Attri­bu­tion mod­els are great, but the real ques­tion is what are you going to do with them? In this post I will share Attri­bu­tion Tips employed by suc­cess­ful com­pa­nies. And I promise that you will not shed a tear wav­ing good­bye to last-click track­ing after read­ing this article.


Edu­ca­tion is one of the most impor­tant steps you will take when you begin to look at your attri­bu­tion imple­men­ta­tion method­ol­ogy. Under­stand­ing when and how to uti­lize attri­bu­tion logic lets you opti­mize your cus­tomer expe­ri­ence and mar­ket­ing tactics.

For exam­ple, imag­ine the paid search depart­ment see­ing their cost-per-order increase because it wasn’t actu­ally the last action before plac­ing an order by the cus­tomer. Then imag­ine the dis­play depart­ment cheer­ing because they are receiv­ing more credit for help­ing the con­ver­sion prior to the last click.

  • Attri­bu­tion tip: This type of dis­rup­tion across all depart­ments is not some­thing easy to imme­di­ately com­pre­hend. Really suc­cess­ful com­pa­nies edu­cate their depart­ments on what they can expect to see before they even look at the num­bers. Attri­bu­tion mod­els will help you, but don’t expect their imple­men­ta­tion to be easy or serve as a golden ticket to resolve all of your problems.
  • Attri­bu­tion tip: Con­vey to all depart­ments that they need to work more closely together with attri­bu­tion to improve the over­all ROI across all channels.

It’s inter­est­ing to see teams work­ing together to push the over­all ROI.  Typ­i­cally each chan­nel works in silos and less as a group look­ing to opti­mize one of the most impor­tant metrics…the over­all ROI. The exam­ple below out­lines the over­all ROI as $8.10 for all the chan­nels.  This num­ber should be shared across all teams to get them to improve that return.

  • Attri­bu­tion tip: Users can mon­i­tor mul­ti­ple screens and pre­form actions that are not tracked in an attri­bu­tion model. Some clients do have the capa­bil­ity to track known cus­tomers across screens but it is dif­fer­ent for each imple­men­ta­tion. The Uni­fied Cus­tomer Process in Adobe Insight has the keys to unlock this com­plex issue.



Dedu­pli­ca­tion is the remov­ing of dupli­cate events and vis­its from ana­lytic report­ing. If you have mul­ti­ple track­ing sys­tems and mul­ti­ple sources, Insight will help you “dedu­pli­cate” those leads in their attri­bu­tion models.

Con­sider, you have 100 Orders but there are really only 80 unique orders because you employ two track­ing sys­tems.  If you clicked on both ads and then con­verted, each would get credit because those sys­tems don’t com­mu­ni­cate with each other. This issue alone is extremely help­ful to look at by your mar­ket­ing depart­ment to see what is really happening.

  • Attri­bu­tion tip: Look at a month’s worth of data and define each of the chan­nels included.  Who per­forms the real last click or view? How does that com­pare to all of the silo reports that you get from each depart­ment? What is their true cost per lead and/or order?


My two favorite attri­bu­tion mod­els are the Starter/Players/Closer model and the Even model.

In the fol­low­ing exam­ples, I will focus specif­i­cally on Starter/Players/Closer. If you are not famil­iar with the Starter/Players/Closer model, here is a high-level explanation:

For the Starter/Players/Closer attri­bu­tion model: The Starter (ini­ti­at­ing) touch receives a set per­cent of credit for the con­ver­sion, the Closer (last) touch receives a set per­cent­age of credit for the con­ver­sion, and the remain­ing Player touches each receive an equal share of the remain­ing credit for the con­ver­sion. See the Advanced Web-Store Attri­bu­tion in Insight blog from Michael Halbrook.

The Even attri­bu­tion model accom­plishes the same but applies an even per­cent­age break­out of rev­enue and orders. Typ­i­cally the Starter will get 30%, Play­ers gets 30% and Closer will get 40%.

This attri­bu­tion can be a dream come true for large orga­ni­za­tions who strug­gle to under­stand how to break out rev­enue. If you do not like the sub­jec­tiv­ity of assign­ing a per­cent­age, then you can sim­ply change to the Even attri­bu­tion model.


