Case study - Promoting music with Adwords
In the case with Adwords the keywords are mostly relevant, the typical approach is to start out with a big list and widdle down getting more relevant as you do, I’ve done this a lot with Bing search ads and I’m trying to replicate it with Adwords/YouTube.
“do you have a recommendation on what we should try setting as the bid amount?”
On Adwords, bidding for display placement with any number of targeted keywords I’ve found a comfortable bid to start with is .05, work up slowly if you dont get your ads served, or add in more keywords.
The video played to percent for in-stream and its .02 average CPV (cost per view) seems to good to be true. There must be something going on with this placement. I thought it was me messing something up and that’s why I was saying I was having issues with Adwords, but every camp is showing this trend. I believe something about the users in this placement is causing the data to skew. Maybe they mean 100% of the 5 seconds a user has before they can skip. Or maybe the bulk of the users served to in this placement are away from keyboard or asleep. Maybe I’m missing something here since I never use Youtube.
As you can see some “trimming” will need to be done ultimately based on relevance and average view duration.
The Adwords campaigns have definitely developed more insight and i have a number of screenshots to share. As I stated before I think there is something skewing the data’s face value for in stream placement, which seems to be more evident with the latest look at the data.
As you can see although in stream reports much higher “view duration” rates, there are next to no results beyond that, in terms of social activity, which all but confirms its lack of effectiveness.
Another approach I’m adapting is lowering the bids to .01 for both, as it seems with the bulk keywords approach, you will end up paying your bid no matter what, and filling this daily budget seems to be no issue at all, since they both cap early every day.
In terms of keywords you can see how I might now begin to trim, by only displaying keywords with x number of views, and sorting by either viewed to 100% duration or just number of total views, and finding the keywords least relevant to delete from our list, for each respective campaign, which have their own keywords list, but are identical lists nonetheless.
Okay I have much to share and it has been a longer stretch of time between updates, so I’m gonna try to cover all the important stuff.
First off, in-stream placement is useless, don’t use it. If you check the stats coming off ads in that placement, they result in less earned views, even though the reports say they garner ten times more raw views. They also get next to no social interactions compared to in display, even though they get more raw clicks. My theory on this is that they are accidental clicks.
In terms of keyword specificity, I ran a few ad groups at different bids to see what approach looked best. I had a 10,000 long non targeted list with a .01 bid, not expecting it to get much placement, but if they can be had for a penny it would be worth seeing if the performance rate multiplied by the price is more than full priced targeted list. If they are equal then the better choice is the broad list, because it expands our demographic considerably, and also has more potential for optimizing. Always think big like that.
Otherwise the two most notable ad groups were using keyword lists that were specific, but one had roughly 130, while the other had closer to 300. They both had bids of .05 for the larger and .08 for the small I believe. The smaller got much more placement at a higher price ($0.07) while the larger sent about a third as much traffic at a lower price ($0.06). The bigger list had better retention though which is surprising.
Secondly, I’ve made a significant discovery. Around the first of the month I tried an idea. I replaced the tags for my video from ones that were most accurate, to the most popular keywords from the list I built, that were still generally accurate. You can see the screenshots for examples. As soon as I did this I started receiving more organic traffic from the suggested videos widget on Youtube. So yesterday, after realizing this, I uploaded a few more videos and used the same approach. They are included in the screenshots so you can see multiple variants of the approach. Today I received more organic traffic then any other single day. This seems to be most effective strategy if developed on. This is almost exactly the approach one Reddit commentor employs with his Youtube marketing strategy so hats of to VideoGameDJ. I’m thinking by bolstering the video with paid views, you can jump-start it’s organic reach potential. I wanna see what happens after I turn the ads off, after one more day of collecting data on the ad group I left on today.
I left the 10,000 long keyword list at $0.01 big, and the 300 long targeted list down to $0.05 and paused the rest. The daily budget still sits at $30.