Skip to main content
Orgo answers questions about your data in three ways. Ask Orgo is a chat where you type a question in your own words and get a table or chart back. Analytics is a set of prebuilt dashboards for membership, geography, activity, and money. Reports is a builder you fill in yourself, for the extract you download again every month. Built for admins who need a number now and do not want to file a report request or export three spreadsheets to get it. Replaces ad hoc CSV exports, spreadsheet pivot tables, and asking a developer to run SQL. Ask Orgo welcome screen with a greeting, a question box, and suggested question chips such as active members per local center

​
Where to find it

Analytics needs the analytics module active. The module switch is on Settings → Modules → All Modules (/tenants/settings/features), under analytics. The Minimum access level and the per-tab roles are on Settings → Modules → Users & Profiles → Security & Permissions, which collects those dropdowns from every active module onto one page.

​
Ask Orgo

Type a question and press Enter. Questions are capped at 1000 characters, and you can attach one image (PNG, JPEG, GIF, or WebP, up to 5 MB). The welcome screen offers eight starting points, including Active members count per local center, Revenue by product type this year, Members who have never made a payment, and Top 10 members by discussion posts. Ask Orgo replies with one of three things:
  • a result: a summary sentence, a table, and a chart when the shape of the data suits one;
  • a clarifying question, when the request is too vague to turn into a query;
  • a segment proposal, when you asked it to build or save a list rather than show one. It shows the criteria and an estimated member count, with Open in editor to finish it as an email list.
Follow-ups stay in the same chat and modify the previous query, so “add their email”, “only active ones”, “break it down by month”, and “show as a chart” all work. Name the values you want, in business terms, rather than pointing at rows on screen: the assistant cannot read the result table. A conversation keeps its most recent 20 messages; your chat list shows your own last 50 conversations, newest first, with a search box. You can delete a chat, and rate an answer with the thumbs up and down buttons. Ask Orgo carries a Beta badge, and the page warns that it can make mistakes and that database entries are worth checking.

​
What Ask Orgo can reach

Ask Orgo has no free access to the database. It works from a fixed catalogue of 55 record types and, within each, a hand-picked list of fields. Some are reachable only through a related record: registrations always through their event, prices through their product. Not in the catalogue: credentials, tokens, and session data. Identity verification records expose only the upload date, validation status, and validation date, because ID and personal numbers are stored encrypted and are not queryable (see Identity validation). Voting records expose only whether a member voted, never which option they chose (see E-voting).
Values stored in encrypted custom fields come back in their stored encrypted form through Ask Orgo, because results are read as raw rows rather than as full records. Read those values on the member profile instead.

​
How a question becomes an answer

1

Your question goes to the model

Orgo sends Anthropic’s Claude API your question, the catalogue above (record and field names with their descriptions), and the structured queries from earlier turns in the same chat. It does not send result rows, which is why a follow-up cannot refer to “those ones” and has to name the values.
2

The model returns a query description, not SQL

A structured description: which record type to start from, which related records to join, filters, columns, grouping, and sorting.
3

Orgo validates it

Unknown record types, fields, and operators are rejected. A query may use at most 6 joins, 20 filters, 20 columns, and 5 aggregate filters, and the page size is clamped to 500 rows. Queries that would inflate totals by combining several one-to-many relations are refused with an explanation rather than returning a wrong number.
4

Orgo runs it against your tenant only

Before anything else is added, the compiler pins the query to your organisation, and every joined record hangs off that filter. Only SELECT statements are allowed, and a non-SELECT is refused before it reaches the database.
The round trip has a 30 second budget. A query that is too heavy comes back asking you to narrow it: a date range, one local center, or a count instead of the detail rows.
Ask Orgo runs with full tenant-admin reach: an ADMIN_TENANT user can ask about any member in the organisation, including data they would have to click through several profiles to see. Grant ADMIN_TENANT accordingly. Local admins do not get Ask Orgo at all.

​
Working with a result

Ask Orgo conversation showing a summary sentence above a result table with column headers and a row count Column headers come back in the language you asked the question in.

​
Analytics

Analytics Platform tab showing the membership and financial summary cards, contact, chapter and subscriber tiles, and a row of activity sparklines with a period selector Tabs appear only if you hold the role configured for them. Defaults: The Platform tab is also open to ADMIN_LOCAL and ADMIN_PARENT_LOCAL when local centers are in use, scoped to their own local center and its descendants, never organisation-wide. Membership and geography tabs accept a local center filter and check the local equivalent of the tab role against it. Period selectors offer 7 days, 30 days, 90 days, 12 months, and a custom range. Daily totals are aggregated overnight, at 04:00 UTC for the previous day, and the page says so. Today’s figures are computed live and added on top, and responses are cached for five minutes. Reset cache next to the page title clears that cache and reloads, which is the fix when a number looks stale after a bulk import or a large payment run.

​
Database Queries

Anything you save from Ask Orgo lands here as a named query that admins can run again without the chat. Opening one runs it and shows the results, with Export for a download. It is stored as the finished statement, with your organisation’s filter baked in, so it keeps returning your data only. FINANCIAL_TENANT admins see every saved query in the organisation; everyone else sees only the queries shared with them, listed under Extra access on the query. Every run is recorded in the activity log with the query name and who ran it.

