The talent intelligence system you talk to.
You already employ hundreds of the people who could be running this business in five years. You can’t say which ones, and you can’t develop them all the way you develop the top 50.
Spherikal is coaching every one of them will actually use — and the evidence you’ve been missing builds while they use it.
Most tools hand you a confident number. We hand you the evidence — and how sure we are.
We say the thing the other tools won’t: insufficient evidence is an answer. We never fake a score. If we can’t stand behind a read, we show you the gap instead of painting over it.
Why this reads 4.2
The point is 4.2 and the band around it runs 3.9 to 4.5. It is that tight because three separate sources point the same way — and one of them is not the participant.
One governed loop, not five disconnected tools.
Every other stack makes you the integration layer — exporting from one tool, re-keying into the next, losing the thread. Spherikal closes the loop and keeps the signal moving through it: the three stages are one system, and each stage’s output is the next stage’s input.
- — which sharpens the next program. The loop compounds.
Your talent signal is a byproduct, not a program.
The assessments and feedback rounds in Spherikal aren’t done to your people to produce a report for somebody else. They’re the same instruments that produce that person’s own coaching — so the read your succession and investment conversations have always lacked arrives as a byproduct of the development itself, not as a separate program you have to run.
Five line items collapse into one system, one signal.
You’re already paying for most of this — in separate contracts that don’t talk to each other. Spherikal is the consolidation, not another tab.
Pick a line item to see what stands in its place.
Assess. Coach. Develop. Sense. Decide.
Connected capabilities on one spine. Turn on what you need, when you need it — nothing you switch on is a silo. Open a layer to see what it holds.
Assess
Take in whatever evidence about a person actually exists, weight each source by what it can genuinely support, and be explicit about where it runs out.
Included in the base platform.
- An embedded battery — Short, separately-takeable self-assessments built from public-domain items, taken inside the product. Nothing to procure, nothing to schedule.
- The reports you already own — Upload a personality inventory, a 360, or a development report. Spherikal reads it, classifies its scales, and puts it to work without re-testing anyone.
- Fusion that weights the source — Sources carry different reliability, and a self-rating that outruns the disposition behind it is shrunk rather than believed.
What you getOne profile per person, competency by competency, carrying the evidence behind it and how far that evidence can honestly be pushed.
Coach
A coach every person can talk to that already knows what their assessment said — and that nobody else can read.
Included in the base platform.
- Private by construction — An individual’s coaching is theirs. The isolation is enforced at the database layer, not promised in a policy document.
- It starts from the measurement — Profile, goals, program context and everything shared so far are in hand before the first word, so nobody re-explains their job every session.
- Commitments with dates — What gets committed to in conversation becomes a dated row. Whether it slipped is arithmetic, not an impression, and the next session opens with it.
What you getA coaching relationship that compounds instead of restarting, and a plan that gets updated by the work rather than by a form.
Develop
Turn a read into reps: named next actions, evidence that they happened, and progress that only moves when something actually accumulated.
Included in the base platform.
- Chosen, not assigned — Sphere proposes focus areas and actions; the person picks. Each action carries a plain-language “done when”, so finished isn’t a matter of opinion.
- A ladder, never a percentage — Four named rungs, from getting started to ready to validate. The rung is a qualitative read on the evidence, not a count of check-ins.
- Your guidance, not generated advice — Load your organization’s own approved coaching language and resources, and the coach leads with yours before it personalises anything.
What you getGrowth with a trail behind it: what was worked on, what was tried, what got in the way, and what changed.
Sense
Ask the group instead of the individual. Themes come back; individuals never do.
Modules — the full 360, team listening and engagement surveys are each switched on per organization.
- Team listening rounds — An admin can run one across the organization, a program or a cohort. A manager can run a qualitative round on their own team, entirely in conversation.
- Engagement rounds — Dimension scores, driver analysis and manager-to-upline rollups, riding the same anonymised engine as listening rather than a second one.
