Using Experiments to Get Crazy(ier) But Still Be In Control By Susan Wenograd

21
#SMX #11B @SusanEDub Using Experiments to Get Crazy(ier) But Still Be In Control

Transcript of Using Experiments to Get Crazy(ier) But Still Be In Control By Susan Wenograd

Page 1: Using Experiments to Get Crazy(ier) But Still Be In Control By Susan Wenograd

#SMX #11B @SusanEDub

Using Experiments to Get Crazy(ier) But Still Be In

Control

Page 2: Using Experiments to Get Crazy(ier) But Still Be In Control By Susan Wenograd

#SMX #11B @SusanEDub

Hi! I’m Susan and I’m a paid media nerd.About me

Page 3: Using Experiments to Get Crazy(ier) But Still Be In Control By Susan Wenograd

#SMX #11B @SusanEDub

SEM and Social Ads Nerd at Five Mill, managing multiple six figures in spend per month.

Paid media for the past 10 years (good grief, seriously?)

Live in Virginia with some very cute people.

About me

Page 4: Using Experiments to Get Crazy(ier) But Still Be In Control By Susan Wenograd

#SMX #11B @SusanEDub

Someone says, “We should test this.”

Someone else involved usually reacts like this:

Testing can cause anxiety…

Page 5: Using Experiments to Get Crazy(ier) But Still Be In Control By Susan Wenograd

#SMX #11B @SusanEDub

What if you could test in a less scary way? (You can!)

Page 6: Using Experiments to Get Crazy(ier) But Still Be In Control By Susan Wenograd

#SMX #11B @SusanEDub

Duplicate

Test

Rejoice

Using Drafts & Experiments

Page 7: Using Experiments to Get Crazy(ier) But Still Be In Control By Susan Wenograd

#SMX #11B @SusanEDub

Using Drafts & Experiments

Click into a campaign and create a draft

Use a descriptive title to make things easier later.

Page 8: Using Experiments to Get Crazy(ier) But Still Be In Control By Susan Wenograd

#SMX #11B @SusanEDub

Edit your draft….

Note that performance data carries from source

Page 9: Using Experiments to Get Crazy(ier) But Still Be In Control By Susan Wenograd

#SMX #11B @SusanEDub

Side note about Drafts…

Work “in-progress” outside of AdWords Editor

Easy to QA/Share with team/client

Page 10: Using Experiments to Get Crazy(ier) But Still Be In Control By Susan Wenograd

#SMX #11B @SusanEDub

Time to make your changes… Edit the draft for your test - Campaign settings

alone are a huge area of potential tests!

Don’t forget to add any necessary internal tracking used for this separate segment!

Page 11: Using Experiments to Get Crazy(ier) But Still Be In Control By Susan Wenograd

#SMX #11B @SusanEDub

Apply the changes and kick it off

Page 12: Using Experiments to Get Crazy(ier) But Still Be In Control By Susan Wenograd

#SMX #11B @SusanEDub

Choose your dates & percentages Be clear about what you are testing

Name so it makes sense (especially if you do a lot of them)

Choose your dates– Think ahead for stat sig amounts– Choose your traffic split

(minimize risk with smaller %)

Page 13: Using Experiments to Get Crazy(ier) But Still Be In Control By Susan Wenograd

#SMX #11B @SusanEDub

Where to find it once it’s live…

Page 14: Using Experiments to Get Crazy(ier) But Still Be In Control By Susan Wenograd

#SMX #11B @SusanEDub

Where to find it once it’s live…

Experiment-level data will populate in the Experiment view, but show separately in account totals:

Page 15: Using Experiments to Get Crazy(ier) But Still Be In Control By Susan Wenograd

#SMX #11B @SusanEDub

Understanding your results…

All metrics for the experiment are available in normal data when in the experiment view

Icons helpfully indicate what is stat sig and not (breakout on the next slide) Specific experiment data appears below this table in a normal Ad

Group/Keywords, etc view

Page 16: Using Experiments to Get Crazy(ier) But Still Be In Control By Susan Wenograd

#SMX #11B @SusanEDub

Understanding your results…

Page 17: Using Experiments to Get Crazy(ier) But Still Be In Control By Susan Wenograd

#SMX #11B @SusanEDub

Example outcome…

Five Mill Bidding Platform vs. Google CPA Bidding:– Google’s CPA 86% higher – Google’s conv rate 30% lower – Our tool had +30% margins (147% ROAS), Googles tool had -25% margins (75%

ROAS) – our ROAS is 200% better – Google sent 80 questionably incremental converted clicks at an additional cost

of 2217eu,  making the CPA on those incremental conversions 27.70 which is 4x our CPA target for this campaign.

*** TOTALLY DEPENDS ON THE QUALITY OF YOUR CPC BIDS/PLATFORM/STRATEGYP

Page 18: Using Experiments to Get Crazy(ier) But Still Be In Control By Susan Wenograd

#SMX #11B @SusanEDub

Another example outcome…

“We want to try a new landing page.” Unlike rotating ad versions, this gives us complete control over traffic

flow Early on, but already seeing reassurance our original landing page was

spot-on.

Page 19: Using Experiments to Get Crazy(ier) But Still Be In Control By Susan Wenograd

#SMX #11B @SusanEDub

What else?You can’t test some automated bid strategies (Booooooooo) Target search page location Target ROAS Target outranking share

If your Experiment does better, you can convert it into a “normal” campaign!

Page 20: Using Experiments to Get Crazy(ier) But Still Be In Control By Susan Wenograd

#SMX #11B @SusanEDub

Duplicate

Test

Rejoice

…Now it’s your turn! Go!

Remember, three easy steps…

Page 21: Using Experiments to Get Crazy(ier) But Still Be In Control By Susan Wenograd

#SMX #11B @SusanEDubLEARN MORE: UPCOMING @SMX EVENTS

THANK YOU! SEE YOU AT THE NEXT #SMX