Negative Keywords (in google ads)

[LarryATX]

Hello,

Curious if anyone has thoughts on this idea/topic:

When should you negative a term? What to negative?

1. you just know it's not a good term
2. you see clicks but no conversions
3. you see clicks, but how many clicks is a valid amount of clicks before you negative
4. you see no conversions, but (what if) the conversions are tracking correctly on purpose or accident? (meaning, some logical reason why conversion are not tracked, now showing)
5. you see way too many impressions and not enough valid engagements
6. do you add in all the various "what if" topics (seasonal, website issues, etc)

YES, the data (number of terms), can be overwhelming.

my point of this post, question:
what is your strategy for negatives?

A) just negative whatever looks good and move on... Be Fast. Think Less. Use a tool, don't look back. Just move forward fast.

B) carefully look for negatives in the top X amount of terms with the most clicks or most money spent. After that top list is processed, execute item A above.

C) carefully look at all clicks and come up with valid logical reasons why terms have lots or few clicks.

D) Look at impressions only, as that is the beginning of wasted clicks. you can prevent wasted clicks by viewing the impressions. ==>YES, this is time consuming.


FINAL QUESTION
have you negatived something and then realized later, it was a mistake? ==> if so, what what the issue or reason or situation?

I'm trying to have an open discussion on negatives and a process to implement it well.

Thank You for the feedback, input!

-Larry

 

[eliquid]

Even though this sounds like a basic topic and question, this is one of the topics where having someone that has done this for years really matters.

Because what you put into the negatives, can have long standing impacts on your campaigns.

I typically identify the terms that are 2-3x over my allowable CPA, or way under my ROAS goal... that have been under/over goal for at a min of 90 days. They ALSO have to be over $XXX dollars of spend.

It has to match all 3 conditions ( goal, time, money spent ).

If you do it any other way, I can guarantee you issues later down the road somewhere.

At some point, you will want to run this for the last 9 months too, to find the "ankle biters" ( the terms that have been under $XXX dollars of spend for the 90 day period you have been looking at ) and kill those off too.

^^ that would still follow the goal + time + money spent rule ( all 3 needed ), you are just changing the time to last 9 months to catch those that slipped prior because they were always under the money spent threshold for 90 days.

 

[LarryATX]

I typically identify the terms that are 2-3x over my allowable CPA, or way under my ROAS goal... that have been under/over goal for at a min of 90 days. They ALSO have to be over $XXX dollars of spend.

It has to match all 3 conditions ( goal, time, money spent ).

Thanks eliquid!
I can agree with your reply and rule/match conditions. (and even tweaking it a little depending on account or specific needs).

Question: do you let your rule override the thought process?

Meaning, what if a perfect search term, (call it perfectXYZ), or at least it seems perfect. It seems or looks very similar or just like other terms that convert well., BUT perfectXYZ fails the rule.

Do you negative it anyway?
Just let the rule be the deciding factor? Is it YES or NO, don't let logic or bias or opinions get in the way?

thanks

 

[MustLoveDogs72]

Thanks eliquid!
I can agree with your reply and rule/match conditions. (and even tweaking it a little depending on account or specific needs).

Question: do you let your rule override the thought process?

Meaning, what if a perfect search term, (call it perfectXYZ), or at least it seems perfect. It seems or looks very similar or just like other terms that convert well., BUT perfectXYZ fails the rule.

Do you negative it anyway?
Just let the rule be the deciding factor? Is it YES or NO, don't let logic or bias or opinions get in the way?

thanks

eliquid gave solid advice. It's pretty much what I do, as well.

Another thing I pay attention to is how many clicks the term received. Depending on your site's average conversion rate, you want to give each term a shot at converting. If your typical CVR is 1%, 50 clicks isn't enough data to determine that the term won't work. You want to balance this with all of the factors eliquid mentioned.

As far as the perfect search term scenario you mentioned, if it's a perfect search term, it will convert. I don't let my gut make decisions, I let the data speak. If you follow the schedule outlined above (and have a decent budget), you'll find that for most terms, that's sufficient time to determine your winners and losers.

 

[eliquid]

Thanks eliquid!
I can agree with your reply and rule/match conditions. (and even tweaking it a little depending on account or specific needs).

Question: do you let your rule override the thought process?

Meaning, what if a perfect search term, (call it perfectXYZ), or at least it seems perfect. It seems or looks very similar or just like other terms that convert well., BUT perfectXYZ fails the rule.

