Les Binet and Sarah Carter get a little bit angry about some of the nonsense they hear around them… like the idea that advertising is a risky investment.

Last week we played host to advertising guru Paul Feldwick, who came into the agency to discuss his latest book. During questions afterwards, a planner raised his hand. "Failure often teaches us more than success, so looking back, which campaign was your biggest failure?" he asked.

Paul thought for a bit, then replied that it was actually hard to think of any real disasters. Not because he was infallible, but because it's actually quite hard to fail disastrously when it comes to advertising. This struck us as an important thought. Clients often approach advertising with nervous caution. What if the message is wrong? Or our tone of voice isn't right? Or we alienate customers? Or we go against the zeitgeist? Best delay a bit, do a bit more research, get it perfect.

As we've written before though, you have to get it spectacularly wrong to generate significant hostility to your advertising; the worst that usually happens is indifference or mild hostility. And on the rare occasions when brands do provoke a backlash (e.g. the recent 'Beach Body Ready' uproar in the UK), there is always the suspicion that the free publicity generated may actually increase sales. (Understandably, there is little publicly available research on this.)

In nearly 30 years of research, we've never seen a single example of an ad campaign that was shown to have a negative effect on sales. Not one. And it's hard to think of even anecdotal examples. Strand cigarettes? That was 50 years ago, and we've never seen empirical evidence that the ad did depress sales.

One reason why it is so hard to fail is what psychologists call the Mere Exposure Effect. Exposing people to any ad for a brand, even just the brand's logo, will increase their propensity to choose that brand, no matter what it says. Ads tend to work better if people like them, of course. But even advertising that is disliked can be effective, as the Grand-Prix-winning IPA paper for Radion showed many years ago.

So most advertising works, to some extent. In all the econometric models we've ever seen, only one TV campaign failed to have a positive sales effect. And that was a single execution with a low number of ratings behind it. (Other media are often harder to measure, but we suspect that most of them work just as reliably.)

Of course, increasing sales is not quite enough. A decent ad campaign should also generate enough cash to pay for itself and more. Not all manage that, but experienced advertisers can usually avoid making disastrous investments if they follow a few simple rules. Concentrate your firepower on markets and brands with decent profit margins (the single biggest factor determining RoI). Choose media that combine low cost per exposure with high emotional power and/or good response rates. Allocate most of your budget to stuff you know delivers a decent RoI, and a bit to testing and learning from new stuff. Piece of cake!

But many clients don't see it like that. They test and tweak in artificial conditions, over-intellectualising and procrastinating. We know of one company that spent the last four years getting their next advertising 'just right', while sales continued to gently decline. They were lucky to get away with it. An analysis we once did revealed that two years off air can often be enough to kill a brand, especially if competitors grab the opportunity to ramp up spend and maximise share of voice. There is a crowded graveyard of once proud, high-share brands that died from this scenario.

So, yes, advertising is unpredictable. Yes, it's never possible to know in advance exactly how the public will react to our communication until it's out there. And in that sense, it's always risky. But a far bigger risk comes from things that we can control – and doing nothing is usually the biggest risk of all.