How Mosaic works: one profile per club member
How Mosaic joins club systems into one member profile: sources, match rules, plain-English questions, segments, consent checks and its current limits.
Mosaic takes the systems that each know part of a club member (membership, point of sale, gaming, bookings, email and SMS, the club app), links their records into one profile per person, and lets staff ask questions, save the answers as segments and build campaigns with consent and exclusions removed first. It runs in your own cloud account or on your own servers, so the profiles stay in a database you own. This post walks through what happens at each step, what a team does in the first week, and what Mosaic does not do yet.
Step 1: Each system becomes a data stream
Every source system is set up as a stream with its own status, record count and refresh time, so staff can see when a feed has failed before they trust a number. Fields from each system are mapped to one member model: mobile_no in the bookings system and phone in the POS both become Mobile. A field-mapping view shows each source field drawn to the member field it fills.
What this means in practice: a failed sync is visible on the stream, and you fix it there. Connecting each live system is done with you during set-up. There is no self-serve connector you switch on yourself.
Step 2: Records are matched into one member
Mosaic keeps every source record as it arrived and links it to one member profile, often called a golden record. Match rules are written in plain words and listed in priority order, for example:
- Exact email
- Exact mobile
- Membership number
- Fuzzy name plus date of birth plus suburb
- Fuzzy name plus postcode, switched off until you trust it
Each rule shows how many records it has matched, and each linked source record carries a confidence score. Records that do not match with confidence go to a list of records waiting for a person to review. The profile then shows which systems it was built from, so a member with six source records shows all six.
Step 3: A profile staff can read
The member page shows tier, member number, visit count, spend, gaming turnover where you run gaming, lifetime value and churn risk, with a timeline that mixes every source: a bistro bill on Friday, a gaming session, a visit to another venue, a newsletter open. Consent appears per channel (email, SMS, push, mail).
Lifetime value and churn risk are calculated from the member's own activity, such as how long since the last visit and how often they come in. They are simple, readable scores, not a black-box model. Treat them as a sort order for attention, not a prediction to bet budget on.
Step 4: Ask in plain English
Staff type questions such as "Members who haven't visited in 90 days" or "Show me my top 10 members by value". Mosaic runs a query on the member data and returns a sentence answer, the key numbers, a chart or table, and a readable version of the query it ran, for example a count of members whose last visit was more than 90 days ago. Where the answer is a group of people, it offers to save them as a segment.
Today the questions it answers fall into set types: lapsed and inactive members, top members by value, food and drink revenue by venue, gaming revenue, tier counts, app and push reach, members by suburb, averages and totals, and the gambling-harm risk spread. A question outside those types gets a list of suggested questions, not a guess. The launch post covers how to test this kind of tool.
Step 5: Segments that stay current
A segment is saved as rules, not as a list of names: "status is lapsed and churn risk is over 55", or "last visit more than 90 days ago". Conditions can be all-of or any-of, and cover tier, status, spend, visits, churn risk, age, suburb, days since last visit and consent per channel. Because the rules are stored, the members behind a segment are worked out again when you open it, so people drop in and out as they visit or lapse.
Step 6: Build a campaign with removals shown
Pick a segment and a channel (email, SMS or app push) and Mosaic shows the drop-off before anything is created:
| Stage | What is removed | |---|---| | Audience | Everyone in the segment | | Consent | Members without consent for this channel | | Exclusions, every campaign | Self-excluded members | | Exclusions, gaming offers | Members at medium or high gambling-harm risk | | Will receive | What is left |
A campaign is treated as a gaming offer when its name mentions gaming, pokies, a jackpot, a raffle or a members' draw. That is a check on the name, so name gaming offers plainly and look at the exclusion count before you go ahead.
The first week, day by day
Timings are a suggested starting point, not a benchmark.
- Day 1. List your systems, their owners and how data gets out of each.
- Day 2. Agree the member model: which fields matter, which system wins for each.
- Day 3. Agree match rules and their order. Decide what a person must review.
- Day 4. Map consent per channel and the exclusion lists, with your compliance lead.
- Day 5. Ask five questions you could not answer last month and check one answer by hand.
The by-hand check matters. If one answer is wrong, trace it to the source record before you trust the rest.
What Mosaic does not do yet
Be clear on these before you plan around it.
- It does not replace your CRM or membership system. It reads from them.
- It does not send messages itself. Delivery through your email or SMS tool is set-up work we scope with you, so ask what is connected for your tools.
- There is no separate approval step before a send in Mosaic yet. Keep your current sign-off.
- The question box covers set question types. Free-form questions outside them return suggestions.
- Exclusion rules are fixed to consent, self-exclusion and gambling-harm risk today. Other exclusions you need, such as members who objected to profiling, must be agreed at set-up.
- There is no self-serve sign-up. A Thinkiyo team sets it up with you and stays after launch.
How to judge any customer data platform
Whether or not you choose Mosaic, put these tests to every vendor:
- Matching on your data. Give them 500 real records, including two different people with the same name. Count wrong merges.
- Lineage. Click any field on a profile and ask which system it came from.
- Consent on merge. When two records disagree on consent, which one wins? It should be the more restrictive.
- Show the query. Every answer should show what it counted. If it cannot, you cannot check it.
- Stale data. Break a feed on purpose. Does the product show it, or quietly use yesterday's numbers?
- Exit. How do you export every profile, segment and consent record if you leave?
The single member view guide covers the privacy and marketing rules in the UK, EU, Australia, India and the US. If you are comparing larger platforms, read our Salesforce Data 360 and Segment comparison.
Where it runs
Mosaic runs in your cloud account, in the region you choose, or on your own servers. Profiles, segments, consent and campaigns sit in a database you own. Contact us with your systems list and your hardest member question.
Common questions
What systems does Mosaic connect to?
It is built for membership and CRM systems, point of sale, gaming, bookings, email and SMS platforms and club apps. Each one is connected with you during set-up and runs as a stream with its own status and record count. Ask about your specific systems before you commit.
How does Mosaic decide two records are the same member?
It applies match rules in the order you set, such as exact email, exact mobile, membership number, then fuzzy name with date of birth and suburb. Each linked record keeps a confidence score. Records it cannot match with confidence wait for a person to review.
Does Mosaic check consent before a campaign?
Yes. When you build a campaign it removes members without consent for that channel and self-excluded members, and for gaming offers it also removes members at medium or high harm risk. It shows how many were removed at each stage.
Can staff ask any question in plain English?
Not yet. The question box answers a set of common question types and shows the query behind each answer. Questions outside those types return suggestions instead of an answer.
Where is our member data stored?
In your own database, in your cloud account or on your own servers. Thinkiyo does not run a shared platform for Mosaic customers.
Sources
This post describes Mosaic itself. The rules it refers to are sourced in the linked guide.