Google ACE is an excel­lent tool that came out in Octo­ber last year, which allows for sci­en­tif­i­cally sig­nif­i­cant test­ing across many fea­tures within an Adwords Cam­paign. In this blog post we won’t dis­cuss how to actu­ally imple­ment ACE tests (Google has that cov­ered), but will focus more on the func­tion­al­ity and best prac­tice when set­ting up one of these tests.

So what is ACE?
ACE stands for Adwords Cam­paign Exper­i­ments, and it is a tool that allows you to split poten­tial Google traf­fic at its source. It is applied at the cam­paign level, and traf­fic can be split in a num­ber of ways (usu­ally we opt for 50:50) between a ‘con­trol’ group and an ‘exper­i­ment’ group. You can also put parts of the cam­paign into both groups at the same time. Best prac­tice here is to have every­thing you aren’t test­ing in both groups, the new part of the cam­paign you want to test in ‘exper­i­ment’ group, and the cur­rent part of the cam­paign (i.e. the part that you want to replace if the test goes well) in the ‘con­trol’ group (this is explained in more detail fur­ther down).

What can you test?
You can cur­rently test any­thing you can change within a cam­paign. You can add new key­words, neg­a­tives (Adgroup level), place­ments, dif­fer­ent ad copy and alter­na­tive Adgroup struc­tures. How­ever there is cur­rently no func­tion­al­ity to test any­thing applied at the cam­paign level. I have heard talk that the next thing to be added to ACE is the func­tion­al­ity to test site links, which would be nice, but our wish­list also includes GEO-targeting, cam­paign level neg­a­tives, net­works, devices, loca­tion exten­sions, day­part­ing and cam­paign budgets.

How do you test?

We have found that the biggest lim­i­ta­tion with ACE is its report­ing.  Google was obvi­ously keen to get this use­ful tool in the mar­ket­place as soon as it could, so report­ing seems to have been neglected. The way we get round this at EF is by build­ing dupli­cate Adgroups (a generic exam­ple is below).

Cam­paign: Hol­i­day Des­ti­na­tions:
Con­trol Greece Exper­i­ment Greece
Con­trol Por­tu­gal Exper­i­ment Por­tu­gal
Con­trol Spain Exper­i­ment Spain
Con­trol France Exper­i­ment France
Con­trol Italy Exper­i­ment Italy

This is the start­ing point for an ACE test (the above exper­i­ment adgroups are exact repli­cas of the con­trol groups). We would then make the changes we wanted to test (i.e. a new land­ing page URL) to the exper­i­ment groups only, and start the ACE test to split the traf­fic 50:50 between both groups. Many of our account man­agers now main­tain a con­tin­u­ous ‘test’ cam­paign, purely for ACE test­ing as they always have sev­eral tests run­ning at any given time. This ethos of con­tin­u­ous test­ing is impor­tant in PPC (and a pol­icy that Google itself has always cham­pi­oned) because it is the only way to keep learn­ing ways to increase the per­for­mance of your account / busi­ness in the long-term.

So what is the ben­e­fit of using ACE?
In short, flex­i­bil­ity and accu­racy. Pre­vi­ous tests relied on the before and after tech­nique of mak­ing a change and com­par­ing last week’s data with this week’s. The prob­lem with that approach is that no two weeks are the same. Any­one who has been run­ning a PPC cam­paign for some time (par­tic­u­larly those work­ing on large-scale, com­plex accounts) can tell you that per­for­mance changes because of  a vari­ety of con­flict­ing vari­ables, includ­ing con­sumer demand, sea­son­al­ity, the weather, even the qual­ity of what’s on TV!  Using ACE means both the con­trol and exper­i­ment group are being exposed to the same mar­ket con­di­tions. You can write off all of those vari­ables because they should affect both parts of the account equally. You can be more con­fi­dent that any dif­fer­ences in per­for­mance are due to your changes, rather than a sta­tis­ti­cal fluke.

As well as being more accu­rate, ACE is more flex­i­ble. Pre­vi­ous ad copy tests relied on putting the whole cam­paign on even-rotate. This approach is inef­fi­cient as your worst per­form­ing ads in other ad groups are now being shown more often than they should, drop­ping qual­ity score, inflat­ing CPCs and decreas­ing con­ver­sion rates. Using ACE allows you to test just the two ads you want to test, leav­ing every­thing else in the cam­paign as it is.

Con­clu­sion
Used cor­rectly and exten­sively, ACE is a great way to dis­cover action­able insights into what dri­ves per­for­mance in your account. It doesn’t take long to set up, is fairly intu­itive and can help you drive per­for­mance. I would highly rec­om­mend any­one not using ACE to start using it as much as possible.

Michael Tay­lor, Asso­ciate Account Man­ager, Effi­cient Frontier

0 comments