Every public consumer + B2B review for any brand, fused across Trustpilot · G2 · Capterra · PissedConsumer · Amazon · eBay. Aspect-graded sentiment (price / shipping / quality / support / UX), verified-purchase filter, photo-review flag, brand-response tracking. CSV+Parquet. Pricing on request — send the brand + markets, quote in 24h.
Six platforms covering consumer brands (Trustpilot, Amazon, eBay), SaaS / enterprise software (G2, Capterra), and complaints (PissedConsumer). Cross-platform dedup, brand-name normalization, aspect-level sentiment via Claude Sonnet 4.6 (not generic star-rating). Brand-response tracking shows what % of negative reviews got a public reply and how fast.
Every review LLM-graded across 5 aspects (price, shipping, quality, support, UX) with -1.0 to +1.0 sentiment per aspect. Brand & product normalized across platforms via fuzzy + embedding match. Brand-response tracking enabled where the platform exposes it.
Real rows. Reviewer names anonymized for this sample. Full 5,000-review free sample for any one brand on request.
# Trustpilot · Bose · DE · negative {"review_id":"acx_tp_4729108","platform":"trustpilot","brand_canonical":"Bose", "product_canonical":"QuietComfort Ultra Headphones","category":"headphones", "country":"DE","language":"de","rating":2,"verified_purchase":true, "sentiment_overall":-0.32,"aspect_quality":0.81,"aspect_price":-0.42, "aspect_shipping":-0.78,"aspect_support":-0.91,"brand_responded":true,"response_hrs":38} # G2 · Salesforce · enterprise · positive but with caveats {"review_id":"acx_g2_8174021","platform":"g2","brand_canonical":"Salesforce", "product_canonical":"Sales Cloud","category":"CRM","rating":8,"rating_normalized":0.80, "sentiment_overall":0.42,"aspect_quality":0.71,"aspect_price":-0.63, "aspect_ux":-0.18,"aspect_support":0.31, "topics":["customization","learning-curve","pricing"]} # Amazon DE · Anker · 5-star verified-purchase with photo {"review_id":"acx_amz_de_192847","platform":"amazon","country":"DE", "brand_canonical":"Anker","product_canonical":"Soundcore Liberty 4 Pro","category":"earbuds", "rating":5,"verified_purchase":true,"has_photo":true,"helpful_count":147, "sentiment_overall":0.88,"aspect_quality":0.92,"aspect_price":0.81} # PissedConsumer · Comcast · escalated complaint {"review_id":"acx_pc_3849021","platform":"pissedconsumer","brand_canonical":"Comcast", "category":"ISP","rating":1,"country":"US", "sentiment_overall":-0.94,"aspect_support":-0.98,"aspect_price":-0.71, "topics":["billing","hidden-fees","cancellation"],"brand_responded":false} # eBay UK · Shimano · seller-feedback style {"review_id":"acx_ebay_uk_7102849","platform":"ebay","country":"GB", "brand_canonical":"Shimano","product_canonical":"105 R7000 Crankset 172.5mm", "category":"cycling","rating":5,"verified_purchase":true, "sentiment_overall":0.74,"aspect_quality":0.88,"aspect_shipping":0.62} # Capterra · Hubspot · SMB perspective {"review_id":"acx_cap_5821049","platform":"capterra","brand_canonical":"HubSpot", "product_canonical":"Marketing Hub Pro","category":"Marketing automation", "rating":4,"sentiment_overall":0.51,"aspect_quality":0.62,"aspect_price":-0.38, "topics":["contact-limits","workflows","reporting"]}
Brand-snapshot for one-off competitive analysis or PR audit. Continuous monitoring tier for brand teams + customer-experience SaaS. Daily refresh on enterprise tier. Pricing is per use-case — brand count, markets, refresh cadence, license scope. Send the spec, we send a quote in 24h.
All historical reviews for one brand across Trustpilot, G2, Capterra, PissedConsumer, Amazon (.de/.com), eBay (.de/.com). For PR audit, competitive intel, due diligence.
Track up to 10 brands (your brand + 9 competitors), weekly refresh, delta CSV per delivery. For brand managers, customer-experience teams, agency competitive monitoring.
Unlimited brands, daily refresh, full firehose for agencies + CX SaaS building on top. License covers internal use + derivative signals.
Three things to know before you wire this to a brand-dashboard.
Aspect-level sentiment is computed by Claude Sonnet 4.6 with a 16-shot prompt + a smaller distilled stability-check model. Quality benchmarks ship in the README. Star ratings are platform-native. If you want your own scoring, raw text is in every row — swap in your model.
Same brand may appear as "Bose Corp", "BOSE", "BOSE - Better Sound Through Research". Canonical-brand mapping uses fuzzy + embedding match against a curated 14K-brand registry. ~97% accuracy on top-1K brands, drops to ~88% on long tail. Both raw and canonical fields shipped; you can override.
The likely_fake flag uses 9 heuristics (review velocity, reviewer history, language patterns, photo-EXIF anomalies, verified-purchase status). It's a signal, not a verdict — false-positive rate ~6%. Don't act on it without human review. Useful for filtering noise from analytics, not for legal action.
Email us the brand. We send a 5,000-review sample within 24h covering all 6 platforms so you can validate aspect-grading quality first.
Request Free Sample →