My first day of Facebook ads
As I mentioned in my previous post, this week I’m focusing on finding interested visitors for my FeedSubscription.com. — And yes, I did ask ChatGPT for advice on this. 🙃
Since I’ve used Facebook ads before, I though I’d do a quick experiment and see what comes out. So here is my ad setup:
- Goal: “Get more website visitors”
- Total budget: $5.00 a day x 2 days.
- Location: Worldwide
- Age: 18 - 65+
- People who match: Interest: Blogs
Yes, I have messed up the audience ☝️ criteria: instead of “Interest: Blogs” I should have set it to “Job title: Blogger.” Fixed it after the first 15 hours.
Now, as I do every morning, I’ve looked at my tracking to see what’s happening. For historical reasons, I have 3 tracking systems at the moment which run in parallel:
- G Analytics — no custom tracking, just plugged in the script; I know it can do much more, but I only needed visitor count at the moment.
- access.log + GoAccess, aka “access-report” — this and G Analytics is to get a birds-eye view: how many visitors I get per day, and how does that evolve over time;
- in-app interaction tracking, aka “tracking-report” — this is the most reliable and clear for me because it allows me to track specific user actions: page loads, scrolls, clicks on CTA elements, or other interesting UI elements like navigation, and filling in forms.
It’s true that they are different in kind, but having all of them at the moment allows me to triangulate issues, and seeing the same numbers from three perspectives gives me confidence that they are real.
So, most of the entries in tracking-report yesterday had URLs with ?fbclid=something or having the referrer of (m|l|ml).facebook.com (← that’s a regex), which is perfectly expected given the launch of the Facebook ad.
Besides me messing up the audience, I’m a bit disappointed from the results so far: I have 193 unique devices although Facebook Ad Center says “469 click links”. I understand that, technically, there is some “distance” between click in Facebook and landing on the website, but it still seems too much of a discrepancy between the numbers.
Now, because I have the most trust in my own tracking-report, every morning I’m looking at the events recorded on the day before. If something looks fishy or unclear, I cross-check with access.log: look at what request have been made from that IP, what’s the user-agent string. — If you’re wondering, yes, it does takes me some time, but, as I already mentioned, I need to get some sense of clarity about what is happening. 🙂
Today I saw quite a few fishy patterns:
- Fishy homepage-only load patterns:
- lots of single load event, with no other interaction;
- 2, 3, or 4 load events, seconds, minutes, or tens of minutes apart, with no other interaction.
- Fishy registration form submission:
- repeated click the submit button without completing the email and password first;
- the form was being GET-submitted, as if there is no onsubmit event handler; — To ensure my website is not broken, I tried to reproduce this ☝️ behavior with the in-app browser of the iOS Facebook app, but the form validation still triggers properly in that scenario, so the form is not submitted.
- Fishy navbar clicks: a human can’t actually click on the element (I’ve tried!!) because it’s a container for the navbar elements, so clicking the navbar would record a click on the specific thing clicked, e.g. logo, mobile menu toggle, specific link clicked, etc.
I consider these patterns fishy because I’m doing this kind of inspection daily for the last few weeks and I can already tell how a human interacts, even a confused one. 😁
I should add that some visitors were human-like judging by their interactions patterns, but not as many as I expected: somewhere around 15–20%.
All in all, from the 193 devices, I got no signups. I understand that part of this is my mistake with the audience, and also I’m guessing some of this activity is legit Facebook bot activity like taking screenshots of the website or checking that it exists and isn’t down.
So, I’ll give it some more time because there is still more than one day left for the add to run, but I’m curious if anyone here has seen similar results in terms of behavior, or really, any info related to Facebook advertising would be appreciated. 🙂
