Superset data over MCP — field specification

What it is forThese two use cases are what make a field required or optional

Report

Generating the monthly customer report

ISC reps produce a monthly report for each customer. Today that means exporting CSVs from Superset by hand; we want to generate it through MCP instead. What we report today: call volume, talk rate, appointment rate, companies with appointments, the no-answer / blocked-at-reception breakdown, monthly and weekly trend, and cuts by list, industry and company size.

Churn signal

Detecting churn risk with a health score

We want to catch, across all accounts, when outreach has stopped, when detected companies are going unworked, when an integration has broken, and when logins are falling off. We roll that into a score per contract and work the accounts whose renewal is closest.

The one thing that matters mostIf detection and outreach cannot be joined, both data sets lose half their value

Intent detectionWhich companies moved,
by keyword, visitor or domain

→

OutreachWhether we actually
acted on them

→

OutcomeWhether it led to
a conversation or a meeting

→

Health scoreWhether the account is running,
and whether churn is coming

So please include the target company corporate ID in both data sets

The current Keyword Hit export carries the company name only, which forces us to match on name. With the corporate ID we can show directly how many of the detected companies were actually worked.

Answers to your questionsResponses to the points raised in the thread

Approach resultsYour understanding: the number of sends per channel

That alone is not enough. On top of "was it sent" we need "what happened" (the phase name) and unique counts per company. Neither talk rate nor appointment rate has a numerator without them.

Intent detection results (keyword, visitor, domain)Your understanding: hit counts and the companies

That is right. Counts alone cannot drive any change in how the account is run, so please always include the companies. We would also like the corporate ID of each detected company so we can join it with the activity data — the current CSV has the company name only.

System integration statusYour understanding: Salesforce and other integrations

Beyond whether one exists, we want the last sync timestamp and the most recent error. "Configured but no longer syncing" is exactly what we want to catch as a churn signal.

User login statusYour understanding: login history per user

History plus per-subscription aggregates (active seats against issued seats, WAU / MAU) would let us feed it straight into the health score.

Linear ticketsWhich ticket covers which data set

Linear	Data set
SIG-667	Approach results
SIG-668	Intent detection results
SIG-669	System integration status
SIG-670	User login status

Field specificationField names are the Superset export columns as they are. "Required" means the current reporting cannot be produced without it

Shaded rows are fields that do not exist in the current export — these are the additions we would like to ask for.

Approach results (activity detail)

1 row = 1 activity total 26 fields ／ 24 required

Equivalent to the Superset "活動一覧" (activity list) export. One row per call or send, including what came of it.

Field (Superset column)	Suggested field name	Type	Required	What we use it for	Notes
Subs企業名Subscription company name	subscription_name	string	Required	Identifies the customer
ターゲット企業名Target company name	target_company_name	string	Required	Identifies the company being approached
ターゲット企業法人IDTarget company corporate ID	target_company_id	string	Required	Join key against the intent detection data	The single most important field. Without it we cannot measure detection-to-approach conversion
母業界Industry (major)	industry_major	string	Required	Appointment rate by industry
子業界Industry (minor)	industry_minor	string	Required	Drill-down by industry
正社員数Employee count	employee_count	number	Required	Drill-down by company size
正社員数範囲Employee count range	employee_range	string	Required	Appointment rate by company size
売上高範囲Revenue range	revenue_range	string	Required	Drill-down by company size
資本金Capital	capital	number	Optional	—
資本金範囲Capital range	capital_range	string	Optional	—
都道府県Prefecture	prefecture	string	Required	Results by region
アプローチ方法Approach channel	channel	string	Required	Per-channel results; the denominator for call volume	Call, email, form, ad, SNS, letter
フェーズ名Phase name	phase_name	string	Required	Talk rate, no-answer rate, blocked-at-reception rate	Not "was it sent" but "what happened". This is the pivotal field
フェーズカテゴリーPhase category	phase_category	string	Required	Rolled-up view of phases
フェーズカテゴリー（normalize)Phase category (normalized)	phase_category_normalized	string	Required	Aggregation after normalizing wording
部署情報ありHas department info	has_department	boolean	Required	How call outcomes differ when a contact name is known
担当者接触Contact reached	contact_reached	boolean	Required	Contact-reached rate (the official flag)
活動日Activity date	activity_date	date	Required	Monthly and weekly trend
直近30日シグナルありHas signal (last 30d)	has_signal_30d	boolean	Required	Appointment rate with vs. without a signal	Used to verify whether intent-based targeting works
直近30日キーワードシグナルありHas keyword signal (last 30d)	has_keyword_signal_30d	boolean	Required	Effectiveness by signal type
直近30日ビジターインテントありHas visitor intent (last 30d)	has_visitor_intent_30d	boolean	Required	Effectiveness by signal type
（新規）直近30日ドメインインテントあり(new) Has domain intent (last 30d)	has_domain_intent_30d	boolean	Required	Effectiveness by signal type	The deal list has "domain intent" but the activity list does not. Please align them
アポありHas appointment	has_appointment	boolean	Required	Appointment rate and number of companies with appointments
アプローチ者Approached by	approacher	string	Required	Results per rep
商材Product	product	string	Required	Filter by product	Required for customers running more than one product
コールリスト名Call list name	call_list_name	string	Required	Results per call list

Approach results (per company)

1 row = 1 target company total 44 fields ／ 38 required

Equivalent to the Superset "案件一覧" (deal list) export. Which channels reached each company, and its latest status.

