Hacker News new | past | comments | ask | show | jobs | submit login

"Never listen to a google ads rep if they call."

Excellent advice. I was helping manage an Adwords campaign for an online shop, only had search ads turned on, was carefully watching and tweaking ads and managed to get a pretty decent CTR which had a big impact on cost per click.

I was in a different time zone (15 hours ahead) than the company. I went to bed one night and next morning I checked and found that our CTR had gone through the floor, was way less than 1%. And as a result our cost per click had sky rocketed from 10 to 15 cents to several dollars.

Turns out that a Google ads rep had called the office and convinced someone to turn on display ads, which killed our CTR. Our advertising budget was depleted, our Adwords account had received the dreaded low CTR slap which meant it would be very costly/impossible to get CTR back to a reasonable level. We were done with Adwords.

Things may have changed since then. It was years ago and I haven't touched Adwords since.




> which meant it would be very costly/impossible to get CTR back to a reasonable level

Sorry - is there some kind of hysteresis? You can't revert the setting and get back to exactly where you started?


Once you have thousands of ad impressions with very few clicks, it takes thousands of new impressions at a very high CTR to get your overall account's CTR back to something reasonable. And now that your cost per click has skyrocketed, even if you can get a high CTR for the new impressions your cost has skyrocketed.


That’s a sampling bias. You are confusing how your metrics are affected by the adverse event with how your metrics are now.


No confusion. I've been through this with several accounts I helped manage. Once an account has a history of low CTR the CPC goes up. Starting a new campaign doesn't help; the CPC stays high until you can get the CTR back up. And that costs dearly.

But again, that was years ago. Things may have changed.


Unfortunately, things have not changed. You are 100% correct. If anything, things have gotten worse, because now we have even fewer options to manage accounts.


So there's no way to set over which period the metrics are calculated?

Or is it a thing where the client only cares about the default metrics calculated over the full history of an account?


It’s on Google’s end. “I have inventory (a placement) to sell to an advertiser. I can sell it to X with a CTR of (10% or 1%) and a CPC of $0.50 or to Y with a CTR of 3% and a CPC of $1.”

Whether X’s CTR is 10% (what it used to be) or 1% (what it is now) drives the decision of which ad is better to serve in that inventory slot.


There are some misconceptions here (i.e. display CTR effects search ads quality score)

Here are some direct from the source help pages that might clear up your understanding of the system. https://support.google.com/google-ads/answer/156066 https://support.google.com/google-ads/answer/6167118


The GP does not appear to have any misconceptions.

From your first link, "how relevant your ad text is to searches, how likely people are to click your ad, and the quality of their experience once they reach your landing page"

This is all data that's generated based on the overall learning that Adwords does against your site presence. It's why it takes weeks to really get your data settled in, and why if you (or your Google rep) start to mismanage your account, things will rapidly go bad.


The GP story was based on the premise that opting a campaign into the display network increased CPC on search ad clicks. I know for a fact that it is not possible. I am not at liberty to go through each issue the GP had, so I linked to the best source of truth on this subject.


The effect of lowered account-wide CTR on CPC is something that has been experienced by a lot of people who have run Adwords campaigns. I saw it discussed a lot on forums about affiliate marketing. They always said, whatever you do, avoid the "account slap" that comes if your CTR plummets. I saw it myself on more than one account.

Maybe it isn't causal? But the correlation is pretty near 100%.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: