The locations of young people who are most at risk of having poor outcomes later in life have been revealed - as have the "stark" gaps in support for them.
The proportion of young people who are most at risk has been mapped by region, with analysts able to drill down further.
In Auckland, more than 10,350 teenagers aged 15-19 are in at least one of five "target populations" - such as those with health or disability issues and Child, Youth and Family history.
Those factors can stack the odds against them. The most at-risk individuals are much more likely to drop out of school, go on a benefit or end up in jail.
Some areas have a bigger problem than others. In Papakura, the percentage of at-risk teenagers is 21 per cent, compared to 13 per cent in Manukau and 9 per cent in the rest of Auckland.
Such mapping is part of the Government's "investment approach" to social spending, which aims to identify where up-front spending can cut costs later.
Deputy Prime Minister Bill English said "stark" shortcomings had been revealed, including in the health system which "has a peripheral and occasional interaction with our most challenging people".
"Despite all the rhetoric about early intervention, the money doesn't start kicking in until the kid is about 3 ... we don't do much with vulnerable pregnant women, other than have a midwife there. That's it.
"Until last week, we wouldn't know that. And now it is stark - that actually we don't do what we think we are doing."
Mr English, who made his comments at a data workshop at Parliament yesterday, said the Government did not have the answers yet on how to find all at-risk individuals, but expected an upcoming review of the Privacy Act could help.
A solution could involve "testing notions of consent" - asking people how much information they are willing to share and have shared.
More cost-benefit analysis to allocate funding has worried unions, including the Public Service Association, which has said the "lottery" of contestable funding causes uncertainty. But Mr English said smarter, more targeted support was needed, rather than universal or core services.
"When you have someone who is 21 times more likely to have a bad outcome, the universal service does not work.
"We are hoping they will show up to the universal service that we like providing. And I can tell you, those are the services that chew the money."
The Integrated Data Infrastructure (IDI) database uses anonymised information from Government agencies including the Ministry of Social Development, Ministry of Health and Ministry of Education.