BroadcastwellStart free
Menu

Account exports

Export reference: what each file contains, field by field.

An export from a Broadcastwell account carries up to four sections: results, receipts, sources and fixes. Each file lists its fields and types, states its rows against a row limit of 2,000 per section (1,000 for fixes), names its dates and carries a SHA-256 checksum line. Only delivered results are exported.

By the Broadcastwell team. Updated 9 November 2026. The limits below are the ones the account applies today.

Open Exports in your accountDownload the fictional sample files

Choose a format

Every format carries the same rows and the same checksum line. Open Exports from the account navigation, or use the export links under the Questions, Receipts, Sources and Fixes views, which export that view's section.

7 formats
FormatSectionsWhat it holds
Excel workbook (.xlsx)Every section, or one sectionOne sheet for each section, plus an About sheet with the fields, dates and checksum.
CSV (.csv)One section in each fileOne section in each file. It opens in Google Sheets, Excel and Numbers.
Word document (.docx)Every section, or one sectionOpens in Word, Google Docs, Pages and LibreOffice.
Markdown (.md)Every section, or one sectionPastes into Notion, Confluence, GitHub and Asana.
HTML page (.html)Every section, or one sectionOne page that opens in any browser without a connection.
PDF report (.pdf)Every sectionResults, sources and fixes in full, with the receipt count and dates. Receipts themselves travel in the data formats.
JSON (.json)Every section, or one sectionThe canonical data the checksum is computed from.

Excel cells hold at most 32,767 characters, so an answer longer than 32,000 characters is cut in the workbook and marked where it was cut. The CSV, JSON, Markdown, HTML and Word files keep the full text. Markdown writes a line break inside a table cell as <br>.

Read the row limit before you download

The account shows each section's rows and its limit beside every download link. A section with more rows than its limit carries the first rows in date order and says, in the file, how many were left out; choose a shorter date range for the rest.

Row limits per file
SectionRow limitOne row isDate rule
Results2,000One row for each question in each delivered measurement.Figures are computed from the scheduled answers observed inside the chosen dates.
Receipts2,000One row for each observed answer, scored or excluded.An answer is included when its observed_at time falls inside the chosen dates.
Sources2,000One row for each page cited in a scored scheduled answer.A page is included when a scored scheduled answer inside the chosen dates cited it.
Fixes1,000One row for each change in the Source to Change Ledger.A change is included when the day it last moved falls inside the chosen dates.

Dates are UTC days. With no dates chosen, a file carries every retained answer from delivered measurements. A date range applies to every section of the file: results are recomputed from the scheduled answers inside it, so a rate always matches its receipts.

Every field, in file order

The field names below are the column headings of the CSV and Excel files and the keys of the JSON file. Markdown, HTML and Word list the same fields before the rows.

text
Text, as stored.
integer
A whole number.
number
A decimal number with one decimal place.
boolean
yes or no. Spreadsheets receive TRUE or FALSE.
date
A calendar day, YYYY-MM-DD, in UTC.
date-time
An ISO 8601 time in UTC, for example 2026-09-02T01:00:00.000Z.
url
A complete web address.
list
Several values in one cell, separated by a semicolon and a space.

Results: 19 fields

Results, in file order
FieldTypeMeaning
run_idtextThe measurement the figures come from.
run_kindtextThe kind of measurement.
method_versiontextThe method version the measurement used.
observed_fromdate-timeThe first scheduled answer counted in this row.
observed_todate-timeThe last scheduled answer counted in this row.
question_idintegerThe question number inside the measurement.
question_texttextThe buyer question as asked.
question_typetextThe question group.
scheduled_answersintegerScheduled answers observed for the question.
scored_answersintegerScheduled answers that could be scored. This is the base of the rate.
excluded_answersintegerScheduled answers left out of the base, each with its reason in the receipts.
named_answersintegerScored answers that named your company.
named_rate_pctnumbernamed_answers divided by scored_answers, as a percentage.
interval_low_pctnumberLower bound of the 95 percent Wilson interval.
interval_high_pctnumberUpper bound of the 95 percent Wilson interval.
own_domain_cited_answersintegerScored answers that cited a page on your own domain.
engineslistThe engines whose scheduled answers are counted.
rivals_namedlistCompanies named in these answers, each with its count of scored answers.
adaptive_answersintegerExtra answers asked again where scheduled verdicts disagreed. They are never added to the rate.