Field (Superset column)	Suggested field name	Type	Required	What we use it for	Notes
Subs IDSubscription ID	subscription_id	string	Required	Identifies the customer; join key against the other data sets
Subs企業名Subscription company name	subscription_name	string	Required	Identifies the customer
ISC担当者ISC owner	isc_owner	string	Required	Managed per rep
ターゲット企業法人IDTarget company corporate ID	target_company_id	string	Required	Join key against both activity and detection data
ターゲット企業名Target company name	target_company_name	string	Required	Identifies the company being approached
母業界Industry (major)	industry_major	string	Required	Distribution by industry
子業界Industry (minor)	industry_minor	string	Required	Drill-down by industry
正社員数Employee count	employee_count	number	Required	Drill-down by company size
正社員数範囲Employee count range	employee_range	string	Required	Distribution by company size
売上高範囲Revenue range	revenue_range	string	Required	Drill-down by company size
資本金Capital	capital	number	Optional	—
資本金範囲Capital range	capital_range	string	Optional	—
最初活動日First activity date	first_activity_date	date	Required	Days elapsed since first contact
最新活動日Last activity date	last_activity_date	date	Required	Detects companies that have gone untouched
アプローチ方法Approach channel	channels	string	Required	Whether multi-channel outreach is actually running	Whether several channels are combined
カテゴリ（アプローチ済み）Category (approached)	approach_category	string	Required	Separates approached from not-yet-approached
最新ステータスLatest status	latest_status	string	Required	Open counts by next action
直近30日シグナルありHas signal (last 30d)	has_signal_30d	boolean	Required	What share of detected companies were actually approached
キーワードシグナルありHas keyword signal	has_keyword_signal	boolean	Required	Coverage rate by signal type
直近30日ビジターインテントありHas visitor intent (last 30d)	has_visitor_intent_30d	boolean	Required	Coverage rate by signal type
ドメインインテントありHas domain intent	has_domain_intent	boolean	Required	Coverage rate by signal type
直近7日キーワードKeywords (last 7d)	keywords_7d	string[]	Optional	—	The last-30-days field is sufficient
直近30日キーワードKeywords (last 30d)	keywords_30d	string[]	Required	Which keyword triggered the detection	Used for per-keyword performance analysis
直近90日キーワードKeywords (last 90d)	keywords_90d	string[]	Optional	—	The last-30-days field is sufficient
直近7日ビジターインテント閲覧ページVisited pages (last 7d)	visited_pages_7d	string[]	Optional	—	The last-30-days field is sufficient
直近30日ビジターインテント閲覧ページVisited pages (last 30d)	visited_pages_30d	string[]	Required	What topics the company is interested in
直近90日ビジターインテント閲覧ページVisited pages (last 90d)	visited_pages_90d	string[]	Optional	—	The last-30-days field is sufficient
リスト名List name	list_name	string	Required	Progress per list
リスト数List size	list_count	number	Required	Size of each list
コールありHas call	has_call	boolean	Required	Which channels reached the company
コール架電日Call date	call_date	date	Required	When each channel was used
メールありHas email	has_mail	boolean	Required	Which channels reached the company
メール送信日Email sent date	mail_sent_date	date	Required	When each channel was used
フォームありHas form	has_form	boolean	Required	Which channels reached the company
フォーム送信日Form sent date	form_sent_date	date	Required	When each channel was used
フォームクリックありHas form click	has_form_click	boolean	Required	Response via the form
フォームクリック日Form click date	form_click_date	date	Required	When the response happened
広告ありHas ad	has_ad	boolean	Required	Which channels reached the company
広告アプローチ日Ad approach date	ad_approach_date	date	Required	When it happened
広告インプレッションありHas ad impression	has_ad_impression	boolean	Required	Ad reach
広告インプレッション日Ad impression date	ad_impression_date	date	Required	When it happened
広告クリックありHas ad click	has_ad_click	boolean	Required	Ad response (a headline metric in the report)
広告クリック日Ad click date	ad_click_date	date	Required	When the response happened
アポありHas appointment	has_appointment	boolean	Required	Companies with appointments (company-based appointment rate)

Intent detection results

1 row = customer x keyword x week (visitor / domain intent is per company) total 13 fields ／ 12 required

Equivalent to the Weekly Hit tab of the Superset "Keyword Hit Dashboard". Hit counts together with the detected companies.

