I talked recently about when partnerships do and don't work at an interesting gathering of dozens of corporates and central government. Nesta is immersed in some very constructive partnerships - with firms like Santander, B&Q and BT over Innovation in Giving; with O2, Café Nero, Facebook and Microsoft on Makethingsdostuff; in research projects with Google, Pearson and others.

I've also seen partnerships work very well in the Studio Schools, which now involve hundreds of firms, from local taxi firms to multinationals involved in space industry.

But I'm not keen on partnership for its own sake, and have seen far too many which are superficial, vague, ineffective and frankly a waste of people's valuable time. Fortunately it's pretty clear what does and doesn't work. So here are a few deliberately tendentious views which draw on discussions with people who have spent years trying to make partnerships succeed, but are by no means scientific.

Try for depth

If you want to achieve anything significant you have to be willing to devote time and resources. What doesn't work is getting a bunch of people to brainstorm for an hour or two on how to solve poverty, homelessness or, for that matter, the UK's future economic growth prospects, though it's amazing how often this still happens. Deep-seated problems demand depth of engagement. They're likely to be complex and to involve many interconnected factors. Anyone who thinks that a simple model, or an app, is going to solve a problem, needs to grow up a bit.

Start with humility and mutual respect

Good things only happen when all the partners start off with humility, and a willingness to learn from what's already been done and tested. The business world shouldn't overestimate its ability to come up with smart solutions to social problems just as government shouldn't overestimate its ability to solve business challenges.

Humility doesn't mean that you have to be trapped by conventional thinking. But you need to understand it before you move onto radical alternatives. Bill Gates is an interesting example in this respect. When he first became involved in charity he gave the impression of thinking that simple business models would quickly solve problems like malaria or poor schooling in the US. To his credit he devoted a huge amount of time to gaining a deeper understanding of these fields. Many of his early projects failed - in some cases on quite a large scale. But over time his foundation has steadily become more effective and more respected.

Move towards precision and value propositions

Partnerships don't work when they're vague and rhetorical - but do when they focus on the precise value they can provide, for example when a corporate can offer skills in finance, logistics or marketing to a charity with deep knowledge of a social field. Each side in the partnership needs to understand what matters for the other - they're bound to have different interests, and different ways of thinking that reflect these interests. It helps a lot when areas of common ground, and precise value propositions, are reached quickly. If after two or three meetings this is not even dimly visible then it's probably worth trying to find another partner.

Grow real relationships and mobilise the power of small numbers

Partnerships all boil down to people. In practice partnerships aren't really between organisations - they're always a partnership between a few, a few dozen, or a few hundred individuals who get to know each other, get to respect each other, and feel energized by each other. These relationships thrive best when they're reinforced by shared experience - the more frontline the better. Doing things together is better than just talking about things together, shared risk and vulnerability is better still. And relationships are helped by compelling conversations in small groups. Big conferences don't create partnerships; even workshops with more than around a dozen people don't. Big events are better for conference organisers - but not much else.

These four comments may seem like statements of the blindingly obvious. I only share them because much, and perhaps a majority, of what I see labeled 'partnership' ignores all of them. There are any number of events which purport to share ideas across sectors and to build partnerships - but many are too shallow, too arrogant, too vague, and too big - and as a result achieve very little despite lots of talent and goodwill.

I may be entirely wrong in my diagnosis and would welcome other views. But over the last few weeks I've seen both the best and the worst. The best examples have been ones where a business has built up deep engagement in a field - like Telefonica on incubators or O2 providing opportunities for young people, Timpsons on opportunities for ex-offenders or B&Q on the environment. My worst recent experiences included sitting with a group of entrepreneurs trying to dream up solutions to old age isolation - without any knowledge or experience, prompted by overheated rhetoric about using business and entrepreneurship to 'solve' social problems. They were well-intentioned, and they're all people who are very good at what they do in their day jobs. But the group reminded me of sitting in on a group of Marxist academics about 20 years ago who were trying to devise new ways to plan the economy. In both cases there was a weird combination of ignorance, overconfidence and blithe lack of curiosity about how the things they were talking about actually worked.

Partnership matters. We need more creative blending of the skills of different sectors. As public spending is even further squeezed there's both a need to find different ways of delivering public services, and an inherent virtue in exploring new models with lateral thinking. But most important things in life depend on humility and a willingness to learn. This is no different.