I was recently reminded at a speaking engagement in Slovenia that success is a dangerous thing in many ways.

Someone in the audience observed that often campaigns that were performing successfully in business terms were rewarded by a cut in the budget (rather than an increase, as I had been advocating) and I started to wonder how common this was.

Certainly when you start searching through the global case studies of effectiveness in the Warc archive you keep coming upon examples of this.

Here's a revealing quote from the case study for Lego Bionicle globally: "The media spending budget for 2002 was reduced by approximately 25 % (with twice as many products) in the aim to capitalise on the 2001 investments."

And another one from the Steinlager case study in New Zealand: "Note, in year 2 we had approximately 1/3 of the budget that we had in our launch year."

Even allowing for the unsustainable level of commitment often given to product launches, this is a savage reduction following such a successful launch. And something similar happened in China after the hugely successful launch of the Motorola Razr "The brand econometric model suggested that cutting ATL spend in the first half of 2007 (from US$30 million to US$7 million) may have lost some four million handset sales, worth more than a billion dollars."

And there are many more examples of share of voice being cut after a period of outstanding business success: ING (UK), Corona Extra (USA), Familiprix (Canada), to name but a few. There is a suggestion of this being more generally at work amongst the collected findings of the IPA databank.

It turns out that the campaigns with lower levels of excess share of voice (i.e. share of voice minus share of market – a measure of the relative level of investment or disinvestment in the brand) were campaigns that worked more efficiently. And this can't be entirely explained by the size of the brand: although these are on average slightly larger brands and therefore to be expected to deliver slightly greater efficiency, the difference is marked.

So perhaps it is widespread business behaviour to reward success with disinvestment. A clue to what might be driving this shows its face in the case study for UK grocery retailer Waitrose. The case study reports how 'effective share of voice' rose from 6 to 11% between 2002 and 2007 whilst actual share of voice fell from around 5% to 4%.

'Effective share of voice' is an increasingly common measure that I believe should be banned immediately on brand health grounds. For those not familiar with it, effective share of voice works like this: imagine my true SOV is 5% and my campaign is twice as effective in business terms as the average campaign in the category, then my effective share of voice is 10%. 'Fantastic' thinks the finance department 'we can halve our budget and still be on target'.

This is of course, the logic of the asylum. No agency can guarantee to deliver exceptional effectiveness and no-one can know how long such an asset will last – it is a hugely valuable and rare asset that should be sweated for all its worth while it still delivers.

The sensible response to having a super-effective campaign is to increase investment behind it: beg, borrow or steal budget from wherever you can, but make the most of the brief window of opportunity for increased return on investment. That is of course unless you are worried about being able to supply the demand you have created.

But that is the kind of dangerous success that most brands can only dream about.