Skip to content

Platform · Sfere ID

Know every fan.
One consented profile.

Sfere ID is a sovereign CDP and real-time identity graph. It resolves fragmented, anonymous touchpoints into one governed golden fan profile, in-Kingdom, consented, and yours to own.

  • Real-time graph
  • Multi-identifier
  • In-Kingdom · PDPL
  • Yours to own

One week, one fan

The story of Sara,
as the graph sees it

Six ordinary moments from one week. No configuration, no rules to learn - just scroll, and watch a blank card become a person. The card on the right is the only thing to follow.

fig. 01 · monday

A stranger browses.

Someone reads two pages on the website and leaves. No login, no name - the only thing anyone saw is a cookie: c_9f2. Your analytics counts a "visitor". That's the whole relationship so far.

→ the card gets its first mark: one dashed chip, and an UNSIGNED stamp.

fig. 02 · wednesday

A ticket buys a name.

At the box office, someone buys a ticket - and gives an email: sara@…. Now the card has a holder: Sara. The ticket number t_8841 is stapled on with it.

→ the HOLDER field writes itself in. Stamp: IDENTIFIED.

fig. 03 · thursday

The cookie gives her away.

Evening: a stream login from a laptop - carrying the same cookie as Monday's stranger. So Monday was never a stranger. It was Sara, before we knew her. History rewrites itself onto her card.

→ the cookie chip pulses: Monday's visit back-fills into Sara's story.

fig. 04 · friday

The phone joins too.

The app opens on a phone, then logs in with the same email. A loyalty scan follows at the stadium gate. Phone, laptop, inbox, ticket, gate - five ways she shows up, one person behind all of them.

→ gold foil slides on. Stamp: GOLDEN PROFILE. Fan № 48211.

fig. 05 · saturday

The tablet tempts the graph.

A family tablet: kids' shows all afternoon, then Khalid logs in, then Lama. One device, two different people. A greedy graph would fuse them into one "fan" and call it complete. This one refuses.

→ the shared device tries to bridge two cards. The guardrail blocks it: HELD APART.

fig. 06 · the count

One week. Nine records. Three people.

Your tools counted nine separate records this week. The graph holds three fans - and one honest "we don't know yet". Nobody invented, nobody merged by mistake, every join can show its evidence.

0records your tools counted
0resolved fans
0held apart, on purpose
Sfere ID · fan cardfig. 00
?
-
-
identifiers - none yet cookie c_9f2 email sara@… ticket t_8841 device d_311 device d_900 loyalty l_2210

waiting for the first touch…

ids: 0 · status: blank

Same records, different rules

What if?

The story you just watched is the graph under Sfere ID's defaults. Flip each scenario to see the same week resolve a different way - because the rules are yours to define.

If email weren't evidence…

Fan #48211 · Saraemail · loyalty · ticket · 2 devices · cookie⚡ golden profile

with email as evidence: one fan

web + streamcookie c_9f2nameless
ticket t_8841no linkable evidencenameless
app + gatedevice d_900 · loyalty l_2210nameless

same week, three "strangers"

Take one identifier away and one fan lands as three separate records. This is why identity can't hang on a single hard key.

If the guardrail were off…

Khalidkhalid@…held apart
Lamalama@…held apart

guardrail on: two people, kept two

khalid@… + lama@…fused by shared device d_tab⚠ over-merged

two real people, one wrong profile

Without the guardrail, the family tablet collapses a marriage into one "fan" - and every message downstream is wrong for one of them.

If a fan is a household…

Khalidkhalid@…held apart
Lamalama@…held apart

definition: one person

🏠 Household H-07khalid@… · lama@… · shared tabletone season-pass household

definition: household - merged on purpose

Decide what a fan means for your business - one person, a family sharing a season pass, a corporate box - and resolution follows your model.

How it works

Identity that holds up under audit

From the first anonymous visit to one governed profile, Sfere ID keeps resolution accurate, explainable, and current at every stage.

  1. 01

    Recognize more fans, in real time

    The same fan shows up as a phone at the gate, a laptop on the stream, and an anonymous browser in between. Sfere ID joins those appearances into one profile, so recognition survives every device switch.

  2. 02

    Real-time stitching, not overnight batches

    The moment a fan acts is the moment to respond. Profiles update in real time as signals arrive, with no waiting on a nightly batch to learn what already happened.

