Image credit: patrimonio / 123RF Stock PhotoA while back I pub­lished a post on find­ing ways to mon­e­tize test­ing on media-based and pub­lish­ing sites. In that post I dis­cussed the basic “page views per visit” engage­ment met­ric baked into Test&Target and then briefly touched on the con­cept of Page Score. In today’s post I want to dive a lit­tle deeper into Page Score to address how pub­lish­ers can lever­age this type of met­ric and when it might make sense.

For con­text, let me share an exam­ple we can all relate to. Let’s sup­pose you’re try­ing to gain weight to become a sumo wrestler.  To track your progress you decide to mea­sure the num­ber of meals you eat. Three meals a day isn’t doing it so you step it up to five. Eat­ing five meals a day gets tax­ing rather quickly so before long you find your­self eat­ing smaller meals with more veg­eta­bles and less meat. Soon you notice you are actu­ally los­ing weight.

You dis­cuss this with your weight-gain coach, who rec­om­mends you track the amount of calo­ries you con­sume, rather than the num­ber of meals eaten. With this approach you are able to eat three high-calorie, but oth­er­wise man­age­able meals. As your total caloric intake increases, so does your weight and before long you’ll be on your way to sumo star­dom. In this rather sim­ple exam­ple, mea­sur­ing meals is akin to mea­sur­ing page views. While the meals we eat often cor­re­spond with weight gain or loss, it is not the meal itself that dri­ves that change. Rather, it is the calo­ries inside that meal that move the scale. In our exam­ple calo­ries rep­re­sent the weight of the page, as cap­tured by page score.

Not All Pages Con­tain Equal Calo­ries
While the stan­dard “page views per visit” met­ric excels in its ease of imple­men­ta­tion (essen­tially requir­ing a uni­form global mbox on every page) it assumes all pages have equal value. This may be a good assump­tion for some, but prob­a­bly not for most in today’s com­plex media busi­ness. Let’s assume most land­ings on your site are on the home­page. Let’s also assume that due to the very wide but shal­low pool of con­tent on that page, ad impres­sions sold on the home page have a lower CPM than cat­e­gory pages like “Fash­ion” or “Sports” where adver­tis­ers can place more tar­geted ads, result­ing in a more lucra­tive CPM. A flat-rate page-views-per-visit met­ric won’t be able to dis­tin­guish between a five-page visit that made up of four home page loads and one cat­e­gory page load and a five-page visit made up of one home page load and four cat­e­gory page loads. All this model sees is five uni­form pages. To make mat­ters worse, what if you test a change that makes the home page more dif­fi­cult to scan so vis­i­tors end up back at the home page or load the home page more often because they can’t find the con­tent they want. In this case, page views may be increas­ing but they aren’t the higher-value page loads you are hop­ing for. Your busi­ness could actu­ally be los­ing sig­nif­i­cant money in this scenario!

With a Page Score met­ric you have the abil­ity to work around this prob­lem. With a Page Score para­me­ter you can tell Test&Target that the home­page is worth 0.1 and the cat­e­gory pages are each worth 0.8. Now you can con­struct a weighted average.

Set­ting Page Score
Set­ting Page Score is easy to do. You sim­ply add a para­me­ter to your mbox call on every page like so:

mboxCreate('global_mbox', 'mboxPageValue=10');

Note that while mbox para­me­ters can take any name, the page score para­me­ter is one of those reserved mbox para­me­ters that must be set exactly like so: “mboxPageValue=VALUE”. This is sim­i­lar to other reserved para­me­ters like: orderId, order­To­tal, and pro­duct­Pur­chase­dId. The value used would change based on the rel­a­tive value you would want to assign to the page. For a more com­pre­hen­sive exam­ple, you can read more about it in Test&Target’s help sec­tion.

Pig­gy­back­ing on Your Ana­lyt­ics Clas­si­fi­ca­tions
To make this met­ric effec­tive, you should have a page score assigned and set on every page. Set­ting the mbox­PageValue on every page sounds tedious. It is tedious. It’s the equiv­a­lent of man­u­ally writ­ing your return address on 200 thank you cards instead of buy­ing a stamp. If you’re like most pub­lish­ers, you have dozens – if not hun­dreds of new pages born every day. For­tu­nately, there is a good chance you already solved this prob­lem when you nailed down your ana­lyt­ics imple­men­ta­tion. Hope­fully each page has a page type or some sort of clas­si­fi­ca­tion. This may be the DART zone or the sales cat­e­gory of that page. That may be the axis on which you can assign page score val­ues. If you’re defin­ing the page type vari­able some­where on the page, you may be able to dynam­i­cally inject logic that assigns the page score based on the page type (home page, arti­cle page, cat­e­gory main, etc.)

var pageType= "article"
if(pageType= "home") var pageScore=1;
else if(pageType= "article") var pageScore=4;
else if(pageType= "categoryTop") var pageScore=8;
mboxCreate('global_mbox', 'mboxPageValue=' +pageScore);

Are You Sell­ing Ad Impres­sions or Ad Inven­tory?
These two may sound the same, but what I really mean is if your site has a num­ber of unsold ads (pages load­ing with poten­tial ad impres­sions that have no buyer), dri­ving up page views isn’t nec­es­sar­ily dri­ving up rev­enue. A met­ric like page score may be able to help if you can get clever with your imple­men­ta­tion. Specif­i­cally, if your sold ad units have a unique class name that dif­fer­en­ti­ates them from unsold ad units, you may be able to employ a script on your page (or a Test&Target pro­file script) that loops through all the divs on the page and counts the num­ber match­ing your sold ad unit class. If that num­ber can be set in a JavaScript vari­able, it could then be dynam­i­cally pop­u­lated into an mbox­PageValue parameter. For example:

var divs=document.getElementsByTagName('div');
var adCounter = 0;
adCounter = adcounter + 1;
mboxCreate('global_mbox', 'mboxPageValue=' +adCounter);

Please note that I am not a devel­oper. The code above will prob­a­bly need tweak­ing for your use case, but it should get the point across. The result is an mbox on every page that com­mu­ni­cates the num­ber of sold ad impres­sions that appeared on that page. If you have mul­ti­ple ad sizes or types with dif­fer­ing val­ues, your scripts can get more elab­o­rate to pass in rel­a­tive val­ues. For exam­ple a right-rail ad unit with class “ad_rr_300x260” may be worth 0.3, but a fea­tured ban­ner ad with class “ad_featured_1000x260” may be worth 0.7.

Don’t Get Par­a­lyzed by Per­fec­tion
I will wrap this up by reit­er­at­ing what I said in my orig­i­nal blog about mon­e­tiz­ing engage­ment. It is easy to get side tracked with the idea of per­fec­tion. If the per­fect met­ric is going to require you to wait a full year to get it imple­mented – it’s not the per­fect met­ric. Start with some­thing you can get work­ing in the next 4–6 weeks. Any­thing beyond that is likely to never hap­pen. I have seen test pro­grams stum­ble and then fail com­pletely because they were par­a­lyzed by perfection.

There are going to be flaws with any mon­e­ti­za­tion model you work out. Pick some­thing that is an improve­ment from where you are today and work on that. Page Score may be the next level for you. If so, start deter­min­ing where the calorie-dense pages are on your site and plug this into your mon­e­ti­za­tion model.  As always, reach out to your Adobe Test&Target con­sul­tant for more ideas on how you page score can help you.