Commentary

Four Easy Ways To Measure Performance Online

I'd like to begin with a bold statement: E-marketers in today's online world are trapped in a box. They are still conceptually tied to “impressions” and “views,” ignoring basic measurements of online campaign success.

This brings to mind the famous legend of the Gordian Knot:

When the Phrygians were without a king, an oracle decreed that the next man to enter the city driving an ox-cart would become their ruler. A peasant farmer named Gordias was that man. After entering the city, he tied his ox-cart to the temple post in a perfect knot.

The cart remained tied to the post for centuries. There was not a single person who was able to untie it. Another prophecy was then made stating that the man who untied the knot wouldl become the conqueror of Asia.

Around 333 BC, after numerous people tried their luck to no avail, Alexander the Great made his way to the city. He stood in front of the knot and examined it from all directions. He then drew his sword, slicing the knot in half, in a sharp, rapid stroke.

And the rest is history…

This story draws parallels to the state of today's display space. E-marketers are currently trapped inside a post-view box, instead of measuring performance the way it has always been and should be measured.

This post-view approach became prevalent as providers started selling highly complex, view-through-based models. They convinced us that clicks, and more importantly, conversions generated by clicks were no longer of value. They offered their own angle to handle this “knot.”

Some providers convince marketers to run A/B tests to validate view-through, but actually use it as a false pretext to charge 95% PV. They cite “uplift,” “lift” and “engagement” to demonstrate value. Ignoring the most basic measurement of success, marketers end up running pure post-view display campaigns.

Anyone who has run an A/B test knows that post-view provides value in display campaigns. But it should be regarded as an added value and not the only or even most important measurement of a successful campaign.

The time has come for marketers to draw their sword of reason and slice the knot by returning to measuring performance according to three basic metrics: CTR, CR (conversion rate) and ROI. Unlike offline media channels, the online world offers us the luxury of having these accurate and telling metrics.

Here are four methods performance metrics:

1)     eCPM – You’ll always pay a higher cost per 1K impressions for a retargeting campaign vs. other channels because of its degree of precision and optimization. However, don’t pay for an over-priced CPM.

2)     eCPA - Measure eCPA by post-click sales only  to get a true CPA model, which is the essence of performance in display advertising.

3)     eCPC – Measure clicks only in relation to actual conversions they generated. Remember, although important, a click is only the first step toward the goal. Marketers should focus on conversions rather than clicks. After all, you can't go to the bank with clicks, can you?

4)     ROI - For each $1 spent, measure how much you made from post-click conversions only.

While I am not advocating against post-view conversions or running campaigns on a straight CPM basis, I am stressing the point that advertisers should adopt models that adhere to basic measurements of success.

When comparing providers, no matter which business model you choose to adopt, remember the most important goal of a successful online display campaign: driving post-click conversions.

Reconnect with these performance metrics and the rest will be history -- not only for Alexander the Great, but for you, too.

6 comments about "Four Easy Ways To Measure Performance Online".
Check to receive email when comments are posted.
  1. Craig Mcdaniel from Sweepstakes Today LLC, August 23, 2012 at 1:50 p.m.

    You are short one "performance metrics". This is "Billboard". Billboard is charging a monthly/weekly/or daily fee for the ad(s). I went back to this method because it was simple to bill, required no extra programming and most of the success of the ad went back onto sweepstakes sponsor or ad agency. If the sweepstakes is by a well-known named sponsor and has great prizes, it will do very well. For Sweepstakestoday.com this solves the myth that “people just enter for the prize” which is not true. This method also eliminates “click fraud”. Further, we “pre-qualify” the ad by giving our members a “description page” about the sweep promotion.

    While it is forgotten in the modern digital age, it works extremely well for what we publish. My hope is that the ad agency, marketing companies and ad networks start to understand that “CP’s” methods doesn’t fit all publisher’s websites. This old and proven advertising pricing method still works well in the new digital age.

    It works for us. Why not consider this method again?