  3. 03

    Fan definitions that fit your model

    Decide what a fan means for your business: one person, a family sharing a season pass, or a corporate box holder. Resolution follows your definition instead of forcing one shape.

  4. 04

    One journey, read as one story

    Anonymous behavior is tracked from the first touch and joined to the profile once a fan identifies, so the pre-signup history and everything after it read as one relationship.

  5. 05

    Stitching rules you define

    Resolve on any combination of identifiers rather than one hard key, and apply your own data-quality rules so identity stays accurate as profiles merge.

Consent & residency

Governed at the data layer,
not bolted on

The hardest part of a fan graph isn't collecting data; it's owning it responsibly. Consent, residency, and deployment are built into Sfere ID from the profile up, so the same record that powers activation is also the one that keeps you compliant.

  • Consent is captured and enforced at the profile itself: governance built in, not patched on top.
  • Fan data is resolved and stored in-Kingdom, PDPL-governed at the data layer.
  • Deploy in your own cloud, on-premise, or fully managed. The choice is yours.
  • Your fan data never leaves your control. Activation decisions run inside your environment.
See data residency

Connect your data

Connect every source, keep it yours

Sfere ID sits on an open connector framework and a warehouse you self-manage. Bring the touchpoints you already have, with no rip-and-replace and no vendor holding your fan data hostage.

  • 01

    An open connector framework ingests events from ticketing, streaming, app, web, on-site, and social, with no parallel database to maintain.

  • 02

    Map your existing user IDs with custom identifiers and attributes, so the profile fits the data you already keep.

  • 03

    Stream to a columnar analytical warehouse you self-manage in-Kingdom, so the platform stays composable with your stack.

  • 04

    A clean REST API and client SDKs cover anything custom, so every capability is programmable.

Read the developer docs

How resolution works

A graph that gets identity right,
not just complete

Resolution matches signals on any identifier, merges them into one golden record, and applies guardrails so distinct fans stay distinct.

  1. 01

    Identifiers, not one hard key

    Email, device, app ID, ticket or order number, loyalty ID, custom IDs: each is a first-class identifier with its own merge rules. Resolution connects events on any of them, so you are not forced to hang identity on a single field that is often missing.

  2. 02

    A live graph that merges and splits

    New signals either match an existing profile, create one, or merge several into a golden record as the graph learns the relationships. Contested matches follow rules you set, rather than silently collapsing two real people into one.

  3. 03

    Guardrails against false merges

    Per-identifier limits stop a shared device or family login from over-merging distinct fans. When a limit would be crossed, the system holds the records apart instead of guessing. Correct identity beats complete-but-wrong identity.

Put the profile to work

One resolved profile drives segmentation, journeys, and paid audiences in Sfere Activate.

See Sfere Activate

Questions

Identity questions engineers ask

Is resolution real-time or a nightly batch?
Real-time. Profiles update as signals arrive, so the fan you recognize reflects what just happened rather than what a batch job will report tomorrow. Historical and warehouse data map into the same graph so batch context is not lost.
Deterministic or probabilistic matching?
The model is graph-based and rule-driven: it connects events on shared identifiers you define, with per-identifier merge limits and priorities you control. That keeps resolution explainable and auditable rather than a black-box score.
How does consent fit in?
Consent is captured and enforced at the profile itself, not patched on downstream. The same governed record that powers activation is the one that keeps you compliant, and it is resolved and stored in-Kingdom, PDPL-governed at the data layer.
Do we have to move our data to you?
No. Sfere ID sits on an open connector framework and a warehouse you self-manage in-Kingdom. Bring the sources you already have, keep the data in your environment, and stay composable with the rest of your stack.
What keeps the graph from getting polluted by dirty data?
Resolution runs alongside schema validation as events land, so graph hygiene is enforced on the way in rather than repaired by a nightly cleanup job. You apply your own data-quality rules (normalizing formats, catching typos, filtering internal test traffic), and per-identifier limits hold contested records apart instead of guessing a merge.

Know the fan in front of you - while they're still in front of you

Bring your messiest export to a walkthrough. We'll run it through Sfere ID live and show you the golden profiles, the merge evidence behind every one of them, and the records the guardrail refused to guess on.