I currently have one proven landing page, so the only different variations are the offer combinations.
I have almost 20 offers to test from, but because of the geo only getting about 300k impressions a day, I'm unable to test quickly.
How do you recommend spliting up the campaigns to test different groups of offers (Galaxy S9, Iphone X, Vouchers, Etc?)
My one concern is running multiple campaigns to different offer groups. But that would mean that I'll be bidding against myself. Is this a good strategy?
Or is there any other way to test all these offers quickly rather than throwing them all into one campaign and waiting for something to happen.
300k impressions a day is PLENTY!
Each "prize" (samsung galaxy, iphone x etc.) will need a separate set of landers showing that respective prize, so you'll need to group offers accordingly and use their own set of landers (for each prize). Whether you set that up as different paths within the same campaign (with each path consisting of offers and landers of one particular prize) or as different campaigns (one campaign per prize) is your choice. I wouldn't worry about competing with yourself - if you set the same bid across the campaigns (for different prizes), you'll likely get the same amount of total traffic or more, compared to if you had set up a single campaign with the same targeting and bid. And setting them to the same bid will give the best chances of getting similar amounts of traffic to each campaign (on some traffic networks and geos, if you set up multiple campaigns with the same targeting but different bids, most of the traffic would go to the camp with the highest bid - which is something you probably won't want if you're wanting to test all prizes simultaneously).
As for testing speed: The fastest way to test all 20 offers is by testing them simultaneously, because the more offers you're testing, the faster the inferior offers will reach statistical significance meaning the faster you can eliminte them. Please see my signature on the bottom on how to compare offers using a stats tool to identify a winner.
Having said that, if you're not wanting to wait for all 20 offers to finish running, you don't have to run all of them at the same time. You can run in the first batch the ones that you feel hold the most promise (e.g. offers recommended by your AMs or other contacts, offers showing good trends in spy tools, etc.), cut down to a winner, and then test that winner against the rest of the offers in another round of split testing. This way you'd feel less impatient. However, be aware that overall speaking, you'd be spending more money and time on the testing compared to running all 20 at once. This is because the more offers you run simultaneously, the greater the chances that you've included the best and worst offers, so the gap in performance would be greater, which would lead to statistical significance being reached faster. Also, you'd be testing at least one offer twice, as the winner from testing the first batch of offers, would need to be included in the testing of the second batch of offers (it is called the "test control" - like a current-best offer you'd be trying to beat). Psychologically though you may feel better because each individual test would likely be shorter, although you'd be doing 2 rounds of testing in total.
There IS another advantage to testing offers in smaller batches: If you test the more-promising ones first, and find one that converts well enough to turn your campaign green (optimization may be necessary), then you may choose not to test the rest of the offers for now, and just scale the heck out of that offer first. You can always siphon a smaller percentage of the traffic to these remaining offers to test them, while still sending most of your traffic to the profitable offer. If one of the new offers end up converting better than the original profitable offer, you can always choose to switch to that new and better offer.
So - different approaches, and pros and cons to each. Choose one or try all approaches - your call.
Lastly - there is no need to be twiddling your thumbs and tapping your feet while waiting impatiently for the testing to get done! There is ALWAYS more testing you could set up! More landers, more offers, more geos and more traffic networks. When doing testing, it would actually be better to have several campaigns going on at the same time - that way you're not so eager for any one campaign to succeed. This kind of focus can have negative effects on your campaign. It's synonymous to being an over-protective parent to your only child (it may not be a very appropriate analogy but hopefully good enough to get my point across). Holding your breath for the only campaign you're running to succeed can invoke anxiety to cloud your judgment. Whereas if you have several running, you wouldn't be expecting every/any one to be a winner. Also, when running several test campaigns, the chances of finding something promising would be greater. The only downside is you'd be spending more on testing, which can potentially lead to cashflow issues if budget is very limited. But other than that, I'd encourage affiliates that know all the basics of campaign optimization to run multiple campaigns in parallel, for all the benefits mentioned above.
Hope I haven't just confused you further! Have fun and feel free to ask further questions.
Amy
Sent from my SM-G930W8 using STM Forums mobile app