  • Attri­bu­tion tip: Look to see how each of your chan­nels fall into Starter/Players/ Closer with rela­tion to orders.  Typ­i­cally you will see that Ban­ner view/clicks have more rev­enue on starter/player than a Closer.  Typ­i­cally, the SEO/Paid search will usu­ally have more rev­enue toward Closer because they usu­ally get last-click revenue.
    • In my mind, you will want to make sure that Closers have a large enough bud­get to fully run and close the deal. Play­ers are impor­tant because they con­tribute to the final closed trans­ac­tion. Starters will begin the con­ver­sa­tion and get vis­i­tors inter­ested to buy.
      If an orga­ni­za­tion wants to drive more rev­enue, they can take a look at those Starter chan­nels and increase fund­ing.  For exam­ple, when plan­ning your ban­ner buy, can you exam­ine what is the SOV on that site? Is there room to grow and tar­get more of those users? It may not be a direct cor­re­la­tion, but this can be an inter­est­ing way to deter­mine where to spend your budget.
    • Attri­bu­tion tip: Now that you have dedu­pli­cated the leads, what is the ROAS per depart­ment with SPC per­cent weights in place? What is your ROAS for Even attri­bu­tion rev­enue? You can adjust and opti­mize your mar­ket­ing bud­get to areas that drive suc­cess and lower the areas that are pulling down the total ROAS.
    • Attri­bu­tion tip: Pre­dic­tive mar­ket­ing can also help you fig­ure out the value of those weights.  If you decide to go down this route, Adobe has a team that is strictly ded­i­cated to pre­dic­tive mar­ket­ing.
      My col­league Trevor Paulsen talks about pre­dic­tive mar­ket­ing in his blog post Attri­bu­tion Mod­el­ing– It’s Like Ulti­mate Fris­bee.


Attri­bu­tion is great, but what if I have dif­fer­ent cus­tomers buy­ing dif­fer­ent prod­ucts. Do you think a per­son buy­ing a lawn­mower has the sim­i­lar mar­ket­ing touch points as some­one buy­ing a crib? Insight allows you to cre­ate cus­tom fil­ters to ana­lyze at a cat­e­gory level.

Let’s layer this onto the attri­bu­tion models…wow this is get­ting exciting!

  • Attri­bu­tion tip: Look at your lines of busi­ness and see how the behav­ior of the cus­tomer changes by cat­e­gory type. Con­sider how you will spend your mar­ket­ing funds in rela­tion to the prod­uct cat­e­gories. What would you do if the CEO told you to sell more lawn­mow­ers tomor­row? If you are going to sell more lawn­mow­ers you need to take a detailed analy­sis on that seg­ment type and see if you can push harder on those chan­nels that work to drive rev­enue volume.

I know my mom very well. When she wants to visit a spe­cific web­site she goes to Google and types in the actual URL in the search box. It’s just adorable. But ask your­self: Who should get the rev­enue asso­ci­ated with her pur­chase later in that visit?

  • Attri­bu­tion tip: Adobe Insight allows you to put that rev­enue toward Direct instead of SEO. This type of logic can be put into Insight as a con­di­tional state­ment to do that type of rev­enue dis­tri­b­u­tion. This alter­ation is done in the mar­ket­ing chan­nels con­fig­u­ra­tion with sim­ple logic rules and order. This is an open box so you can feel free to define them to meet your spe­cific busi­ness needs. Other com­pa­nies that do attri­bu­tion are very cum­ber­some and pro­vide only for­ward look­ing attri­bu­tion.  Insight allows you to adjust attrib­utes at any time and can reflect upon its his­tor­i­cal data.


Most com­pa­nies will take a look at the ROI when they incor­po­rate vari­able costs. That’s great, but they may be under­es­ti­mat­ing some expenses: your salary, tech­nol­ogy fees, agency man­age­ment fees, and a whole mess of other costs. Obvi­ously some of this infor­ma­tion can be aggre­gated and shared with a lim­ited amount of peo­ple to defray costs across the orga­ni­za­tion, but it is still a sig­nif­i­cant issue.

There is also a very sub­jec­tive way to share the total cost of Ana­lytic soft­ware based on each customer’s vary­ing approach to orga­ni­za­tional goals.

  • Attri­bu­tion tip: If you want to hire more peo­ple into a spe­cific chan­nel, does the ROI jus­tify it? To find out, layer on Starter/Players/Closer and Even mod­els to see how well each depart­ment is really per­form­ing. Organic growth might have the most com­pelling mar­ket­ing ROI, but with a closer look, Paid Search may have the actual strongest ROI based on reported met­rics. Con­se­quently, the CXO may real­ize that the Organic chan­nel is not pro­duc­ing the best ROI after you fac­tor in the hard cost to get a Com­pany ROI. If he had the option to invest more in Organic or Paid Search, you can be sure the invest­ment will be made in Paid Search after defin­ing its real value.

I come from the Agency and Client side which makes me appre­ci­ate attri­bu­tion a lot more than most. I under­stand that the last click mod­els have held a lot of large com­pa­nies back from truly under­stand­ing their data and track­ing rev­enue for each of their mar­ket­ing channel.

Attri­bu­tion is a sig­nif­i­cant imple­men­ta­tion and invest­ment issue that deter­mines ongo­ing track­ing and sales goals. If you want to learn more about Attri­bu­tion Mod­els, reach out to some­one at Adobe. We are here to help you cus­tomize your Adobe imple­men­ta­tion based on your spe­cific needs!


Ryan Green