Google Ads monitoring/script idea... competitor alerts

[LarryATX]

So, trying to have an open discussion if I'm going down a never-ending rabbit hole OR is there a viable, useful solution that can help an account keep performing well.

I was trying present this problem/idea to AI:

Let's say a google ads campaign and ad group are running for 12 months and are performing at a 100% ROI every month and without any changes that I made, the campaign dropped to 70% ROI without any seasonal changes, economy did change, no external macro changes. The only reason can be is the competition and competitors had made changes. What exactly and how would I see in these competitor metrics change, that can alert me or show signs of the drop in ROI before it actually happens?

NOTE, to keep things on topic and specific for this case:
1) The account, campaign, or ad group will not have any changes.
2) No need to worry about keywords or targeting, as the past year has been perfect, every month.
3) consumers are not changing their behaviors.

AI Output:

These examples will help illustrate how to interpret significant shifts in the metrics and what they might mean in terms of competitive dynamics.

1. Overlap Rate​

Metric: Overlap Rate
Example: Last month, your overlap rate with Competitor A was 35%, but this month it has increased to 50%.
Interpretation: This increase suggests that Competitor A's ads are now appearing in the same auctions as yours much more frequently than before. This could mean they have expanded their keyword list to include more of the keywords you are bidding on, or they have increased their budgets, pushing more aggressively into the markets you serve.

2. Outranking Share​

Metric: Outranking Share
Example: Previously, your outranking share against Competitor B was 60%, but it has now decreased to 40%.
Interpretation: A drop in this metric indicates that Competitor B’s ads are now appearing above yours more often than they used to. This might be due to Competitor B increasing their bid amounts or improving their ad quality scores, which enhances their ad rank.

3. Position Above Rate​

Metric: Position Above Rate
Example: You notice that the position above rate for Competitor C has gone up from 20% to 45% over the past few weeks.
Interpretation: This change implies that Competitor C’s ads are now more frequently displayed above your ads when both are shown. It could be a result of Competitor C employing higher bids or having enhanced the relevance and performance of their ads.

4. Search Lost IS (Rank)​

Metric: Search Lost IS (Rank)
Example: Your Search Lost IS (Rank) has increased from 10% to 25%.
Interpretation: An increase in this metric indicates that you are losing more impressions due to poor ad rank. This could be because competitors are improving their ads' relevance or bidding strategies, directly impacting your visibility and potential ROI.

5. Search Exact Match IS​

Metric: Search Exact Match IS
Example: Your Search Exact Match Impression Share has decreased from 75% to 55%.
Interpretation: This decrease suggests that competitors are possibly bidding more aggressively on the exact keywords you are targeting, or they have improved their ad quality for these keywords, resulting in them capturing more of these valuable impressions.

Monitoring Strategy​

To effectively monitor these changes, you could set up alerts in Google Ads to notify you when any of these metrics change by a predefined threshold, such as a 10% increase or decrease. Regularly reviewing these alerts and the detailed Auction Insights report will enable you to keep a pulse on competitive movements and their potential impact on your campaign performance.

These examples and interpretations should give you a clearer picture of how to use competitive metrics from Google Ads to preemptively identify shifts in the competitive landscape that might affect your campaign’s ROI.



=========
Back to my discussion, idea.
Has anyone tried to create a script to provide such alerts at an account, campaign or ad group level?
Did it work? Does it help?

OR maybe you think it is great idea and would work?
OR maybe you already know that this is great theory, but doesn't work in reality as too many variables.

thanks

 

[eliquid]

And this is the problem with thinking Ai can solve or replace people anytime soon.

You can set up alerts all day long, for weeks on end, to prompt you to these changes.

But it doesn't mean a competitor did something on their end, to outrank you.

If Google Ads algo changes up for XYZ reason, you could start getting less of everything that has nothing to do with your competitor doing something on their end and outdoing you.

The mere fact Google changed something, could cause all of that to happen while your competitors did 0.

Also, the economy changing and people heading into a recession is a big red flag and people are buying less. Just check out car sales right now as a indicator to the economy and people's fears.

1. consumers are not changing their behaviors - I see it across multiple large accounts
2. The account, campaign, or ad group will not have any changes. - Maybe not from you, but Google's own changes to your account via their algo will cause you to lose
3. No need to worry about keywords or targeting, as the past year has been perfect, every month. - and this is why doing nothing and thinking things will be the same is a bad plan. You will need to adapt because of #2 and #1 above.

You can't rest on your laurels.

This is why people like me ( by myself ) can run circles around a team of 30 combined SEM's. Why I can spend multiple millions weekly during a COVID lockdown and scale an account from $55m to $125m while everyone else is going bankrupt in the world and pulling back their ad spends.

You looking at this this the wrong way.

To answer your question, yes scripts like this are out there. I've created some on my own. They work on a % of change and email you, or show the change in a weekly report.

However, this doesn't mean the competitors did something.

Truth be known, just looking at your CPM/CPC.. ROAS.. CPA and internal numbers ( like NC CPA, etc ) can tell you everything.

Also of note, ROAS is about the worst metric to look at for ecom. It's a fake number to begin with, especially moreso if you have any other marketing channel ( like Direct, social, email, etc )

 

[LarryATX]

Hi @eliquid , great reply (again) Thank you.
I agree with everything and understand the reasoning, etc.

My "faults/false statements/ incorrect logic, etc", not that they are horrible, but rather it's a learning process? You try something and learn what is good and what is bad/false.


