The Business Case For Cost-Per-Click

jean-baptiste-rudelle“Data-Driven Thinking” is written by members of the media community and contains fresh ideas on the digital revolution in media.

Today’s column is written by Jean-Baptiste Rudelle, CEO and co-founder at Criteo.

Digital advertising moves quickly — sometimes too quickly.

The market and technology has evolved significantly over the past five years to drive the best possible price for digital ad inventory for publishers and maximize impressions for the brands they serve. But, of course, there are growing pains when you’re working in a fast-paced, rapidly evolving market like display advertising.

But there’s a solution that’s picking up speed that can solve some of the biggest challenges advertisers are tackling with display today. These challenges can be broken down into three buckets:

1. Am I really getting what I am paying for?

The quality of ad inventory relies on four key parameters: the audience (who’s looking at the ad), the placement (where the ad is located on the page), the context (what’s surrounding the ad) and the format (what’s the shape and size of the ad).

In theory, inventory is supposed to be priced according to its intrinsic quality. However, the market is far from efficient, and for the same price, quality can vary quite a bit. The issue is advertisers’ ability to verify audience, placement and context of the ad (it’s pretty straightforward to verify the format).

Amazingly, users never see nearly 30% of all paid display impressions, particularly when they are below the fold. For a client, it’s almost impossible to ensure they will pay the right price for each impression. All too often they end up paying an average of all impressions, which is highly inefficient.

A common tactic to guarantee a certain level of quality is to restrict the buy to easy-to-verify premium inventory, like high-end websites. However, as competition for this well-controlled inventory is very high, there is a huge premium in price per impression. In the end, the choice is between high uncertainty or premium price. Sound familiar? Not very appealing.

2. What is the right level of capping per user?

Everyone knows that beyond a certain number of impressions, the impact of advertising diminishes significantly, and could even become negative at some point. Too much advertising kills advertising.

So how do advertisers set the right limit? In practice, hard capping is set at an arbitrary number per day or in total. This creates inefficiencies, as some users receive too many impressions while others are underexposed.

3. What’s the value for each impression and how do we measure the ROI of the campaign overall?

This is a direct consequence of the lack of visibility into actual quality. Some impressions might have a lot of impact, while others are completely useless, especially those below the fold that we’re still charged for. Unfortunately, tracking software has no way to know that the ad is positioned below the fold and it automatically attributes the same weight to all impressions regardless of their actual quality. Amazingly, there are still several bad apples in our industry that push this dirty cookie-stuffing model — blasting the Web with super cheap, almost invisible impressions and taking credit.

In this context, how would an advertiser know if this particular interaction has done anything for their campaign? It’s almost impossible to definitively say if it’s had an impact or not. And this creates a very serious limitation on how brands can measure ROI of an online display campaign, to say nothing of the serious fraud issues.

A Different Approach

The good news is there’s a solution to these challenges. But it requires a change in approach, a change in how the industry defines value for display advertising, and a change in what’s bought and sold by advertisers and publishers. Impressions are great, but they’re a bit ambiguous and it’s hard to really measure the value, which is why cost-per-click (CPC) is gaining more momentum in the market for display. Advertisers want to know what they’re getting for their investment, and clicks mean more than impressions.

So, how does buying on CPC solve the quality issue for the advertiser?

It automatically adjusts the price of each impression, depending on the actual quality.  It’s easy to understand that highly visible banners shown to the right audience at the right time have a much higher click-through rate than those randomly buried down in the site. And interestingly, it does not mean that an impression without any click has no value. CPC is just a powerful mechanism to guarantee that each impression will be paid at the right price. In other words, by paying per click, the advertiser has the guarantee that they will never overpay for any impression, which is important.

And similarly, the CPC model also ensures automatically optimal capping per user.  For each new impression shown to a user, the likelihood to engage tends to decline. With this fast-diminishing click through rate, the advertiser is in a much better position to put the brakes on the ad as needed. This will automatically cap the number of impressions per user at the most optimal level – and ensure again that advertisers are not overinvesting in wasteful impressions.