- Qualitative 360 — Open-ended questions, no rating scales. Raters answer through a link — no account, ever — and nothing is synthesised until the anonymity floor is met.
What you getAn honest read on how a team or an organization is doing — which informs coaching and never becomes somebody’s competency score.
Decide
When development turns into a decision, the evidence bar has to move. The product moves it rather than trusting anyone to remember.
Modules — analytics is switched on per organization; validation runs as an engagement with your Client Executive, not as a switch.
- The Verified lens — A separate view that narrows to people with independently administered evidence behind them. Self-report alone never appears in it, and seeing it is its own permission.
- A potential model you own — A weighted read across your own approved competencies — seeded from the research on what predicts future capability, then tuned by you rather than hidden from you.
- Local validation, when you need it — Criterion validity and adverse-impact analysis against your own outcomes. Every statistic is computed by a fixed engine and only then narrated, and it refuses to report at all under twenty matched cases.
What you getA defensible answer to “who is ready” — and an equally defensible “we don’t have enough on this person yet.”
Built to survive a works-council review.
There is no admin view of a coaching thread, because there is no query for one.
Two columns. The admin console's only read of the coaching table asks whose session it is and when it was last touched — that is the “last active” column on the people page. The messages themselves are never selected: not redacted, not filtered, never asked for.
A projection built by hand. When someone invites a mentor or their manager to follow their plan, that supporter receives a view assembled field by field — focus areas and the rung each has reached, live actions, the titles and dates of completed ones. Reflections, coaching transcripts, the accumulated coaching profile and every competency score sit outside that list by construction, not held back by a filter someone could later loosen.
A test that tries to break it. It signs in as a genuine supporter holding a live, active link and reads the private tables directly — plans, coaching sessions, the profile. Each returns zero rows. It runs against a real database rather than a mock, and it is checked in beside the policies it guards, to be re-run whenever they change.
It is also why coaching can be candid. Nothing said in a session quietly becomes a score — when Sphere thinks a demonstrated strength deserves to count as evidence, it says so out loud and offers a route you are free to decline.
One module, not each feature, decides whether a person's read leaves.
One module, every exit. Whether an individual's composite read is visible to HR at all, whether that person is counted into an aggregate, whether a reminder email may be sent — all three are answered from the same stored policy in the same place, so they cannot quietly drift apart. The visibility dial itself runs all the way down to nobody: not the HR team, and not the individual either.
Silence is never a yes. Under an opt-out default, taking part is the default and withdrawing is a single switch. Where a jurisdiction is set to require affirmative opt-in, an unanswered consent counts as no — the person still gets the nudge in their in-app inbox, but no email is sent and their read stays out of the org's view.
A list you can add to, not subtract from. Quebec, the EU, the EEA and the UK sit on the opt-in list by default and are merged back in every time the policy is read, so an org that saves a narrower list does not quietly lose them. Every change to these settings is written to an audit log.
Person-to-person sharing works the same way. Inviting a mentor or a manager to see your plan is itself the act of consent — one named person at a time, up to five, pausable without ending it and revocable outright.
A suppressed group leaves the totals as well as the table.
Three is the floor, not the setting. An org can require a larger cell; anything below three is clamped back up on the way in. A client can be more careful than the default — it cannot be less.
Suppressed means out of the arithmetic. A group under the floor is dropped from the breakdown, its headcount never reaches the browser, and its members are removed from the “overall” figure too. Otherwise subtracting the visible groups from the total would hand back exactly the people you just hid.
Enforced at the source. Suppression happens on the server, before the numbers are serialised. The withheld figure is not in the payload, so there is nothing in the page to un-hide. The same floor gates team listening: below it, a round declines to produce results rather than producing thin ones.
And team listening is deliberately not a scoring input. Group sensing tells a leader how their team is doing; it never becomes any individual's score.
Book a 25-minute walkthrough.
See the working product: a measured profile with its evidence, the governed loop, and the trust architecture — with your questions, not a canned deck. No slideware, no self-serve trial.