Concierge CDP · The analytics accelerator for associations
Answers that show their work.
Ask a question in plain English. Concierge returns the number, the chart and exactly how it was read: the definition, the filters, the query and the members behind it.
How it was read
View the query
01 The problem
One board question, two ways to answer it.
Most associations already have the data. What they don't have is a way to get a trusted number out of it without a queue of hand-offs.
What most associations do today
- 1
The director asks an analyst for the board report.
- 2
The analyst writes up the requirement.
- 3
A developer builds the query or report.
- 4
QA checks the numbers.
- 5
The figures go into the board deck.
With Concierge CDP
- 1
The director asks Concierge in plain English.
- 2
Concierge returns the KPI in whatever form is needed: a chart, a table, a map or an Excel file.
- 3
The director drills to the lowest grain, the members or orders behind the number, and sees that it is right.
Time
Days per question
Seconds per question; follow-ups refine the same answer
Effort
Four people, four hand-offs
One person, no hand-offs
Accuracy
Checked after the fact
Every number drills to the records behind it
Consistency
Depends on who built it
One governed definition, used by everyone
02 Governed AI, not guessed AI
Same question. Same number. However you word it.
Many AI analytics tools hand the model your raw tables and let it work out the rules each time, so the answer can move with the wording.1 In Concierge, the model reads only the question. The semantic layer decides what it means.2
Concierge CDP
Pick a wording. Watch what it resolves to.
Typical AI analytics
The model reads the tables and writes new SQL each time.
| Typical AI analytics | Concierge CDP |
|---|---|
| AI reads raw tables and writes SQL for every question | Concierge works on a warehouse built for questions |
| The model infers your business rules each time | Rules are applied once, in the data load, for everyone |
| The number can change with the wording | Same question, same number, and every drill-down ties out |
| Hard to see how an answer was reached | Every answer shows its definition, filters and query |
| Everyone sees everything, or nothing | Each role sees only its own views; sensitive fields stay out |
03 How it fits together
Start most of the way there, not from zero.
Most analytics projects begin with a blank page: extract, model, argue about the rules, build dashboards, and only then think about AI. Concierge arrives with the warehouse, the association business rules, the semantic layer and the AI agent already built. Your team maps your data and tunes the rules.
-
Sources
Your systems
AMS, events, store, email, finance. They stay exactly as they are.
-
Connect
Plug in*
Ready-made connectors, or loads from the extracts you already have.
-
Rules
Data load
Grace periods, retention, new versus renewing: decided once, the same in every report.
-
History
Warehouse
Every member as they were each month, and every order line by day.
-
Meaning
Semantic layer
Definitions, synonyms, guardrails and who may see what.
-
Answers
Ask Concierge
Questions, boards, maps, alerts and assistants.
The semantic layer is what keeps the AI honest. It tells Concierge which view answers a question, what each column means, and which values and rules apply, so it picks the right objects instead of guessing.
* Ready-made connectors: we're working on it. Today, we load from the extracts and copies you already hold. The brand of your source system does not decide whether Concierge fits.
04 Every member, every month
A warehouse that remembers.
Your AMS knows who a member is today. Concierge keeps who they were in every month, so “how many were in grace last March?” has a real answer, and the history is there to learn from.
Snapshot
Mar 2025
- Status
- Member type
- Term ends
- Dues billed
Ask Concierge
Members in grace at each monthly snapshot · illustration with sample figures
Renewal Outlook
Not only what happened. Who is likely to renew next.
Illustration with sample figures.
- Every upcoming renewal, scored.Each member whose term ends in the next 12 months gets a probability of renewing, a band from “very likely” to “least likely”, and the reasons behind it.
- A forecast you can ask for.“Projected renewals for the next six months” or “dues at risk by state”, answered as a chart or a map.
- Built the governed way.The model learns only from what was known at the time, is tested on a year it never saw, and is scored in the monthly load like any other rule.
05 One platform, ten roles
Everyone gets the answer they need, and only that.