Field (Superset column)	Suggested field name	Type	Required	What we use it for	Notes
subscriptionidsubscriptionid	subscription_id	string	Required	Identifies the customer	Existing column in the Keyword Hit Dashboard
companynamecompanyname	company_name	string	Required	The detected company	Existing column
（新規）検知企業の法人ID(new) Corporate ID of the detected company	company_id	string	Required	Join key against the activity data	Not in the current CSV. Please add it so we do not have to match on company name
weekweek	week	date	Required	Weekly hit trend	Existing column
targettarget	target_product	string	Required	Operating results per product	Existing column. The same keyword is sometimes run for several products
keywordkeyword	keyword	string	Required	Hits per keyword	Existing column
keyword_start_datekeyword_start_date	keyword_start_date	date	Required	When the keyword started running	Existing column
keyword_expire_datekeyword_expire_date	keyword_expire_date	date	Required	When the keyword was rotated out	Existing column
keyword_statuskeyword_status	keyword_status	string	Required	Distinguishes running from stopped	Existing column
hit_30dayhit_30day	hit_30day	number	Required	Hits in the last 30 days	Existing column
hit_90dayhit_90day	hit_90day	number	Optional	—	Existing column. The 30-day figure is sufficient
（新規）ビジターインテント検知企業(new) Companies detected via visitor intent	visitor_intent_companies	object[]	Required	Which companies visited the customer's own site	Company name, corporate ID, visited pages and detection date as a set
（新規）ドメインインテント検知企業(new) Companies detected via domain intent	domain_intent_companies	object[]	Required	Detection at the domain level	Company name, corporate ID and detection date as a set

System integration status

1 row = customer x integration total 9 fields ／ 9 required

Not only whether an integration exists, but whether it is still alive (last sync, errors).

Field (Superset column)	Suggested field name	Type	Required	What we use it for	Notes
Subs IDSubscription ID	subscription_id	string	Required	Identifies the customer
連携種別Integration type	integration_type	string	Required	What the customer is integrated with	Salesforce, HubSpot, others
連携有効フラグIs enabled	is_enabled	boolean	Required	Whether an integration exists	This is the "SF and other integrations" you asked about
連携方式Auth method	auth_method	string	Required	How the integration is set up	OAuth, API key, etc.
連携設定日Connected at	connected_at	date	Required	How long the integration has been in place
最終同期日時Last synced at	last_synced_at	datetime	Required	Tells us whether the integration is still alive	We want to detect "configured but no longer running"
直近同期ステータスLast sync status	last_sync_status	string	Required	Whether the sync succeeded	Success / failure
直近同期エラー内容Last sync error	last_sync_error	string	Required	Identifies why the sync stopped	We want to pick this up as a churn signal
同期対象オブジェクトSynced objects	synced_objects	string[]	Required	How deep the integration goes	Lead, account, opportunity, etc.

User login status

1 row = customer x user (plus per-subscription aggregates) total 10 fields ／ 10 required

Field (Superset column)	Suggested field name	Type	Required	What we use it for	Notes
Subs IDSubscription ID	subscription_id	string	Required	Identifies the customer
ユーザーIDUser ID	user_id	string	Required	Identifies the user
ユーザー名／メールUser name / email	user_name	string	Required	Identifies who is actually using the product	Used to detect when the internal champion has moved on or left
権限ロールRole	role	string	Required	Whether the administrator is active
アカウント有効フラグIs active	is_active	boolean	Required	The denominator of issued seats
最終ログイン日時Last login at	last_login_at	datetime	Required	Detects disengagement	A core input to the health score
期間内ログイン回数Login count in period	login_count	number	Required	Usage frequency
（集計）発行ID数(aggregate) Issued seats	issued_seats	number	Required	How widely the contract is actually used	Please return this as a per-subscription aggregate
（集計）期間内に稼働したID数(aggregate) Active seats in period	active_seats	number	Required	Utilization (active seats / issued seats)	Aggregate per subscription
（集計）週次／月次アクティブユーザー数(aggregate) Weekly / monthly active users	wau_mau	number	Required	Trend in usage, up or down	Aggregate per subscription

What we would like from the tool interfaceThe part the field list alone does not settle: how it is called and how much comes back

Scope and its limitsFilterable by Subs ID. Internal use spans all subscriptions; when this is exposed to a client it must be restricted to their own subscription. (Permission boundaries are hard to retrofit, which is why we raise it now.)
Date rangeA from / to range over activity dates and weeks.
Control over how much comes backWe need an "aggregates only" mode, plus row limits and pagination on detail queries. An MCP response goes straight into the AI's context, so a tool that returns every detail row — several thousand for one customer-month — is effectively unusable. This is the one real difference between this and simply putting the Superset screens behind an API.
Data freshnessThrough the previous day is plenty (the same as Superset today).