​
Troubleshooting

Give it a scope: “how many active members joined each month this year” works better than “membership trends”.
Two settings are checked before the page renders, and failing either one redirects you to the dashboard with no message. The analytics module has to be active, and you have to clear the analytics setting labelled Minimum access level, which ships as any signed-in member.Worked example: an organisation raises Minimum access level to HR_TENANT so that ordinary members stop seeing the numbers. That works, but it also takes away the Geography and Leaderboards tabs, which default to any signed-in member and were the two tabs the organisation actually wanted everyone to have. The minimum is a gate on the whole page, evaluated before any per-tab role.If you get in and the page is empty, the minimum passed and none of the tab roles did.
Finance needs the role and online payments switched on. The tab checks both, so an organisation that collects fees by bank transfer only never sees it, whoever is looking. Revenue totals still show on the Platform tab.DB Query is the other tab keyed to FINANCIAL_TENANT, and it does not check the payments setting, so seeing Database Queries but no Finance is the normal shape of that combination rather than a fault.
No, on both counts. The route, the sidebar entry and every endpoint behind the chat require ADMIN_TENANT, and the sidebar entry is written into the menu in code rather than drawn from your configurable menu, so it cannot be added to another role’s menu or moved.Someone without ADMIN_TENANT who follows a link to Ask Orgo is sent to the login screen rather than shown a permission message, which reads like an expired session and is not one. The same is true of any page in Orgo that they lack the role for. There is no read-only or delegated version of the chat; if a colleague needs an answer regularly, save the query and give them access to it, with the caveat in the next answer.
Sharing works, the way in does not. Adding somebody under Extra access lets them run that query, and once at least one query is shared with them a queries entry appears in their sidebar. That entry points at the Database Queries tab of Analytics, and the tab itself renders only for FINANCIAL_TENANT, so they land on Analytics without it.Send them the query’s own address instead, the one in your browser bar when you have the query open. That page has no role gate of its own and checks the share, so it opens for them and records the run in the activity log under their name.The related setting is the analytics hasQuery flag: turning it off removes the sidebar entry for everyone but leaves the tab in place for finance admins.
It refuses when the arithmetic would be wrong rather than returning a number that looks fine.Worked example: ask for total fees paid and the number of events attended, per member. Both hang off the member as separate one-to-many branches, so joining them multiplies every fee row by every attendance row: a member with 4 payments and 5 events contributes 20 rows, and the fee total comes back five times too high. Orgo detects the shape, blocks the sum, and tells you to query each metric separately.Ask it as two questions, then line the answers up by member in a spreadsheet, or ask for counts rather than sums when you only need magnitude.
The export from the chat asks for pages of 10,000 rows, but every query, including that one, is capped at 500 rows per request. So a result of more than 500 rows exports as a single file holding the first 500, and the completion message still says the export finished.Worked example: a question matching 3,000 members reports 3,000 in the chat and downloads a CSV with 500 rows in it.For a full extract, use Save to store the query, then run and export it from Database Queries. Saving deliberately drops the paging limit, so the saved statement returns the whole result set. The other option is to narrow the question until it fits, one local center or one month at a time.
Saving removes any row limit of 50 or more, on the assumption that it came from the default paging rather than from your question. Limits below 50 are kept, because those look like something you asked for.Worked example: “top 100 members by discussion posts” answers with 100 rows in the chat. Saved and re-run from Database Queries, it returns every member ranked by posts, which for a large organisation is a much heavier query than the one you tested. “Top 25” saves as 25.If you want a fixed cut-off in a saved query, ask for fewer than 50, or add the real filter you mean (a date range, a status) instead of relying on the count.
A conversation keeps 20 messages: the first two, which hold the original framing, and the most recent eighteen. Turns in the middle are dropped, so in a long session a follow-up can no longer see a constraint you set twenty turns ago. Restate it, or start a fresh chat for a new line of questioning.Reopening an old chat shows at most 50 rows of the last result, because that is the size of the snapshot stored with the conversation. It is a record of what you saw, not a live result. Re-run the question to page through everything or to pick up data that has changed since.
Your chat list shows your own conversations only, and there is no shared inbox of everyone’s chats. But conversations are stored per organisation rather than sealed per person, and another organisation admin who has the address of one can open it.In practice this changes little, because the feature is limited to ADMIN_TENANT and any admin can ask the same question themselves and get the same data. Treat chats as a shared admin workspace rather than private notes, and delete a conversation when you want its text and its stored rows gone.
Reset cache requires ADMIN_TENANT, while the tabs themselves can be opened by HR_TENANT, FINANCIAL_TENANT or ordinary members depending on your settings. A finance admin looking at a stale figure after a large payment run has no button to press and has to wait out the five minute cache, or ask an organisation admin.Reset cache also only clears the five minute cache. It does not rebuild the overnight aggregation, so a figure that is a whole day out rather than minutes out will not move until the next nightly run, however many times you press it.
Check the assumptions in the summary line. Ask Orgo defaults to active members, successful payments, and published events unless you say otherwise. Ask “include inactive members too” to widen it.
Result rows are never sent. What leaves Orgo is your question, the catalogue of record and field names, and the structured queries from the same chat; an attached image goes with the question and is not stored on the conversation afterwards. The conversation itself lives in your organisation’s own data as the message text, the structured query, and a snapshot of up to 50 rows from the last result, so reopening a chat does not re-run the query. Deleting the conversation removes all of it.