Receipts: 15 fields

Receipts, in file order
FieldTypeMeaning
receipt_idtextThe stored identifier of the answer.
run_idtextThe measurement the answer belongs to.
pass_idtextThe dated pass inside the measurement.
pass_kindtextscheduled, adaptive or monitor.
question_idintegerThe question number inside the measurement.
question_texttextThe buyer question as asked.
enginetextChatGPT, Claude, Perplexity, Google AI Overviews or Google AI Mode.
observed_atdate-timeWhen the answer was observed.
namedbooleanWhether the answer named your company.
own_domain_citedbooleanWhether the answer cited a page on your own domain.
competitors_foundlistCompanies the answer named.
cited_urlslistEvery address the answer cited.
excludedbooleanWhether the answer was left out of every base.
exclusion_reasontextThe plain reason an excluded answer was left out.
answer_texttextThe engine's words, exactly as observed.

Sources: 14 fields

Sources, in file order
FieldTypeMeaning
urlurlThe cited page.
domaintextThe host of the cited page.
own_domainbooleanWhether the page is on your own domain.
times_citedintegerScored scheduled answers that cited the page.
engineslistThe engines that cited the page.
question_idslistThe questions whose answers cited the page.
answers_naming_youintegerOf those answers, the ones that also named your company.
first_cited_atdate-timeThe first answer that cited the page.
last_cited_atdate-timeThe last answer that cited the page.
queue_statustextnot_started, pitched, placed or declined, from your source queue.
you_appeartextyes, no or not_checked, from your source queue.
ownertextThe owner recorded in your source queue.
contact_routetextThe contact route recorded in your source queue.
notestextYour notes from the source queue.

Fixes: 18 fields

Fixes, in file order
FieldTypeMeaning
entry_idtextThe ledger entry.
changetextThe change to test.
statetextDrafted, Approved, Sent, Acknowledged, Published, Measured again or Declined.
ownertextThe person or role responsible, or Unassigned.
question_idslistThe questions the change addresses.
rivaltextThe company named instead on those questions.
source_urlurlThe cited page behind the change.
receipt_urlurlThe receipt behind the change.
run_idtextThe measurement the change came from.
drafted_ondateThe day the change was drafted.
approved_ondateThe day the change was approved.
sent_ondateThe day the change was sent.
acknowledged_ondateThe day the change was acknowledged.
published_ondateThe day the change was published.
remeasured_ondateThe day a later scheduled run asked the questions again.
declined_ondateThe day the change was declined.
updated_ondateThe day the change last moved.
work_rankintegerThe position in the work queue, when the change has been assessed.

What is never exported

Adaptive answers, the ones asked again where scheduled verdicts disagreed, stay in the receipts with pass_kind set to adaptive. They are never added to a results rate.

Check that a file is unchanged

The checksum line is the SHA-256 of the file's data written as canonical JSON: object keys in sorted order, no spaces. The same data gives the same checksum in every format, so a CSV and a workbook downloaded together can be matched. The JSON file carries that data as its data member, with the rule written beside it.

A matching checksum shows that the rows are the ones exported. It does not show that an answer is correct; the receipt holds the engine's words for that.

Import a section into Google Sheets

  1. In Exports, create an import link for one section. It lasts 1 to 30 days, 7 by default, and is shown once.
  2. Paste the formula the account shows, =IMPORTDATA("..."), into cell A1 of a sheet.
  3. The first 6 rows are the file's label, dates, row count, checksum line and a link to this page. The field names follow on row 7.

