-
Serious question for PPC Experts
Posted by on February 14, 2026 at 3:48 amIf a campaign is performing well,
how often are you tweaking:Sub-Locations,
Age,
Gender.Do you optimise just because you can…
or only when the data tells you to?I’ve seen people over-touch campaigns that are printing money. I’ve also seen accounts plateau because no one challenged the targeting.
So what’s your rule?
• Leave it alone if CPA is stable?
• Test quarterly?
• Always push for incremental gains?Curious to hear how real operators think about this.
(Imaging you are in a running in a service based google accounts country wide)
-
1 Reply
-
In the words of noted PPC expert Johann Sebastian Bach, “If it ain’t baroque, don’t fix it.”
-
> (Imaging you are in a running in a service based google accounts country wide)
we’re going full circle i am being spoken to like I’m an AI agent
-
if it is performing well, why would you tamper with it at all??
-
The robot is almost always going to weigh those factors better than you can. Even if that was false, mostly people “optimize” for irrelevant data points like CTR when they pull those levers, especially in small accounts.
Service based, nationwide doesn’t make much sense unless you’re a saturated franchise or a lead seller with a broad customer base. I generally don’t buy outside the service area.
-
If it works you don’t change anything
-
I’m mostly on the search and programmatic side and yeah I would constantly be tinkering for improvement. Those small little 5% gains add up and then when you’re looking YoY suddenly you realize how far you’ve come.
-
You need to monitor impression share, rank lost due to budget, ad copy performance, and CPA changes on a weekly basis, and make adjustments aligned with your goals.
-
my rule is simple — don’t touch what’s working unless you have a hypothesis backed by data.
for demographics specifically (age, gender), i almost never proactively exclude on a well-performing campaign. google’s smart bidding is already adjusting bids by demographic segment behind the scenes. if you hard-exclude 65+ or whatever, you might be cutting off a segment that converts at a lower rate but still profitably. i’d rather let the algorithm bid down on weak segments than remove them entirely.
sub-locations are a different story though. geo performance can vary wildly and the algorithm doesn’t always catch it fast enough. i’ll pull a geo report monthly on anything spending real money. if a metro or region is consistently dragging down roas, that’s worth an exclusion or bid adjustment.
the over-touching problem is real and honestly more common than the “never looked at it” problem, at least in my experience. seen people kill campaigns by making 3-4 changes at once because they got anxious during a slow week. then you can’t even tell what caused the performance shift.
my general framework: if cpa is stable and within target, i check in weekly but only act on 2+ weeks of a clear trend. quarterly i’ll do a deeper audit on geo, device, and audience segments. the goal is to find pockets of waste, not to optimize for the sake of optimizing. -
If it’s printing money and CPA is stable then touching it is just gambling with working math… the people who overtinker profitable campaigns are usually bored or trying to justify their existence to clients.
Only time to optimize demographics or geos is when you have statistically significant data showing a segment is dragging down blended performance… and that usually takes 90+ days minimum to know for sure.
-
Post click—-optimize the lander, conversion funnel
-
If it is performance well, the age old adagium goes: don’t just do something, sit there.
-
if its working, i generally just talk about things during the client meeting and do nothing. We’ve made changes before and things can break. Its actually a feature of the system if you raise budgets, performance drops but people are too scared to wait for the figures to catch up so end up making only small adjustments.
-
First, bid adjustments do not work for locations, age, or gender unless you’re using manual or max clicks bidding. And if you’re using either of those methods you also certainly shouldn’t be changing them that often because presumably you have very little conversion data to work with.
That aside, you should make adjustments of any kind once you have statistically significant data to drive those changes. So, for example, if you want to consider excluding an age group make sure your sample size is big enough and you’re not doing it based on gut feeling.
But adjustments aren’t the only kind of tweaks. The most important tweaks are to creatives which should be pretty frequently, but again based on the data.
For example, if you’re running an ad variation test, wait until you have at least 90% (aggressive) or 95% (best practice) confidence before stopping that test.
It’s also good to experiment more broadly with dedicated budget. That might look like running a separate DemandGen campaign to fill the TOF and measure uplift, for example.
-
Just run tests. Unlike a lot of the other commenters, my clients don’t pay me to do nothing. Test ad copies. Test landing pages. Test new campaigns. Test ad extensions. Test anything.
I’ll also sometimes carry out routine hygiene tasks on campaigns that are working. I’m not too worried about breaking things.
Log in to reply.