# Ecommerce platform teardown: coding notes, codebook v2.0 (collected September 26, 2026) The published codes are in coded_v2/. They come from two independent AI readers (Claude agents) who coded every answer blind from gold2/READER_PROTOCOL_v2.md (shipped as reader-protocol.md), with a third AI reader deciding every code the two disagreed on. The readers are AI models, not people. The older rules-based coding in coded/ is retired and not used anywhere on the page. ## Method and agreement - Six codes per brand per answer: picked (the answer's own verdict, overall or for a case), recommended (assigned to a stated case or fit, not as the verdict), listed, passing, anchor (the brand being replaced in an alternatives question), not_a_mention. The page's "recommended" column includes picked. - 411 brand codes per reader. Reader vs reader: 97.8% at the six-code level, 100.0% at the recommended level (gold2/CODED_V2_SUMMARY.md). - 9 codes decided by the third reader, all picked vs recommended: 8 went to recommended, 1 to picked (details below). - The rules-based first pass (v1.6 rules) agreed with the final codes on 96.8% of brand codes at the recommended level. It was used only to build the candidate lists and as a check; the reader codes are what is published. - Final mix: 376 recommended (285 of them picked), 6 listed, 9 passing, 20 anchor. 93 answers pick at least one brand. ## Third-reader decisions (gold2/out/J02.jsonl) - q066 wix alternatives for ecommerce: Shopify is picked by both readers ("I'd look here first"). WooCommerce, BigCommerce and Squarespace sit in the same practical shortlist but each carries a fit ("particularly interesting if you want ownership", "worth considering for large catalog, B2B", design-first stores). Decided recommended: described, not chosen. - q087 which ecommerce platform is best value for money: WooCommerce decided picked. The answer's own practical take says WooCommerce is better value if the buyer is comfortable with WordPress, a conditional verdict, alongside Shopify as the overall value pick. - q091 works best if I also sell on Amazon and eBay: Ecwid decided recommended. It is #4 of a ranking with a best-for label (a smaller operation); Shopify, BigCommerce and WooCommerce carry the "I'd choose" lines. - q093 artist selling prints: Squarespace decided recommended. It fits the portfolio-plus-store case in the catalog but is not in the answer's practical recommendation (Big Cartel, Shopify, Etsy, Printful are). - q097 also sell in person: Square Online, Wix and WooCommerce decided recommended. Shopify is "the first one I'd investigate"; the other three get "particularly compelling if mostly face-to-face", "worth considering if the website is the priority" and "makes more sense if committed to WordPress". This is why Square Online is picked in 3 of the 4 local and in-person questions, not 4. ## Judgment calls that matter for this category - No-pick full answers. The readers agree that four long answers give every brand a fit and never choose: q001 and q002 (the two broadest questions, "best fit" / "best suited for" tables), q030 high-volume sellers ("strongest fit for DTC", "compelling for complex catalogs", "consider for extensive customization") and q036 shopify vs magento ("Shopify tends to make sense if" / "Magento tends to make sense if"). All brands recommended, none picked. - One-sentence replies. q044 commercetools vs salesforce commerce cloud only offers to compare: both brands passing. q054 shopify or etsy for handmade goods names neither brand: no codes. q094 which ecommerce platform should I switch to from wix asks what the buyer sells: Wix passing (the buyer's current platform, inside a clarifying question). This leaves the advice questions at 19 of 20 with a pick. - Decision tables count as picks. q011 top rated ecommerce platform: a "Quick choice" table that assigns a platform to each need is coded picked for all five brands (decision rules phrased as instructions). q010 most popular: the direct answer (Shopify and WooCommerce) is picked; BigCommerce and Wix, offered only in a follow-up comparison, are listed. - Verdict phrasing the readers read as picked: "strongest all-around choice" (q017, Shopify), "the first platform I'd price out" (q068, BigCommerce), a "My picks" or "My shortlist" heading over the list items (q070 WooCommerce, Square Online, Big Cartel; q075 BigCommerce, Magento, OroCommerce), an arrow after the brand under a "What I'd choose" heading (q099: Medusa, Saleor, WooCommerce and Shopify all picked). - Specialists as add-ons. Printful, Printify and Etsy are coded on what the answer tells the buyer to do: in q093 "use Etsy alongside your own store" and "fulfillment via Printful" are picks; in q016 Printful and Printify are picked next to Shopify. ## Reader-added brand mentions - Both readers flagged one mention the alias list missed: q020 best ecommerce platform for food and beverage brands names Magento only by the link text [business.adobe.com] in a table row for enterprise food manufacturers. Coded recommended. The summary table counts this as 2 (one flag per reader). - Link text that is a brand's own domain in subject position names that brand (protocol rule), so table rows such as "[shopify.com]" or "[bigcommerce.com]" count as naming the brand. ## Collection All 100 ChatGPT tasks and 100 Google tasks were posted and fetched through the DataForSEO standard queue between about 8:15 and 8:30 PM US Mountain time on September 26, 2026 (the raw files carry UTC timestamps, early September 27). Zero failed tasks, no retries. Three answers are a single sentence (q054, q094, q044, see above); the median answer is about 3,000 characters. q049 magento vs bigcommerce vs salesforce commerce cloud is a full answer with no citations. ## Dictionary notes (still apply) - Changed after collection, before the final coding run: extra vendor domains only (shopify.dev for Shopify; bigcommerce.co.uk and bigcommerce.com.au for BigCommerce). No alias changed. Details and effects in DICTIONARY_CHANGES.md; the drafted dictionary is brands_original.csv. The domain changes affect citation classes, not brand codes. - Folded before collection (see INPUT_NOTES.md): Shopify Plus into Shopify; Adobe Commerce and Magento Open Source into Magento; Square (any capitalized "Square") into Square Online. Of the 25 answers with a Magento row, 14 use only "Adobe Commerce", 8 use both names, 2 use only "Magento" and 1 names it only by its Adobe address. In the 18 answers to questions that do not name Magento, 14 use only "Adobe Commerce". - adobe.com counts as Magento's own site, and squareup.com as Square Online's. - Amazon and eBay are left out as marketplaces; Amazon is named in 6 answers. Subscription apps (Skio, Appstle, Recharge) are not in the dictionary, so the two subscription answers show Shopify as the only picked brand. - Case-sensitive aliases (capitalized only): Square, Medusa, Swell, Faire, Lightspeed. Every "Square" hit was read; all refer to Square the company. - coded_v2/out_of_dictionary.csv is empty for this category because the scraper's brand entities carry the name under "title"; the out-of-dictionary scan was done from the raw answer text instead (only Amazon and Next.js, 6 answers each, reach 5 or more).