Do you negative it anyway?
Just let the rule be the deciding factor? Is it YES or NO, don't let logic or bias or opinions get in the way?

thanks

The rule kills it, unless someone higher up the chain ( the owner of the company ) tells me for XYZ reason that terms needs to run.

For example, a lot of your competitor terms ( if you bid on competitor names ) will fail this test. Like 8 out of 10 times.

But sometimes these brands and owners just want to F over the competitor. They are willing to lose money over it. SO I let it be.

Otherwise, it can be a perfect term that is the same term that is converting.. an example would be the an Exact Match of termABC doing well and fine, but Phrase Match of termABC is not. The one not doing well even though its the same damn term ( just different match ) is getting killed if it is doing bad.

There is a bit more behind this strategy and concept, that goes really F'ing deep. But most people can't handle that.. so I will keep it high level for now. But sometimes you don't even want to kill these low performing terms off, but that's only if you know things like Market Awareness level and LTV/AOV hacks. If you dont know these, stick with what I wrote above.

Once you get to grand master level, you will need to start adding in other things to truly know if you need to kill a low performer.

Of note, most of the terms that drive in new customers, are low performers. But you won't know that unless you know Market Awareness levels, LTV, LTV cycles, and AOV numbers to go along with your ROAS and CPA numbers. You really need to hone in on New Customer CPA and New Customer ROAS, but most people's head explodes when I go into that.

 

[LarryATX]

ONE:
The rule kills it, unless someone higher up the chain ( the owner of the company ) tells me for XYZ reason that terms needs to run.

TWO:
There is a bit more behind this strategy and concept, that goes really F'ing deep. But most people can't handle that.. so I will keep it high level for now.

THREE:
Of note, most of the terms that drive in new customers, are low performers. But you won't know that unless you know Market Awareness levels, LTV, LTV cycles, and AOV numbers to go along with your ROAS and CPA numbers.

Thanks eliquid, very much appreciate the discussion. I've tried to have this type of discussion on other "ppc places"... its all crickets.

ONE: i'll start doing this, let the rule kill it.

TWO: I get the idea/understand this goes deep, as many items (business, strategy, ad account) ALL go into the decision. This brings up this topic: attribution sucks or is getting worse as of late??
I'm assuming you are somehow factoring that into the equation? Either by feel/your knowledge WITH bottom line numbers (did company grow/sell more this month, etc).

THREE:
I realize lots depends on what IS the business, its goals, and what is the market. There is huge different strategy comparing, for example: commodity online ecom business vs a service company with specific lead gen

I guess at this point, for the simple thing: let the rule kill the term, based on simple data... then compare months of the bottom line and, obviously, the conversion stats.


ONE new thought:
do you have custom tool or scripts or how do you deal with this issue:
= you look at actual search terms (whether you are bidding on exact, phrase, broad).
= a term could appear in any campaign/ad group
= then you see a term appear and it fails the rule in a campaign or ad group. GREAT. you will negative it.
==> then you see the SAME term appear in another campaign/ad group and it is performing well.

being able to see ALL terms regardless of ad group. how do you do that? your own scripts?

thanks!

Another thing I pay attention to is how many clicks the term received. Depending on your site's average conversion rate, you want to give each term a shot at converting. If your typical CVR is 1%, 50 clicks isn't enough data to determine that the term won't work. You want to balance this with all of the factors eliquid mentioned.

As far as the perfect search term scenario you mentioned, if it's a perfect search term, it will convert. I don't let my gut make decisions, I let the data speak. If you follow the schedule outlined above (and have a decent budget), you'll find that for most terms, that's sufficient time to determine your winners and losers.

Thank you.
I'll just trust the data more
and make sure to compare apples to apples

curious, about your thoughts on "attribution getting worse"?

example:
started a new account on a business with a couple years of review data.
the new campaign is showing conversions in the ad data, but looking at the company reports,
revenue is growing much more than the extra revenue shown in ads reporting.

I even compared same time period to prior years/months... the business is increasing revenue and ads is the only new thing added.

IMO, that tells me: the ad campaign is working, but attribution is missing conversions.
it isn't a coincidence that sales increased 50% at the same time the ad campaign started, but ad campaign reporting is only showing 20% increase.

then how to optimize keywords you can't see conversion?
I have ideas, but they are not efficient

 

[eliquid]

This brings up this topic: attribution sucks or is getting worse as of late??
I'm assuming you are somehow factoring that into the equation? Either by feel/your knowledge WITH bottom line numbers (did company grow/sell more this month, etc).

