Home > Paid Traffic Sources > Facebook & Instagram

---

Facebook Troubleshooting -- Campaign so close to scaling big! (12)

---

I'm having some facebook results that are difficult to control/predict. Some days the CPA looks great, then some days the campaign CPA is terrible. Looking for advice on my current setup.

I've got an offer going where I've narrowed the targeting currently down mainly using the following targeting techniques:

age - 45-54
gender - female
relationship - status in relationship
Geo - all us (its been an even spread of states)
Device - all (but mainly mobile comes in)
Placements - FB newsfeed only
Bidding - Conversion Pixel CPA bidding - $30-40 target CPA
Overall estimated audience size - 1.8MM

Attached is performance data for this month segmented by day. Please take a look before you read below commentary.

So I'd had some previous conversions and demo data that led me to target a certain segment. The campaign is immediately making me money from 8/1 to 8/2. Great CTR, great CPA, looks awesome! Then after that it dies off. CPA goes way up, reach goes down. Then yesterday it decides to fire back up again. Does excellent. Then this morning its done the worst its ever done. I'm having trouble understanding the wild performance this pixel is putting me through. Does this data look weird to anyone or is this pretty normal when relying on the conversion pixel? I was hoping this pixel would get more data and get MORE precise with targeting, but its kinda gone the opposite direction. I'm also wondering if the conversion pixel is still trying out random audiences at its own discretion? Like on 8/11 it spent $300+ on an audience segment that didn't work at all...Thoughts?

---

I don't yet have enough experience with FB to answer this question - but thought I would reply to give this thread a bump to get more eyeballs - hopefully from FB experts that could shed some light on your problem.

One thought: If you're suspecting the pixel of being the culprit: Have you tried to NOT use the pixel, and run a traffic camp instead of optimizing for conversions?



Amy

---

I have often seen wild swings like this when using the FB pixel. I wanted to "rely" on the pixel to help with targeting, but I haven't found a consistent formula just yet. Here are a few questions for you, just to help clarify your situation:

1. It sounds like you are not adjusting the daily or overall campaign budget. Is this correct? (I've seen wild swings in results when trying to scale too fast.)

2. Are the desktop and mobile targeting in the same ad group? It might have separated the 2, but at this point since your campaign has been running for a while I'm not sure if it's a good idea.

3. You show 11 days of campaign data. Is that the total length of the campaign? I have never seen good pixel behavior without 200 or 300 results (conversions). But I sell a lower priced product, so it may be harder for you to get this amount.

Sorry I don't have any magic answers for you, but these are just things I'd look at if I use my experience. Caveat: I am still in the not profitable stage, some days great, other days poor. But I've "invested" a lot of money in my education.

---

If you're optimizing for a purchase event directly, then you might want to have at least 100 or so conversions per week in order for the pixel to have any decent effect on this. If it keeps being this inconsistent I'd swap it over to add to carts until the pixel has enough data and then attempt to switch to purchases.

---

@amy

I have tried some CPC campaigns before, but since I had a larger audience this round I decided to let the pixel do the work.

Next on my list of things to do is to duplicate this campaign with a CPC bid if no better ideas come of this thread.

---

1. Yes, I made some budget & bid adjustments across the date range. In hopes of trying to fire it back up I brought CPA to $40 and budget from $400 to $1000. The $400-1000 daily budget adjustment happened in $200 increments across a few days.

2. Yes same ad group. 90% of traffic is mobile so I'm not too caught up on separating it out.

3. This pixel has seen 145 conversions over the last 2 months. I launched some smaller versions of this campaign before with interest targeting & took insights from that to target a larger group (which we're discussing now).

---

I think I know what you're getting at. Optimize 1 more step up the funnel to get more data points going. I've considered doing this myself, but the problem is that my next point up the funnel metrics don't correlate well with conversions unfortunately. Check out the screenshot I've attached. You can see Voluum CTRs from lander to offer and its pretty steady despite wildly changing conversions. Basically it appears CTR to offer does not correlate well with conversions.

Attachment 16321

---

@gettinleadz, I'm very interested in hearing if you find any reasonable solutions, so please keep us posted. I personally only scale in 20% increments every 3 or 4 days, after some hard lessons on trying to scale by dollar amounts daily. But even this produces inconsistent results, albeit more predictable than scaling by just dollar amount.

---

An update to this thread.

The CPA campaign mentioned above naturally just turned itself off.

Next up I launched a CPC campaign with a 1% lookalike audience to test if the conversion pixel/optimizer was the problem.

Much like the CPA campaign, the CPC campaign (with 1% lookalike based on previous converters) came out crushing it. Over 100% ROI and maxing out current (small) budgets. Then, within 5 days it crashed again. My ads were still getting CTRs at 10%+, just people actually converting died off. CPA went way up as you can see below:



I paused the campaign on the final day seeing it was going to repeat the previous trend.

Total audience size for this 1% lookalike is 1,500,000 people. In the above image you can see I only reached 250,000 before the campaign died. It doesn't seem to be an audience issue to me?

A possible problem with this is that I'm trying to vertically scale budgets too fast. In the above image I've scaled spend from $200/day to $700/day over the course of 3-4 days. But that doesn't really explain why CPA was fine at 8/21 spending $700, then just gets worse. If I scaled to $700/day too fast then I would've thought the first day spending that would be wild, but it spends $700 for 2 days before the campaign dies.

CPC went up from .05 to .08 as I scaled budget.

Any thoughts?

---

What are you optimizing for?

If you are optimizing for purchases, and the campaign dies after showing it to 250,000 out of 1.5 million, that mean's FB showed it to the people most likely to buy (in that 250k) and after that it starts falling off because the rest of the audience, aren't such vivid shoppers.

What you should do is refresh your creatives; make creatives that are different, change the video main thumbnail to something else. You will see more success this way.

---

@Attila

Thanks for reply.

Optimizing for sales as conversions.

Regarding audience, I am surprised to see a 1% lookalike on converters is "maxed out" after hitting 250k of 1.5M.

I have not refreshed crwatives this whole time.

---

@gettingleadz: any updates on this? I'm running into the same issue now where the first few days are amazing, converting extremely well and then all of a sudden, conversions slow way down. Similar audience size as well and optimizing towards conversions.

---

Home > Paid Traffic Sources > Facebook & Instagram