What is the main focus of Blackbird Ppc
Enterprise experience combined with high-touch client service
NPS is our core KPI - Client first mindset
We hire for ambition and pride in craftsmanship
Intense immersive training - All channels are taught to everyone
Always measure incrementality/True value/Conversions that matter
General Strategies and Concepts
Scale and efficiency are two sides of the same coin. If you want more conversions then you are going to need to pay more for them
Make sure that you are using whatever UTM or tracking parameters that your client needs
Exclude near term converter from your remarketing to avoid “ghosting” people through the internet.
Reduce investment in poor performers, increase in better performers.
sem top differentiators
Bids: Make sure that bids reflect the value of the click. Any KW can be profitable with the correct bid. The goal is to trade scale for efficiency.
Account Structure: Are queries moving across many different KWs? Inversely are KWs aggregating too many high volume queries.
Traffic Slashing: Often times potentially good targeting is turned off instead of optimizing. ie budget caps, geo exclusions, over dayparting, pausing relevant KWs, etc. There is no such thing as bad traffic, just bad bids.
CRO: By optimizing landing pages we can improve CVR up to 20% - 30%.
Maximize Extensions: They are free to add and improve QS. No need not to use these.
Modifiers: Try to find segments that perform differently from the average traffic and use the appropriate bid modifiers. Beware of using too many with too high deltas.
Test tCPA: Target CPA is able to use many more second party data points to improve bidding. Generally this out performs manual bidding and potentially can drastically improve performance.
Negatives: Exclude irrelevant queries. True broad needs extra attention.
Keyword Expansion: Test true broad. Promote high volume queries into keywords. Look at Spyfu, KW planner and dynamic search ad campaigns. When you look at top queries the majority of the high volume should be added.
Data Density: Do you have enough data to make decisions? If not then aggregate your numbers in either time, theme or targeting. Additionally, you can start optimizing to proxy metrics like add to cart, quality score, time on site or key page views.
True Value Measurement: For brand, remarketing and view throughs we need to ask ourselves if the purchase would have happened even if they had not seen an add. On/Off tests for brand are the best way to measure.
Diversify your ads by using different types of creative. You want diversity.
Create ads in video, carousel, static, dynamic. Test new ad units as the come out.
Facebook likes consolidation. Use automatic placements with all devices enabled. Verify device/placement performance afterward.
Customize ads to the placements when possible. Most important to include the vertical size within Instagram feed and stories.
Target CPA is an absolutely must.
Lookalikes are going to be your best performing targeting 90% of the time. Create a variety of different LALs to add to your prospecting.
Use Campaign Budget Optimizer with caps on ads ets. This will allow you to add many different types of targets but still have Facebook optimizing to the impression it thinks is more valuable.
If any ad set is not performing you can always lower its daily max budget. This lowers the CPCs.
Testing opening up your LALs or even “No Targeting” as an audience. Facebook’s bid algorithm will try to hunt down the best performing people anyways.
Test and learn is a powerful tool to understand your programs efficacy.
If you are not getting enough conversions per ad set either roll up your targeting or optimize to an up funnel conversion event.
Refresh your ads on a 2-week rolling schedule. If you do not have those assets you can just duplicate existing ads.
Right hand rail ads do not have a frequency cap.
Remarketing campaigns will always overreport view through conversions.
Warm colors contrast with the largely light-blue background of FB.
Use Organic ads that have positive engagement in your prospecting campaign. Make sure that the post has a link to your website.
Best targeting from a DR perspective are in-market, custom affinity and custom intent.
tCPA is great if you can get enough data.
Be sure to add placement exclusions on offensive and poor performing websites.
Use category exclusions for liberally.
User observation targeting to gather data
Responsive ads for display are generally strong performers
Watch out for your frequency. We don’t want buy all inventory on a given page.
Target ROAS has generally been a poor performer
Smart v. Reg
Always use an audience over 200k. Any smaller and the ads will not serve appropriately.
As of now the CPC bidding is better than the automated
Don’t lower your bids too much as this will lower the quality/CVR
CTR and CPCs are closely aligned. Optimizing toward CTR will give you better CPAs.
Test Lead gen campaigns, they often create cheap leads though with lower quality
Always look at demographic and company data. You can see if your targeting is incorrectly skewing into one type of consumer.
Exclude irrelevant industries, job titles, job functions, seniority
Have seen most success when segmenting campaigns by company size and seniority
Use more than 4 ads but swap the poor CTR variants
True Value and Incrementality
One of the strengths of digital marketing is the ability to quantify the efficacy of our campaigns though there are still blind spots that many agencies and in-house teams fail to address. This is largely because in finding true value often times the result is that we overcounted conversions. One way that we differentiate ourselves is to aggressively champion finding the most accurate numbers we can.
Main Issues with Measuring Incrementality
Incremental costs in bids when trying to maximize profit
Cross-channel conversion duplication
Strategies to test for Incrementality
Native tests. This this is the preferred answer. Facebook has “Test and Learn” and many of the retargeting third platforms also have them. The channel will take your targeting and exclude a portion of that targeting from seeing any of our ads. If they still convert without seeing our ads then we know that those conversions are not incremental. We take the difference between the the on and off to create the true uptick that we can associate to the ads. This is also valuable to measuring VTs.
On/Off. In this test turn the ads on and off to measure the numbers when the program is live vs. paused. This will be most commonly used with Brand tests. When the ads are off we expect that organic will increased. The question then becomes how much does it improve. On(Brand +Organic)-Off(Organic) = Incremental Conversions.
PSA. For a public service announcement test we would add a “fake ad” usually American Red Cross into our targeting to see how many conversions it receives. There are usually issues stemming from poor quality score/relevancy score. Most clients will resist paying for another companies ads.
Geo-testing. Here we take two similar groups of DMAs and expose the experiment side to the variable you want to test. For example, how does San Diego perform with Retargeting vs. Memphis without retargeting. Plethora of issues here but the main on is that there are always local trends that are impossible to isolate and to get statistically significant data we would need to spend aggressively and run the test for a long time.
Experiments. Adwords and Facebook both have native tools to introduce variation. This is particularly useful when tracking the incremental costs of increased bids in Adwords.
Advanced Blind Spots
Measuring Halo effect
Struggling with Data Density
Local v. Global Optimization