Question

I was running a nationwide campaign. I took the top 15 cities that were performing the best and getting the cheapest leads (all below 70 cents) into a new campaign. Now leads are costing double or triple on average. Why is that?  

Answer 1

How long did the old campaign run? How long has this one run?  

Comments

@citizen3 about 2 weeks on nationwide campaign, about 3 days on targeted campaign.  

---

Three days isn’t really that long. And I’m assuming the audiences is smaller. I’d give it a little more time. How many leads did you get from the nationwide campaign? Why not try a lookalike of of those leads instead of messing with geography?  

---

So three days. did you launch the campaign on a Friday or Saturday?  

---

@citizen3 For a reason with my sales team, certain cities are assigned to certain sales people. But yes that makes sense - weekend could be worse

Answer 2

The first person that sees your ad sets the path for the rest of that campaigns life. It was going good when you showed Jeff from Chicago that first ad. Then you started over and Kenny saw the first Ad. Different paths, you messed with the system when you thought you knew.  

Comments

@ader748 Thanksgart Makes sense! I appreciate the insight. So anytime I start a new adset/audience it’s basically luck of the draw? And going forward that determines the results of the rest of the campaign?  

---

@burlington126 I honestly can’t tell you that I know but it kinda makes sense from what I’ve seen. It’s not exact as you’d think, there are humans involved after all  

---

@ader748 Thanksgart for sure! I totally agree as much as we can run the numbers on our end of things we can’t actually control humans to take action (as persuasive as we might be ) lol

---

@filmer bring something more than that if you’re gonna call it nonsense.  

---

@filmer the first person has a big effect, the first, as in the first friend you had in school. The first girlfriend you ever had. Ads are “if this - then that” Hell life is that way. When people duplicate an ad set and then see different results often times it’s the “new path” that the ad was out on with billions of different “if this then that” decisions being made. First guy in the first go round had a red hat. Second round guy had a blue hat. Red hat guy bought a product so I think maybe we’ll look at the red hat wearing folks and try to sell to them. Whoops round two the blue hat guy bought the product let’s go look at maybe blue guys. LLA are done the same way they try to figure out combinations of attributes that made someone do what you wanted them to do. Much more complex than a group setting like this can facilitate. For the sake of my sanity (whatever is left) I’ll concede that you are right, I’m wrong, my thoughts are complete nonsense, I’ll go back to the little kids table where I belong. Have a great rest of your week.  

Answer 3

Have you switched off the first campaign or removed the top 15 cities from it ad set? Both campaigns could be competing against each other and driving up the price in the auction.  

Comments

@saudra Yes the nationwide campaign is off. They are not competing with eachother.  

Answer 4

I have had the exact same thing happen, must just be something about FB rewarding broader targeting with cheaper conversions

Answer 5

Did you start with a high budget in a learning phase? I would tighten up the spend until FB algo catches on again.  

Comments

@somewhere about $20/day

---

@burlington126 How long did you give it?  

---

@somewhere original campaign ran for about 2 weeks, new campaign has been running for about 3 days. I’m not worried or anything just wondering why it happens and if anything I can do on my end or should know about.  

---

@burlington126 be patient!  

Answer 6

Man. all of what the guys in here wrote is valuable. However one law of business is: when you scale, whatever you scale, your margin will ALWAYS feel a downward pressure

Comments

@saporific can you explain what you mean?  

---

I mean that if you are doing math with 10-50 leads a day. the cost per lead is easily increased if you want to do 100-500 a day when you scale any sistem (not only fb ads) it is very likely that you'll see an increase in complexity = costs

Answer 7

Its just what happens allways. Sometimes its even best to run it as broad as possible as lead cost from your regions will be cheaper overall than to target only areas you need. Not sure why it is, but fb does that.  

Answer 8

Ads saturation, really common

Answer 9

So much nonsense here. Don’t worry about learning phase, just do what you need to. It’s about your budget to audience size ratio (mediaweight) = relative cost for that audience. Audience quality determines baseline cost and how much budget/audience size ratio fb can support at your CPA. So broad audiences that are low quality can still support some conversions as long as your media weight is low. Fb does fine auto optimizing across regions. It won’t look perfect but it’s better than what you can do manually. If you adjust your budget down proportionally to how much audience you trimmed down you’ll have relatively same costs or better costs if you trimmed down to a higher quality audience. If you keep the budget same then your media weight goes up by more than the audience quality and that’s why your costs spike.  

Answer 10

When you restricted the campaign to the top cities, it became more specific, and fb charges more for that. - that's just my hypothesis. I've tried doing the same optimization, but on, say, age, gender, or mobile vs desktop, and you'd think it would lower costs, but not really.  

Answer 11

Generally what you need to do is run a nation wide campaign with many ad sets each targeting a smaller geographic area. That way you can turn off loosers and concentrate on the winners.  

Answer 12

If it were that easy to do FB ads, everyone would do that. Unfortunately, decreasing the scale that much gives less opportunities for your ad to be served overall which means it will cost more to deliver. Try the opposite. Keep the national campaign, but exclude places (or demo's, placements) with the worst performance.