The True ROI

Last, but not least, CPC pricing makes it easier for brands to understand the true ROI of their campaign. Multiple large-scale studies have shown repeatedly that clickers are much more likely than non-clickers to engage later with brands and ultimately buy. Getting a user who is willing to interrupt their browsing to click on a banner ad is a rare event but it is an incredibly strong signal. Again, this doesn’t mean that impressions without clicks don’t have any value. Clicks simply have much more value.

The market is evolving and more advertisers are making the transition to a pay-for-performance style of media buying, which is why the CPC model is picking up speed.

The shift in defining ad value won’t happen overnight, but it’s happening fast and we can expect to see the demand for clearer forms of ROI and results across the board — beyond just display — because technology is advancing and advertisers want more tangible results for their money.

Follow Jean-Baptiste Rudelle (@jbrudelle), Criteo (@criteo) and AdExchanger (@adexchanger) on Twitter.

Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!


  1. Marc Rossen

    What you have failed to mention is the only place a CPC performance metric might have some merit is within re-marketing display. In other words, after a consumer has already raised their hand and shown brand interest.

    Our industry needs to focus on metrics for mid to upper funnel optimization. This is where the growth is and where optimization metrics area weakest.

    If you have studies showing how CPC proves to be a good optimization metric for mid to upper funnel media, I would love to see them.

  2. Alejandro Correa

    I really liked all your points except for the main one. If marketers pay for clicks, they are essentially paying for outcomes; although it sounds great, it’s not because it gives suppliers of inventory all the wrong incentives. Imagine if instead of paying for coffee by the pound one paid for coffee by the “buzz”; coffee suppliers would be incentivized to add all sorts of stimulants to inflate the price of otherwise cheap product. Same goes for advertising impressions (another commodity); if site revenue was based not on content, but on how hard it is for the user to avoid clicking on a banner, you’ll start seeing either a lot of horrible sites or more click bots. Either way, the marketer loses.

  3. Does CPC solve for “Am I really getting what I am paying for”? It’s no different then CPM. With CPC you get and pay for clicks. For CPM you get and pay for impressions. You are getting what you paid for.

    It also doesn’t “solve for the right level of capping”, given different frequency will return different valued clicks. It also doesn’t “solve the problem of the value of an impression” (there is a value), it ignores it.

    CPC does not, as you state “automatically adjusts the price of each impression, depending on the actual quality”, it automatically adjusts the price of each impression to correlate with the click-thru rate and cost per click, even if those clicks are worthless.

    CPC is simply a metric used to defined when payment is due. In some cases it is easier to work with then CPM. In others it is not. It may or may not be more correlated with your ultimate success metric.

  4. I can’t agree CPC is the way forward. A key reason is articulated by Alejandro in his comment. Another reason is the CPC model allows Criteo to arbitrage the low cost CPM on which it buys, into a high-priced CPC which it sells on to advertisers. The resulting high margins are at the cost of the advertisers. This article is a defence of a business model rather than a contribution to the debate on the value of digital display. Digital media is now more or less universally accessible so why not separate the cost of buying services from the media rather than bundling them in a non-transparent CPC?

  5. The fact is that 8% of people contribute 85% of the clicks out there. These people almost always skew, very young or very old, with low income and low education. Are these the kind of prospects most brands want to be targeting? There have been multiple case studies that have echoed this fact, here is just one done by comscore.

    Unfortunately, optimizing to clicks is about the worst tactic you could possibly do, as you are typically targeting people who can’t afford to purchase most of the products you are selling.
    Viewthroughs work (as shown by hundreds of thousands of AB tests) and are much more effective at driving people to convert.
    By having a CPC model, you will be targeting low value prospects, or all your partners will focus all of their efforts on retargeting. When an advertiser decides to focus on retargeting rather than prospecting, it stops filling the funnel with new prospects and your net new customers will actually decrease.
    Getting new customers is the principal driver of growth for a company, so I would say the business case for the CPC model is inherently flawed.