1 week FB - What I have (not) Understood (1)
Finally we’ve decided to expand the business to facebook. Here is the first campaign! We picked a utility app with mass appeal, for quick and cheap learning. Hope we can provide some learning aspects here, as well as selfishly learn faster. Zeno’s FB guide was really helpful for –at least partly- understanding that beast.
GEO: India
Payout: 0.18$
Offer: Android App Install - Flash Keyboard (authorized)
CAMPAIGN SETTINGS: Campaign objective is App Installs. Ad sets are set to optimize towards App Installs, while we are paying for impressions (paying for app installs is min. € 40 daily budget / ad set).
BIDDING: Starting for the suggested bid (or a bit higher) per install. Then aiming for lowering it, as facebook generates data about its feasibility.
OPTIMIZATION: From a first broad test, we picked the most promising placement, demographics, and seemingly best creatives. However, fb’s algorithm makes it very difficult to later on split test new creatives, as it seems to prefer old ads with history.
We’ve been experimenting with two ways of adding fresh creatives:
1. Putting new ads into existing ad set with good history
Most of the time the ads simply do not receive sufficient traffic:
Strategy: Looking at it fairly often to see if the ads stick or not. And if not, start launching new ones and repeat.
2. Generating new ad sets with equal targeting and adding new creatives.
Problematic, because even if we duplicate the proven ad sets, we have to start at a higher bid to receive traffic, as fb is obviously missing conversion data (remember we bid on app installs). Which finally leads to a higher CPM/CPI until the ad set gets optimized. Not easy to compare creative performance this way.
Facebook delivers unequal no matter what optimization for ad delivery and charging method we chose:
pay: CPM & optimize for Impression
pay: CPM & optimize for app installs
pay: CPC & optimize for website clicks
At the ad level as expected, better CTR and engagement get support from facebook. But sometimes fb just does not give a chance and stops traffic after a few impressions:
Creative1 and 2 has been started at the same time.
Even more surprising was to see this at the ad set level (only creatives were different), with 3 identical ads in each ad set.
What we’ve learned/experienced:
#Facebook has its own rules and shuts down traffic to specific ads even before they have a chance to perform well.
#Facebook helps you optimize to your campaign and ad set aim. There was a massive difference in optimizing for Clicks vs App Installs vs Impressions.
#The Bid size should be sufficiently high (higher range of suggested) at first to get traffic and prove facebook that it can deliver at a lower price per app install. Later, reducing the bid at the right time can reduce the CPI even more without cutting the traffic (can depend on potential reach and budget though).
#Apps like Flash Keyboard are for the masses, targeting to certain interests on facebook won’t really help the campaign according to our experience
#Keep in mind the ‘budget, bid and potential reach ‘correlations. The key to success seems to be harmonizing these three factors.
Questions:
1. How do you make sure you can objectively test different creatives? Can we enforce FB for more equal ad delivery somehow?
2. Does facebook take over historical data when duplicating an ad set / campaign / ad?
3. If yes: After duplicating existing (proven) ad sets, do you adjust the bid? How?
4. What are your thoughts on the ‘Budget, potential reach and bid’ relationship?
5. Does an ad receive organic traffic (when getting liked/shared)? If so, do we get conversions out of it (remember: authorized App. No affiliate/tracking link)?
6. What Facebook Marketing Partners (PMD) can you recommend for starting out?
Qwaya lacks functionality which is important to us (e.g. video support). However, most other platforms do not support S2S tracking.
Home > Paid Traffic Sources > Facebook & Instagram