01 THE EVIDENCE
Why we need to create Team IQ
Every organisation we are asked to look at has already tried to fix delivery. They bought a process, they trained the managers, or they hired better people. None of those three levers explains why capable teams still deliver unpredictable results. This page is the argument — and the published research behind it, so you can check it rather than take our word for it.
02 WHAT HAS ALREADY BEEN TRIED
Three levers. Each one fails for a different reason.
Lever one
The process
Agile is genuinely more effective than waterfall on complex work. Yet an InnerCircle analysis puts failure at 32% of Agile programmes, costing around £37 billion in the UK alone, and Scrum Inc reports that 47% of Agile transformations do not deliver what they promised. Nearly half.
The process is a map. A map is not the terrain, and it cannot update itself when the terrain changes.
Lever two
The manager
You can put every manager through leadership training. You can add pressure, dashboards and one-to-ones. What you are really doing is making the manager the load-bearing wall of the operating model.
Then they take a holiday, move on, or burn out — and performance returns to whatever the team's default was. A leadership dependency is a single point of failure you have chosen on purpose.
Lever three
The individual
The obvious answer is to hire better people. Except that most organisations cannot fire and re-recruit a whole team — the time, the money and the legal latitude are not there, and "the A-team" is not sitting on a bench waiting.
More importantly: the research says it would not fix the problem. See below.
03 THE EVIDENCE
Four findings that decide where you should spend your effort.
1 · Individual intelligence does not predict team performance
MIT ran studies testing the correlation between the individual intelligence of team members and how well the team performed. Neither the average member's intelligence, at 0.15, nor the strongest member's, at 0.19, was a significant predictor once the group-level factor was included. The group factor itself predicted performance on complex tasks at 0.52.1
The studies found no significant correlation between individual intelligence and how the team performed. What they found instead was a general collective-intelligence factor: a property of the team, not of the people in it.
So the answer to a team problem is not "recruit a genius". Two heads genuinely are better than one — but only if the team is capable of using them.
2 · Google spent four years looking for the same thing
Google's team-effectiveness research ran from 2011 to 2015, starting from exactly the assumption most leaders hold: that adding exceptional individuals must raise the team's output. It does not. What separates effective teams is how they work together — safety, dependability, clarity and meaning.2
Those are trainable, measurable team properties. They are not personality traits.
3 · The spread that matters is between teams, not between individuals
In the dataset behind a 2025 study of the video game industry, revenue generated per team member ranged from 1.33 to 15,158. The authors log the measure, because the distribution is that skewed.3
Those are money figures for team output, not test scores, so they are not comparable to any range of individual measurements — and we are not going to pretend otherwise. What the finding does establish is where the variance sits: across teams doing comparable work, drawing on the same talent pool, in the same industry.
The lever is the team — and that spread is visible in your own teams before you spend anything.
4 · Almost half of transformation spend does not land
Third-party analyses of Agile transformation outcomes place failure somewhere between 32% and 47%.45 When nearly half of a large, multi-year investment does not produce the promised result, the problem is not the framework and it is not the funding.
It is the capability of the teams who have to operate it when reality does not match the plan.
What this evidence does not settle
The collective-intelligence finding has been replicated and challenged. Here is the state of it.
A 2021 meta-analysis in PNAS covering 22 studies, 5,279 people in 1,356 groups found strong support for a general collective-intelligence factor, and found the group's collaboration process mattered more than the average skill of its members. It has also been challenged: a 2017 study in Intelligence found individual IQ accounted for around 80% of the variance in group performance, and a 2024 study in PLOS ONE failed to replicate the three predictors.
So what is settled? That some teams reliably outperform what their members' averages would predict. What is still argued about is the precise measurement. Our claim is the narrower, testable one: inspect the conditions, measure them, and see whether the number moves.
04 WHY NOW
The work got more complex. The process did not get smarter.
Three things changed at once, and each one pushes the burden onto the team rather than the individual:
- Complexity. Modern delivery problems are too complex for one person to hold in their head, however senior. You need a team, and you need it to think.
- Speed. The window between a decision and its consequence keeps shrinking. Teams that cannot decide quickly now pay for it visibly.
- Change inside the frameworks themselves. SAFe released in 2011 and is now on version 6, with more than twenty iterations between. A framework that takes eighteen months to roll out is out of date before it finishes. Processes are a snapshot; reality is not.
The German field marshal Moltke the Elder made the same point about battle plans: "No plan of operations extends with certainty beyond the first encounter with the enemy's main strength."
In complex, unpredictable environments you cannot categorise your way to an answer. You have to sense what is actually happening, then respond. That is the difference between a categorisation model and a sense-making model — and it is the basis of the Cynefin framework.6
If the plan cannot survive contact with reality, the only asset that can is a team capable of adapting together. That is what TEAM IQ measures.
What we are claiming
Collective intelligence is measurable, and then it is trainable.
TEAM IQ Creator takes the property the research identified and gives it a score across six pillars, a set of named measurement tools, and a re-measurement that proves whether the change worked.
If the number does not move, we have not finished.
05 WHAT WE DO NOT CLAIM
Three things you will not read on this site.
No invented numbers
Every statistic here is published and linked. Where a figure comes from our own delivery work rather than a publication, it is labelled as a first-hand observation with the organisation unnamed, and never presented as research. Where we could not source a claim at all, we removed it rather than repeat it.
No client testimonials we cannot stand behind
We have no published client quotes yet, and we are not going to write them ourselves. When an organisation agrees in writing to be named, their words will appear on this site unedited. Until then there is nothing to show, and that is deliberate.
No promise that this replaces your framework
Keep SAFe, keep your PMO, keep your dashboards. TEAM IQ raises the capability of the teams operating them, which is the part that determines whether they work.
06 SOURCES
Check the sources yourself
- Woolley, Chabris, Pentland, Hashmi & Malone, Science 330(6004), 686–688, 2010 — "Evidence for a Collective Intelligence Factor in the Performance of Human Groups". Average member intelligence correlated with the group factor at r = 0.15 (P = 0.04) and maximum member intelligence at r = 0.19 (P = 0.008); neither was significant in the regression once the group factor was included (b = 0.05 and b = 0.12, ns). The group factor predicted criterion-task performance at r = 0.52.
- Google — team effectiveness research (Project Aristotle), 2011–2015: re:Work, "Understand team effectiveness".
- Szatmari, Deichmann & van den Ende, Strategic Organization 23(4), 604–629, 2025 — journals.sagepub.com. Team performance is measured as revenue generated per team member (logged): minimum 1.33, maximum 15,158.12.
- InnerCircle — Agile failure rate and UK cost: "UK wasting £37b a year on failed Agile IT projects" — 32% failure rate.
- Scrum Inc. — transformation outcomes: 47% of Agile transformations fail.
- Dave Snowden — Cynefin, and the distinction between categorisation and sense-making: "A Leader's Framework for Decision Making", Harvard Business Review, 2007.
Statistics are quoted as published. Where a figure is contested or comes from a single source, we say so on the page that uses it.
07 NEXT STEP
Find out where your teams actually stand.
The scorecard takes eight minutes and gives you a score out of 100 with your weakest pillar. If you need to make the case internally first, the board case page does that for you.
For the champion
If you are the person who has to convince everyone else, start here:
- The one-page business case — printable, no email needed
- The cost-of-inaction model — put your own numbers in
- Full pricing in pounds — with VAT stated