Pay-per-click keyword prices have been falling in many sectors for the past year or so, according to the agencies and client-side marketers whom we regularly speak with in the UK, and as evidenced by the great US-focused research published by Fathom. 

In 2005, my gut reaction to falling keyword prices was that it was merely a correction. After all, growth in spending on paid-search surged massively between 2004 and 2005, driven by much competition and new entrants to the PPC scene.  

New advertisers – and some existing ones – may have over-egged the cake to secure a number one listing, before revising keyword bids downwards once the resulting sales didn’t provide enough (or any) return on investment. Paid-search is seen as one of the lowest hanging fruits for those new to online marketing, and PPC virgins may simply have been a bit too gung-ho to begin with.  

But maybe keyword deflation suggests something more sinister, something that isn’t purely about new entrants bidding too much to attract visitors. Falling ROI may indicate a rising number of fraudulent clicks.

Is click fraud is in fact a bigger issue than the paid-search networks are prepared to let on? And is keyword deflation the smoking gun?

Click fraud defined

Click fraud comes in various shapes and sizes, but we might as well boil it down to sinister folk clicking on ads to drain competitors’ ad budgets.

Many clicks are automated, making the situation far worse than if it were simply restricted to malicious human action. Or maybe your tiny blogger with Adsense is clicking on the ads a few times per day, to earn a few dollars a day. Multiply that by many thousands of tiny bloggers, and you have a problem. Either way, these clicks need to be paid for by advertisers but will not generate any sales.

Depending on who you speak with, click fraud is either a minimal problem or a massive one. Google claims to be on top of it, and has said that fraudulent clicks are typically identified and dealt with before the client even knows about them. And Google is talking about low-single figures, not a double-digit problem.

Yet some clients – and agencies – are far more pessimistic. They see it as a big issue, and one that might damage the credibility of paid-search advertising. In some sectors, fake clicks are estimated to surpass 20%. It is said to be a particular problem in high-value sectors, such as financial services, or travel, where just one click can cost the client more than £10 on competitive keyphrases.

Some newcomers to paid-search have expressed concerns about transparency: they essentially want Google Adwords to provide more information on the clickers. In particular, complaints are coming from the SME sector, which Google is targeting through local search and pay-per-call offerings. So will the PPC networks help to prove or disprove click fraud by increasing transparency? Presumably they will have to.

Ok, let’s consider two possible scenarios…

Scenario #1: Keyword deflation is a correction

Over-enthusiastic advertisers spent way too much on keywords. The resulting sales didn’t provide a decent return on investment. But it took these advertisers a while to figure this out. Maybe some of these advertisers weren’t properly tracking post-click purchase activity when they launched their PPC campaigns. And only recently have they been able to accurate track PPC-to-checkout activity.

So it could be a correction, driven by an improved understanding of PPC and of web analytics. This is about marketers gaining experience of paid-search, minimising schoolboy errors, and generally becoming savvier about internet marketing.

But hold on. This doesn’t entirely stack up. Why? Because savvier marketers will not stop at amending their keyword bids. They won’t only improve visibility by implementing a solid web analytics platform. They will do many other things too.

For example, advertisers will gradually hone their campaigns. They will amend ad text, to generate higher click-through rates (something that is particularly important on Google Adwords, since CTR helps determine ad positions).

They will also tweak their websites, hopefully creating and improve landing pages to drive more visitors towards the checkout. And they may well finesse the entire purchase journey, to increase conversion rates.

Given all these changes, is it possible that ROI from Adwords will be in freefall? You’d think not. And if keyword costs are falling, then that too should improve ROI, since visitors can be lured for less outlay. Advertisers will attract more visitors, for the same budget.

It doesn’t make a lot of sense, does it?

Scenario #2: Keyword deflation is a reaction to click fraud

So here’s the theory. Keyword costs are falling as a result of declining sales from PPC-referred visitors. That is a correction of sorts, but what are the underlying reasons for the fall in sales? And how is it that conversion rates are falling when marketers are improving their websites? The primary driver of this decline in ROI could be click fraud.

Those who believe that click fraud is a big issue will say that a hefty proportion of clicks paid for by advertisers are not valid, and as such are not going to provide any returns on investment. And the ROI problem is exacerbated as click fraud increases.

Advertisers affected by click fraud will be forced into reducing the amount they are prepared to spend per click to attract visitors. Even when they don’t know that they are a victim of click fraud. Paying the same amount per click will undoubtedly lead to a decline in ROI.

Or, for those advertisers that have optimised landing pages and purchase journeys, they may be treading water, with little or no improvement in ROI despite making positive changes to their websites.

Is click fraud actually driving down the costs of keywords? Is it a much bigger deal than the networks are letting on? Is the gun smoking, or is it aflame?


Some say click fraud is a simple fact of online marketing life. It will never go away, and to deny that it exists is foolish. It is the online equivalent of shrinkage, which every high street retailer needs to factor in. And as such, the market – the advertisers – will have to adjust their budgets to accommodate it. Each sector will achieve some kind of pay-per-click equilibrium.

But advertisers affected by click fraud want two things to happen. The first is an admission, from the networks, that click fraud exists and is a big problem for certain high-value sectors. The more competitive the sector, the worse the problem is.

The second thing advertisers want to see is greater transparency and better tools to help identify instances of click fraud at network level (rather than by sifting through their in-house web analytics data, long after the dodgy clicks occurred).

With many advertisers still reasonably new to paid-search the problem of click fraud may go unchecked, as least for a while. You need time to properly figure out whether you are paying over the odds for keywords. You need a few months of campaign data to analyse trends to be able to spot inconsistencies. To be able to spot click fraud.

Whether the networks can comprehensively police fraudulent clicks on a massive scale is somewhat doubtful. This means it is up to you. So keep a close eye on post-click activity, and red flag unusual visits. Then inform the networks accordingly. Remember too that some user behaviour might suggest click fraud, when in fact it won’t be anything of the sort. It is undoubtedly a grey area…

And keep an eye on the US, where Google has been forced to pay a settlement to advertisers in lieu of click fraud (something which has not been accepted by all affected advertisers).