Home » Articles » The Great (Facebook Ads Strategy) Revamp

The Great (Facebook Ads Strategy) Revamp

November 9, 2021

Firstly, if you’re a client of mine whom I’ve not reached out to and explained all that follows, it means that your account is not affected by this phenomenon – so don’t stress over it. This blog posts is going to explain some things about how I’m trying to get ahead of an issue I’m not entirely sure is even an actual issue yet – this is me being preemptive, that’s all.

What people pay me to do (mostly)

I run Facebook Ads for podcasters – usually corporate podcasters – and this is how I do it:

  1. I setup Facebook Pixels on their relevant online entities (website, newsletters, link builders, whatever they’ve got), and I let that data accumulate for 30-days.
  2. On day 31 I create a Look-a-Like Audience (LALA) from that data so that I can use it in that client’s ad targeting.
  3. I create a very simple image carousal ad, or <30-second video ad.
  4. I further limit the delivery of that ad to iOS devices and only when those iOS devices are connected to WiFi.
  5. I link the ad directly to their podcast on Apple Podcasts

For 30-days I run this ad to marginal success. There’s always something learned in the first 30-days that can be used to modify the targeting decided by the LALA. For example: women may convert on the ad more often and for an average for $0.48/click while men convert less often and for a greater cost-per-click. In the 30-days this initial ad build runs, I can figure things like this out. On day 31 (now a full 60-ish days into working with the client) I make these modifications and build a new ad with the lessons learned.

This new ad runs for 90-days, on a specific schedule (every weekday during daylight hours), uninterrupted.

Typical Results

A podcast with, for example, 10 downloads per episode before the campaign, would perhaps see something like 1000 clicks in the first 30-days and pay something like $0.50 – $1.25 per click (the industry/niche varies the cpc). Of those, in this example 1000 clicks, the client might see an increase from 10 downloads per month to 400-800.

I was never able to connect click to listener in a 1:1 way, so I would attribute all growth to the ad campaign unless the growth exceeded the number of clicks received from the ad campaign. This was, at the time, the most fair way I knew of measuring the success of the campaign. I would also, in the reports submitted to clients, explain this limitation and suggest the growth attributed to the ad campaign if only 50% of the new listens were a direct result of the campaign.

At the end of 90-days we’d pull the plug for 30-days and see what % of the average listenership of those 90-days, stuck around in the absence of ads. It was usually between 60-80%. My clients, on average, pay something between $0.80 and $1.30 per new “retained” listener. So, 10,000 listeners (hypothetically) could cost them $8,000-$13,000 to generate and would *usually* take between 8-12 months.

I charge my clients $1800 a month for this service and so, all in, a company (like I said, I’m usually working with corporate) is spending, over the course of a year, between (on average) $29,600 and $34,600. The cost of a low-wage full-time employee (less the benefits).

I’ve never lost a client.

The Evo Experiment +1

For those of you following along with the Evo Terra experiment, you know that it appeared to be a failure. I have my own suspicions for why that is, and I voiced them in a previous post, but essentially: Evo’s podcast has a high churn rate and isn’t particularly appealing to younger listeners.

But, that not withstanding, and assuming I’m completely wrong, something else happened during this experiment that I’d never seen before: Evo received 500 link clicks over the course of 70-ish days but experienced negative growth in his audience.

A few days later, an individual who is not a client, but whom I’m helping with their Facebook ads in a pro bono capacity, reported the same thing – to the staggering figure of 50% less!

Both of these shows have reasons for lower margins of success, but not negative margins. I raised an eyebrow and I decided that, lest I worry for no reason, I needed to figure out if these two outcomes are outliers or if they’re indicative of a looming problem on the horizon.

Hi, clients 👋 Quick reminder: This isn’t affecting you at the moment, and there’s no indication that it will. I’m being proactive and preemptive here. If you have questions, just email me. Thanks!

Podcasting Sucks Experiment v1

In order to prove my suspicion that the Podcasting Pontifications experiment (Evo Terra) failed because it has a high churn rate and/or because it doesn’t appeal to younger listeners, I need a comprable (in content) podcast to test it against. I also need that podcast not to be inconsistent in their content, and to listen to everything I instruct them to do.

Wait a minute! I’m a podcast for podcasters! I’m consistent, and I always listen to what I tell me to do. Perfect!

So starting today I’ll be running the following ad (but differently than I did for Evo):

Simple ad, nothing special. It started running about an hour ago. Expectations are clearly set by the copy, the creative would surely speak to a podcaster, and the “Listen Now” CTA communicates that clicking this link will take you to a podcast.

The ad is targeting a LALA based on the 10,000+ people who have visited and interacted with my podcast course listings on AppSumo.com (yes, AppSumo let me put a pixel on my pages – thanks team!). Nothing else, when compared to how I run ads for any other client, is different. Except

I’m now using Chartable Smart Links to close the knowledge gap on clicks:listens

Remember before, when I said I was assuming the growth was a direct result of the ads so long as the growth didn’t exceed the number of clicks? Well with Chartable Smart Links, I no longer need to make that assumption. Listen to Episode 75 to understand why.

Just like I would do for any new client, I’m going to let this run, as is, for 30-days. I have my budget capped at $5/day, and delivery is scheduled to occur Monday through Friday during waking hours (weekends don’t convert well, people are busy doing other stuff). In 30-days I’ll share the data and I’ll make changes, then I’ll let it run for 90-days.

Below is the day 0+1 data from Chartable and includes only Facebook’s own testing of the link (which they do during the approval process for an ad), my testing of the link, and one genuine link click that occurred before I remembered to schedule the ad not to start until today (whoops).

Screenshot from Chartable on November 9th, 2021 @720AM US Mountain Standard

Also, for the month of October, I received just shy of 900 total downloads. I was with Captivate for that month, and, according to them, that 900 equated to 397 unique listeners:

There’s a small gap in the data between October 30th and November 3rd because I was migrating. I’ve been with Timber for a few days and here’s where I’m at for November (it’s the 9th) so far:

So, when I revisit this data in 30-days (December 10th, 2021) I will exclude data collected on and before November 9th (today) because my testing spanned into the 9th. That means I’ll exclude any data from the day I’m writing this post.

We’ll see where we’re at in 30-days.

Talk soon.

Photo by Hal Gatewood on Unsplash