In my last post, I addressed the need for dig­i­tal pub­lish­ers to seg­ment their audi­ence to ful­fill demand across the entire demand curve and max­i­mize rev­enue. In this post I’ll review the his­tor­i­cally inef­fi­cient ways media has been bought and sold, and why those meth­ods are yield­ing to more effi­cient trans­ac­tional models.

Direct Media Sales

With the advent of dig­i­tal media in the nineties, pub­lish­ers quickly estab­lished a direct sales model to mon­e­tize online con­tent. For large pub­lish­ers, direct sales offered a sim­ple way to mon­e­tize the most valu­able, pre­mium ad inven­tory while main­tain­ing con­trol over the impact to their brands. The direct sales model does not rely on inter­me­di­aries; it lever­ages rel­a­tively sim­ple tech­nol­ogy, and it is easy for pub­lish­ers to control.

A direct sales model has been (and remains) an effec­tive way to mon­e­tize the upper end of the demand curve for pub­lish­ers with suf­fi­cient scale to sup­port a direct sales team. How­ever, the direct model is not effec­tive at the mid to low end of the curve for a cou­ple of reasons:

  1. ROI from direct sales decreases the fur­ther you move down the demand curve. Sales over­head increases rel­a­tive to ad rev­enue and costs become pro­hib­i­tive for a direct sales team.
  2. Direct sales are an inef­fi­cient means of match­ing adver­tis­ers with gran­u­lar audi­ence seg­ments. Vis­i­tor attrib­utes and seg­men­ta­tion increase the value of ad inven­tory when adver­tis­ers opti­mize against this data, but gran­u­lar seg­ments can­not be bought and sold through direct sales teams.

The fol­low­ing graph illus­trates how direct sales address the high end of the demand curve and mon­e­tize the most valu­able inven­tory. How­ever, direct sales are not an effec­tive way to meet demand across the mid to long tail.

Direct ad sales monetize the upper end of the demand curve

Direct ad sales mon­e­tize the upper end of the demand curve

Dis­play inven­tory is per­ish­able and there­fore loses its value once a page loads with­out an ad. Pub­lish­ers who want to mon­e­tize inven­tory that is not sold by a direct sales team, and pub­lish­ers who are too small to effec­tively uti­lize a direct team turned to “rem­nant” sales channels.

Rem­nant Media Sales

Rem­nant media solu­tions allowed pub­lish­ers to mon­e­tize more inven­tory and allowed more adver­tis­ers to run dis­play adver­tis­ing cam­paigns, cre­at­ing a larger dis­play mar­ket­place mon­e­tiz­ing more area under the demand curve. Early rem­nant solu­tions, how­ever, could only mon­e­tize the low­est price points on the curve because:

  • Large quan­ti­ties of inven­tory put down­ward pres­sure on CPMs
  • Rem­nant sys­tems included few con­trols and pre­sented greater risks, deliv­er­ing less value to both brand and per­for­mance advertisers
  • Rem­nant was pri­mar­ily an undif­fer­en­ti­ated pool of ambigu­ous media

The graph below illus­trates a com­bined Direct and Rem­nant strat­egy for sell­ing media. These two strate­gies taken together mon­e­tized a greater por­tion of the demand curve for most pub­lish­ers but pre­sented addi­tional prob­lems in the form of chan­nel con­flict and data leakage.

Combined direct and remnant ad sales monetize the upper and lower ends of the demand curve

Com­bined direct and rem­nant ad sales mon­e­tize the upper and lower ends of the demand curve

Chan­nel Con­flict & Data Leakage

A com­bined direct and rem­nant strat­egy tapped addi­tional demand for dis­play adver­tis­ing but also intro­duced chan­nel con­flict for pub­lish­ers. Adver­tis­ers will­ing to accept the uncer­tainty and risk of the rem­nant mar­ket some­times found they could get high value ad impres­sions through the rem­nant chan­nel at a frac­tion of the direct price. Pub­lish­ers saw the value of their inven­tory decrease through direct chan­nels and in some cases dis­con­tin­ued rem­nant sales.

Rem­nant mar­ket­places and an increas­ing num­ber of ad net­works pro­lif­er­ated tags on pub­lisher sites to col­lect data and mon­e­tize rem­nant inven­tory. How­ever, along with an increase in rem­nant mon­e­ti­za­tion came an ero­sion of inven­tory value when third par­ties re-monetized data for their own ben­e­fit. Re-monetization through data leak­age occurs both legally and ille­gally depend­ing on the terms of access granted to third parties.

The loss of audi­ence mon­e­ti­za­tion for pub­lish­ers due to data leak­age is com­pa­ra­ble to lost rev­enues sus­tained by the dig­i­tal record­ing indus­try when con­sumers share dig­i­tal IP with­out mon­e­ti­za­tion flow­ing back to the rights holders.

The fol­low­ing graph illus­trates the ero­sion of mon­e­ti­za­tion due to chan­nel con­flict and data leak­age. In addi­tion to the lack of mon­e­ti­za­tion across the mid-section of the demand curve, mon­e­ti­za­tion of pre­mium and rem­nant sales decreases.

Channel conflict and data leakage erode the value of advertising inventory

Chan­nel con­flict and data leak­age erode the value of adver­tis­ing inventory


The dis­play adver­tis­ing mar­ket­place is a large and frag­mented col­lec­tion of adver­tis­ers and pub­lish­ers seek­ing more effi­cient ways of trans­act­ing media for opti­mal adver­tis­ing results. The early mech­a­nisms for buy­ing and sell­ing media were sim­ple but inef­fi­cient. Tech­nol­ogy and inno­va­tion are enabling bet­ter ways to trans­act media and I’ll dis­cuss those in my next post.