From the executive to the member, each role has its own view. Sensitive fields stay out of the views that shouldn't see them.
Ask, pin, and the board keeps up.
Directors ask in plain English and pin the answer. Every card re-runs when the board opens, so the board is always current.
- Analyse any cardOpen a card in Ask Concierge, break it down by member type, state or period, then update the card or pin a new one. Exploring never changes your board by accident.
- PulseTracks each KPI's trend, flags unusual moves and names the biggest driver.
- Any form you needChart, table, map, funnel or Excel, drillable to the records.
The real product, with a demo association and synthetic data.
Every rep opens the call knowing the member.
- The whole member, one viewMembership, balance, orders, events and activity, with no four-screen dig.
- The slipping signal, before helloEngagement score, grace status and balance due, right at the top.
- Verified firstDetails stay masked until the caller is confirmed; money actions need a staff click.
The real product, with a demo association and synthetic data.
Pay $14.00
The member asks, and gets it done.
- Ask in plain languageBalance, orders, events and membership status, from the member assistant.
- Pay in one tapOpen orders and renewals sit on the member's home screen.
- Only their own recordEvery answer is read from that member's data, nobody else's.
Illustration of the member assistant.
Your rules, editable. No code.
Administrators change synonyms, rules and who sees what in the semantic layer. Concierge picks up the change on the next question.
- One meaning per field770 governed fields across membership, revenue by line of business and renewal outlook.
- Guardrails, in writingThe do's and don'ts Concierge follows, visible to anyone who asks.
Illustration of the semantic-layer editor.
06 Before you buy anything
Six questions to ask any analytics vendor.
Including us. These are the questions that separate a trusted number from a confident guess.
If I ask the same question two ways, do I get the same number?
Concierge: yes. Wording resolves through synonyms to one governed definition.
Can I see how an answer was reached?
Concierge: every answer shows its definition, filters and query.
Can I click from the number to the people behind it?
Concierge: any bar, state or row opens the members or orders behind it, and they add up to the total.
Where do the business rules live?
Concierge: in the data load. Grace periods and retention are decided once, with you, and are the same in every report.
How many members were in grace last March?
Concierge: monthly snapshots keep every member as they were, so point-in-time questions have real answers.
Who sees what?
Concierge: each of ten roles sees only its own views; sensitive fields stay out of the rest.
07 Straight answers about us, too
What's built, and what we're working on.
We'd rather show you a shorter list that works in the demo than a longer one that doesn't.
In the product today you can see it in the demo
- Plain-English questions answered as a chart, table, map, funnel or Excel file
- Drill from any number to the members or orders behind it
- A director's board that stays current, with analysis on any card
- Pulse: trends, unusual moves and the biggest driver
- Monthly membership snapshots and daily order lines
- Semantic layer with synonyms, guardrails and role access; personal data kept out of answers
- Renewal Outlook: scored renewals, bands, reasons and dues at risk
- Cohorts by join year, and students who become professionals
- Assistants for customer service and members, with verified actions
- An audit log of sign-ins and sensitive actions
We're working on it not shown as working
- Ready-made connectors for common association systems
- Board-wide filters that slice every card at once
- A question helper: type ? and build a question from your own definitions
- Answers from your documents (rules, procedures, manuals), with the passage used
- Single sign-on
- An email when a watched number moves
- Concierge from Claude Desktop and other AI assistants
08 The pilot
Prove it on one report, in 90 days.
No AMS replacement and no change to how your staff work today. We load your data, set it up with your definitions, and your staff use it for real work.
- Weeks 1–2
Scope
One recurring report, three to five KPIs and a measured baseline.
- Weeks 3–5
Connect and define
Load the data, agree the definitions, reconcile to your controls.
- Weeks 6–11
Use and measure
Your staff run the report and repeat an agreed task.
- Weeks 12–13
Decide
A shared scorecard, and your decision on what comes next.
Bring ten questions your staff ask today. We'll answer them from your own data.
We are taking a small number of pilot associations, of any size. It starts with a working session on one reporting workflow.