Citewake

Open data every edition

Citewake Surface Map

Explore where AI systems retrieve and cite information when recommending products and services.

Accounting software, Edition 2: cited, by surface

ChatGPT60 answersClaude60 answersGemini60 answersPerplexity60 answersVendor and product sitesSpecialist reviews and comparisonsCommunities and discussionIndependent editorialOfficial and governmentAcademic and researchApp marketplacesDeveloper and technicalNot yet classified
Accounting software, Edition 2: cited, by surface. Cited: answers that cited it, out of each AI's answers.
SurfaceChatGPTClaudeGeminiPerplexity
Vendor and product sites53 / 6088.3%55 / 6091.7%45 / 6075.0%56 / 6093.3%
Specialist reviews and comparisons4 / 606.7%35 / 6058.3%22 / 6036.7%9 / 6015.0%
Communities and discussion0 / 600.0%1 / 601.7%26 / 6043.3%0 / 600.0%
Independent editorial1 / 601.7%2 / 603.3%10 / 6016.7%6 / 6010.0%
Official and government12 / 6020.0%0 / 600.0%1 / 601.7%6 / 6010.0%
Academic and research0 / 600.0%1 / 601.7%1 / 601.7%0 / 600.0%
App marketplaces0 / 600.0%0 / 600.0%0 / 600.0%1 / 601.7%
Developer and technical0 / 600.0%0 / 600.0%0 / 600.0%1 / 601.7%
Not yet classified6 / 6010.0%47 / 6078.3%39 / 6065.0%7 / 6011.7%

Hover or tap a ribbon for its exact numbers. Select a surface to see its domains, and a domain to see the questions and answers it appeared in.

Citations and recommendations

Retrieved, cited and recommended are separate measurements, never added together. This is how each AI's citations divide between the brands it recommended and everything else.

  • Of ChatGPT's 181 citations, 58 (32.0%) point to a brand it recommended in the same answer, 9 to a brand it only named, and 114 to third parties.
  • Of Claude's 390 citations, 48 (12.3%) point to a brand it recommended in the same answer, 3 to a brand it only named, and 339 to third parties.
  • Of Gemini's 621 citations, 58 (9.3%) point to a brand it recommended in the same answer, 3 to a brand it only named, and 560 to third parties.
  • Of Perplexity's 276 citations, 87 (31.5%) point to a brand it recommended in the same answer, 9 to a brand it only named, and 180 to third parties.

Across editions

Comparisons appear when an earlier edition researched the same category, with the same panels and benchmark.

How to read it

Ribbons
Each ribbon joins an AI to a surface or a domain. Its thickness is a share of that AI's answers, so the four AIs are comparable. Exact numbers and denominators are on every ribbon and in the table.
Retrieved
The answer's own web search returned a page from the source, out of the answers that searched.
Cited
The answer cited a page from the source, out of all the AI's answers. Each source counts once per answer.
Recommended
The answer cited the source and recommended the brand whose own site it is.
Gemini
Not reported by Gemini: Gemini reports only the sources it used, so what it retrieved but didn't cite can't be observed. It is never shown as zero.
Not yet classified
Sources no published rule or review covers yet. They are counted and shown in grey, never guessed.
Panels
Categories asked per market show the United States and Europe separately. They are never pooled.

Built from each edition's published answers. The method. Data: index (JSON), and each category's JSON and CSV, linked under the map. Licence: CC BY 4.0.

Files for this view: Accounting software (JSON), every source (CSV).