = IPX4 | Apple = IPX4 | Sennheiser = IPX4
• Best ecosystem: Sony = Android | Apple = Apple | Sennheiser = cross-platform
• Driver size: Sony = 8.4mm | Apple = 11mm | Sennheiser = 7mm TrueResponse
• Spatial audio: Sony = 360 Reality Audio | Apple = Spatial Audio | Sennheiser = no
• Price (MSRP): Sony = $299 | Apple = $249 | Sennheiser = $299
summary: Pick the AirPods Pro 2 if your phone is an iPhone - the H2 chip lock-in is real and spatial audio switching only works inside Apple's ecosystem. Pick the WF-1000XM5 if you're on Android and want LDAC for hi-res streaming. Pick the Sennheiser Momentum 4 TWS if you switch between devices across OSes and want a warmer sound signature over best-in-class ANC.
scrb output:
rows:Why this wins: the rows are specific to email marketing buyers' actual decision criteria. The summary names the price crossover point (~8k subs) - a real fact someone would cite in a review article.
• Free tier ceiling: ConvertKit = 1,000 subs | Mailchimp = 500 subs
• Paid plan at 1k subs: ConvertKit = $29/mo | Mailchimp = $26/mo
• Paid plan at 10k subs: ConvertKit = $79/mo | Mailchimp = $100/mo
• Automation builder: ConvertKit = visual (creator-focused) | Mailchimp = visual (business-focused)
• A/B testing: ConvertKit = paid plans only | Mailchimp = all plans
• Landing page builder: ConvertKit = yes (mobile-first) | Mailchimp = yes (templated)
• Segmentation depth: ConvertKit = tags + conditions | Mailchimp = segments + audiences
• SMS: ConvertKit = no | Mailchimp = yes (add-on)
• Free trial (paid features): ConvertKit = 14 days | Mailchimp = no
summary: ConvertKit wins for creators running newsletters and courses - the tagging model fits the "fan → customer" lifecycle and cost scales more gradually past 10k subs. Mailchimp wins for small businesses that need SMS and a wider integration catalog. At under 1,000 subs, ConvertKit's free tier is twice the size. At scale, Mailchimp gets more expensive than ConvertKit starting around 8k subs.
FAQ
How many products can I compare at once?
2 to 5 products per call. Less than 2 isn't a comparison; more than 5 makes the table wider than most mobile screens and dilutes the per-row argument. If you need to compare 6+ products, run two comparisons (the top-3 and the rest) and surface the winner of each.
How does billing work?
Each comparison counts as N generations, where N is the number of products compared. A 3-product comparison debits 3 generations from your plan quota (reflecting the generator's multi-product reasoning). Quota is checked BEFORE the LLM call, so a user at 98/100 requesting a 3-product compare gets a clean 429 instead of a burnt API call that couldn't be billed.
Can I embed the output as an HTML table?
Yes. The web app includes an "Export as HTML table" button that emits a semantic <table> with <thead>, <tbody>, one column per product, one row per feature. Paste directly into WordPress's HTML block, Notion, Webflow Rich Text, etc.
Can I compare products I don't have full specs for?
Yes, but the output quality tracks the input quality. If you give scrb only the product name for each of 3 products, the generator will compare based on publicly-known attributes. If you give the name + 3-5 specific features per product, the rows will be more decision-specific. For affiliate content, the 5-feature version is the minimum to get a table that outperforms a generic writeup.
Does the summary lean toward one product?
The prompt explicitly asks for an honest trade-off summary, not a neutral "each has merits" paragraph. Expect lines like "pick A if your phone is Android; pick B if you're on iPhone." If you want a single winner recommendation instead, post-process by asking the summary field "which wins overall?" yourself - scrb won't answer that implicitly.
How to plug scrb into your workflow
- Web app: pick "product comparison" mode, enter 2-5 products with name + features, get the structured table. Export as HTML, Markdown, or CSV.
- REST API:
POST /api/v1/generate/comparewith a products array. See API docs. - WordPress / Ghost embed: paste the Export as HTML output into a post. The CSS is inherited from your theme.
- Affiliate content pipelines: run the comparison via API, publish to your CMS, and let the structured output inform the meta description + SEO title via the meta description generator.
Pricing
Product comparisons count N generations (one per product compared).
- Free: 5 generations / mo = one 5-product or two 2-product comparisons.
- Starter ($9.99/mo): 100 generations = ~33 three-product comparisons.
- Pro ($24.99/mo): 500 generations = ~166 three-product comparisons + bulk CSV + REST API.
- Business ($49.99/mo): 3,000 generations = ~1,000 three-product comparisons.