  2. Alastair Gibb from Wilkinson & Sonoma, August 24, 2012 at 12:56 p.m.

    As long as the PCCR delivers well, PVCR should be allowed to be counted too. We have been running retargeting for almost 8 months now on six of our portfolio sites. PCCRs have been good on all of them. But we learned that when we started to take into consideration Post View (i.e. we started paying for PV) our retargeting provider put the additional margin into buying more media for us, and that in turn brought more conversions. I agree that PV must be taken into account with extreme care, and it can be a nice addition to the campaign.

  3. Chad Porter from myThings, August 24, 2012 at 1:48 p.m.

    Alastair, I agree that consideration should be given to PV - and you wont' find a firm that won't accept financial consideration, but advertisers seem to want to start with less risk and get comfortable with new campaigns under a PC arrangement. From there, you can partner and agree on how to address scale/increment - if you need to. (ie higher PC price, small PV payment, short PV window, lock in a CPM, or even Cost +). Lots of ways to grow a partnership, the trick is, starting one out on a the right foot.

  4. Sheldon Gilbert from Proclivity Media , August 24, 2012 at 2:18 p.m.

    Interesting article, Yishay!

    You definitely raise a valid concern about "attribution model over fitting" by essentially transposing the broadcast TV and radio ad measurement model to online, which is designed to measure to the fullest extent. Though, your analysis on the positing of clicking and CTR as key metrics is incorrect.

    Unlike search, where the customer is actively looking for a product that he or she has a high likelihood of buying and thus clicks through at high rates, in the case of display, there is often an anti-correlation between CTR and order conversion. In fact, the in-market customer, with a high likelihood to buy a product and knowledge of where to buy it, rarely clicks on the ad but still makes the purchase.

    Further, people who click have a lower likelihood of buying at times, and clicks and CTR are often false proxies of purchase intent outside the realm of search.

    For these reasons, the combination of click-through and view-through revenue must both be considered. Also, the use of A/B tests can and should be applied to both modalities to establish the coefficients of marginal utility to apply for temporal attribution models, especially for view-through attribution, whether 48 hours or 15 days. Let's not forget that applying the statistical models to determine the appropriate attribution coefficient should also be inclusive of many other factors that inform the purchasing/defining timeframe – such as seasonality, types of products (leather couch vs. toothbrush), shopping channel/device (iPad vs. laptop), country, as well as many others. These should all inform the coefficient of both view-through and click-through attribution.

  5. Dave Hendricks from LiveIntent, August 25, 2012 at 11 a.m.

    Much has been said by other commenters that doesn't need to be repeated.

    But given that prior to 1994 or so, advertisers and marketers existed almost exclusively within a PV universe. Lo and behold, business grew year over year. And yeah, the marketers saw that business grew, and the lord was pleased.

    Yes it is more complicated to measure and implement PCCRs and PVCs. And yes it is more difficult to determine attribution credit percentages. And yes, some providers will game this.

    However, the challenge of managing this doesn't mean that it shouldn't be done. Technology providers, agencies, rep firms, ad buyers, etc. all deserve as much credit as possible when a campaign succeeds 'overall'.

    Don't make CTR the only metric.

  6. Yishay Shachar from myThings, August 26, 2012 at 8:28 a.m.

    Thank you all for your replies.
    I just wanted to be clear that I am not suggesting that post view is not a true measure or that it needs to be ignored.
    I am simply raising the fact that many marketers forget the basic KPIs when they look at their campaigns.
    With all the complicated structure you should take a step back and look at the basic KPI (ECPM, ECPC and ROI).

    @ Sheldon Gilbert and Dave Hendricks, by no means I meant that the CTR and clicks are the only key metrics; I said you must not forget these basics. It is still important to look at the basics, which we tend to forget, as the world of marketing becomes more and more complex.
    The Key is to look at the basic KPIs still, and not assume they are less relevant today.
    If you look at your ECPM and see that you are paying $8 and your ROI is good, then this is great (at least you know what your ECPM is).

    @Alastair, thanks for your feedback. I think you got my point.

Next story loading loading..