Google Sheets reads the CSV from the link on its own schedule, so the sheet follows new delivered results until the link expires or you revoke it. We ask for no access to your Google account. Anyone holding the link can read that section, so share the sheet with care. An account keeps up to 10 active links. To keep a fixed copy instead, download the CSV and choose File, Import.

Schedule a data export or a PDF report

An account can keep up to 5 schedules: a workbook, CSV or JSON data export, or the PDF report, weekly on Monday at 06:00 UTC or monthly on the 1st at 06:00 UTC. Each file is a complete snapshot of every delivered result up to its run day, with the same fields, limits and checksum as a download.

Scheduled files are delivered as downloads in Exports, with their history: the last 12 files of each schedule, each with its checksum. A file larger than 1,000,000 bytes is listed with its size and can be downloaded on demand instead. We send no email with a scheduled file.

Download the fictional sample files

Sample data, fictional company. Kalvenor Systems is a fictional company, and every company, answer and figure in this file is fictional. The files were prepared from the public sample on 9 November 2026; that is not a measurement date. Their answers are dated from 15 January 2026 to 12 August 2026.

Kalvenor sample exports, 10 files
FileSectionsRowsBytesSHA-256 of the file
kalvenor-sample-all-2026-11-09.xlsxEvery sectionresults 55; receipts 819; sources 405; fixes 7123,916edefb5230a3fd82aa9e673dea2bdd3927abfdcd2ee31c43b3482bcf86f40cbeb
kalvenor-sample-all-2026-11-09.docxEvery sectionresults 55; receipts 819; sources 405; fixes 794,5085c556fec8efddff3a3fa872b304fa2d3e0142d0c7e9ddaab82bfa5c018614102
kalvenor-sample-all-2026-11-09.mdEvery sectionresults 55; receipts 819; sources 405; fixes 7819,74787b5dbb07d97e51edfd853080b9a59b7847b297025d24ba7b315f84a00ad2a62
kalvenor-sample-all-2026-11-09.htmlEvery sectionresults 55; receipts 819; sources 405; fixes 71,174,239bbd02af480983a500be62cbf08f113d5ad6e3e3ff44e081a266054a6508167b2
kalvenor-sample-all-2026-11-09.jsonEvery sectionresults 55; receipts 819; sources 405; fixes 7991,9797e4ab3450b3f87449c96aba6e7db9f02025d274c13e5b294b38d7a8369e1e273
kalvenor-sample-all-2026-11-09.pdfEvery sectionresults 55; receipts 819; sources 405; fixes 7197,2297b68c432aa5b9c409b3f8853ba2eefb8146b9df27d2731797f1c0ff4feaca223
kalvenor-sample-results-2026-11-09.csvresultsresults 5522,8696d133950bebb495ffc2b95ca5f259bbdb23644eb3b09cbe4014771c1f71dbc73
kalvenor-sample-receipts-2026-11-09.csvreceiptsreceipts 819480,72752a05a338bd664c8ee6fb9daa737e74b70e0f708ad2b635ab897139bdc51e2fd
kalvenor-sample-sources-2026-11-09.csvsourcessources 40586,551b0c9fb27b0cb6c9cecf16d17a23ef3083e7bb63efb094226dfcdff3060a812fb
kalvenor-sample-fixes-2026-11-09.csvfixesfixes 73,990b5ba59da7aeeb81b6258b96c48b4d8ca572e2550c5e12db77ec2f74c44a50e74

The checksum line inside the complete files reads SHA-256 of the data: 625207e8f27b875ba2f584a93ce652b998f88756ae28c8247b9abe43a0ae1826. The manifest lists each file's data checksum. Open the JSON file to recompute it.

Related: reproduce a rate from its evidence, check a complete account archive, review a change and its publication record and read every integration, including the Looker Studio and NinjaCat data sources that serve these sections.