Attribution always sucks, and yes it is getting worse.

iOS, Apple mail, GA4, cookies, chrome privacy, etc....

I don't factor it into the equation. It is what it is. There is no way to get accurate attribution.

Clients don't wanna hear this, but it is what it is. You just try to make the best estimate on multiple models of attribution and blend it to get your real number.

But that means pulling a lot of reports on different time frames, models, etc and coming up with what the attribution has been and will prob. stay.

You can do studies into "lifts" and gain some info that way too.

ONE new thought:
do you have custom tool or scripts or how do you deal with this issue:
= you look at actual search terms (whether you are bidding on exact, phrase, broad).
= a term could appear in any campaign/ad group
= then you see a term appear and it fails the rule in a campaign or ad group. GREAT. you will negative it.
==> then you see the SAME term appear in another campaign/ad group and it is performing well.

being able to see ALL terms regardless of ad group. how do you do that? your own scripts?

I have a ton of custom scripts.

But your example, it's perfectly fine.

Why?

Because Google is going to do what Google wants to do. You can't tell it what to do anymore, you havent for a long time. Anything you put into your Google Ads account is just a suggestion.

So if you have 100 different terms in your account across 5 adgroups and 2 campaigns ( so 20 terms per adgroup, 50 per campaign ) and you have termABC get matched and it's bad and you negative it, but that same term shows up on the other campaign/adgroups and it's fine and good.. you leave that one alone.

It's not about having the perfect ad account setup and structure.

It's about making money.

And if termABC makes money in some random campaign and adgroup, but not another.. you only negative the one not making money.

And BTW, in 90 days.. that good term may not make money anymore. You still kill it when it fails.

This is why when other "guru's" and account reps scream about "your cannibalizing" I yawn and tell them to shut up. Even in SEO.

There is no such thing in PPC ( and not even in SEO ).

This is a much deeper discussion and people are going to try to prove me wrong, so I will just stop there.

the new campaign is showing conversions in the ad data, but looking at the company reports,
revenue is growing much more than the extra revenue shown in ads reporting.

I even compared same time period to prior years/months... the business is increasing revenue and ads is the only new thing added.

IMO, that tells me: the ad campaign is working, but attribution is missing conversions.
it isn't a coincidence that sales increased 50% at the same time the ad campaign started, but ad campaign reporting is only showing 20% increase.

There could be a ton a reasons for this, but mainly it could be:

1. Current and old customers are seeing the ad, not clicking, but purchasing. Thus you wont get credit/attribution

2. Attribution is messed up

3. You dont have your reporting set up right in Google Ads. Remember the default reporting for conversions is Last Click ( AND ALSO ) Date of click ( the conversion happened ) in the Google Ads UX. Date of click is not the same as date of conversion. So if someone clicked your ad 21 days ago and then sat on their hands for 21 days and made a conversion today, your internal company reports would show revenue up $100+ today, but your Google Ads reporting ( unless you changed it ) will not show that today, it will show up in a report from 21 days ago when you backdate the time frame and reporting... 21 days ago will show up as $100+ revenue in Google Ads, not today.

^^ Now magnify that for brands with longer sales cycles and you end up breaking the attribution window time limit and everything is a now broken, not credited, or credited to a new session, or lost click ID

 

[LarryATX]

Anything you put into your Google Ads account is just a suggestion.

It's not about having the perfect ad account setup and structure.

This is a much deeper discussion and people are going to try to prove me wrong, so I will just stop there.

I understand much, agree with all... AND love the above comments.
They made me think a little differently than before


I've been trying to have a conversation like this for a long time.
And it is more than obvious now, that I'm not using scripts like I should. I've used some here and there, use a tool for ngrams. Do you have a resource for getting scripts made or a concise learning spot on the web? Anything you can recommend?

thank you eliquid!!

 

[eliquid]

Do you have a resource for getting scripts made or a concise learning spot on the web? Anything you can recommend?

Only thing you can do is learn Javascript and Google Ads scripting, but I dont know a resource.

There are places out there that have scripts already, but I warn you:

1. Most are old and outdated, on a prior version of Google Ad scripting and wont work.
2. Most are not accurate if they do run, due to #1 above
3. Most are garbage to begin with, telling you things like your CTR. WTF

You can also find a script you think you like and want, find out it doesnt work, and then hire someone on Fiverr or Upwork to edit and modify it for you. I've done this a ton of times.