Yikes. where to start first? (note: I have several years of google ads experience, yet understand others know WAY more than I do.)

Okay, I've already started (meeting you!!!)
and I'm understanding (more now)...
==> changes ARE happening... faster than before, whether I see them or not. In reality, that has always been the case

==> I was thinking negative term list is a best place to start?
I have and tweaking a google ads ngram script, with the idea it is or can be tweaked for EACH account that it's run on. I know how to code and GPT is providing the lines of code for me:

The script processes Google Ads search query data to generate n-gram (unigram, bigram, trigram, quadgram) metrics and writes them to separate sheets in a Google Spreadsheet. It also calculates and displays the total impressions, clicks, cost, and all conversions at the top of each sheet.

Key Steps:​

1. Setup and Initialization:
   • Connect to the specified Google Spreadsheet.
   • Initialize separate sheets for unigrams, bigrams, trigrams, and quadgrams with appropriate headers.
   • Define a list of safe words to exclude certain search terms.
2. Fetch and Process Data:
   • Retrieve search query data from Google Ads for the specified date range.
   • For each search term:
     • Skip terms containing any safe words.
     • Generate unigrams, bigrams, trigrams, and quadgrams.
     • Aggregate metrics (impressions, clicks, cost, all conversions) for each n-gram.
3. Calculate Totals:
   • For each type of n-gram, calculate the total impressions, clicks, cost, and all conversions.
     
     

Tweaks per account:
- safe words
- time period
- account as a whole or specific campaigns/ads groups

Do you think this ngram script is a good direction to go? Good as in a valuable way to increase ROI.
If yes, a HUGE question I have, the rule(s) on how to decide: yes or no for negative?

- It is obvious for certain terms/impressions
- for others that are related/similar... maybe you just give it MORE data, then you will HAVE the answer

OT question:
i'm assuming you use all 3 match types, it just depends on the account and the goals?

thank you for your time!!!!!

 

[LarryATX]

with this script, I believe I can get it to do what I do manually.
But manually is hard, time-consuming, I miss things/make mistakes... AND no way can I compute the "rules" needed to provide the best or highest ROI possible.

Rules and/or realtime alerts like:
- when ROI falls X %
- a specific ad or ad group or product ROI falls X %
- specific search term or variations of a term, the ROI falls X %
- ABC term produces X% ROI and 123 term produces Y% ROI, now should you make a change???


is that pie-in-the-sky thinking OR you might have already done it?

if I'm correct, you have already done it

 

[eliquid]

I would not build a script that does this automatically.

What I mean is, build the script and have it output to google sheets, and then you go into the sheet and find the worst keywords and you negative match them manually.

If you try to automate the finding, selecting, and inserting.. you will end up regretting it at some point in the future.

 

[LarryATX]

oh.. true... that is what I'm working on. My goal is to NOT have it "automatically" add a negative word to the google ads list, BUT (my goal) is to have the script do all the "work" for me.

so I can see the candidates for what term "could" be a negative term.
show me the reasons how/why the term "should" be a negative term.

I agree, not completely automated, as something could and would go wrong

BUT I'm guessing, you agree, this type of script is a great thing to get working, correctly?

I extremely value your opinion and thoughts. The research and knowledge I've done...
tells me... you are one of the people I should listen to

thank you.

 

[LarryATX]

Don't argue.
Listen. He's so right its painful on this one.

Hi, sorry.
I didn't mean or try to come across as arguing. I 100% agree with eliquid. I was hoping to get some clarification on the prior post. I completely agree that an all-automated script will create problems.

I will ask for clarification like this. In the script, there are 4 parts:
1. only writing to a google sheet.
2. using some logic with real-time stats on organizing the list(s)
3. testing/knowing/using better rules will help increase ROI in the long run
4. after looking at the processed list, then I decide which terms are added to the real negative list in google ads. This part should not be automated.

Maybe the confusion I created, was questioning the "automation part".

Do you think only automating Part 1?
I wanted to also automate Parts 2 and 3.
I never wanted to automate Part 4.

Thanks!

 

[eliquid]

Hi, sorry.
I didn't mean or try to come across as arguing. I 100% agree with eliquid. I was hoping to get some clarification on the prior post. I completely agree that an all-automated script will create problems.

I will ask for clarification like this. In the script, there are 4 parts:
1. only writing to a google sheet.
2. using some logic with real-time stats on organizing the list(s)
3. testing/knowing/using better rules will help increase ROI in the long run
4. after looking at the processed list, then I decide which terms are added to the real negative list in google ads. This part should not be automated.

Maybe the confusion I created, was questioning the "automation part".

Do you think only automating Part 1?
I wanted to also automate Parts 2 and 3.
I never wanted to automate Part 4.

Thanks!

Got it.

Yeah, just have it write to a sheet, include the extra data you want with it ( I don't think it's needed, but you might want it ) and then check for the terms manually.

The problem you are going to run into though is this:

1. What time frame is this valid for? - What a lot of people don't realize is, because of market fluctuations in Google Ads, seasonality, errors on LP's that maybe get fixed, changes in your ad account, etc.... you may run this report and find something to negative without fully knowing why and a reason a metric can't explain to you.

An example might be ( and this is one I am just making up ), the client's LP for a category of their website was off/tracking not working or was a seasonal product. You run this script and negative match those terms.. but once things are fixed you now miss out on sales.

OR, lets say you made sweeping changes to the Ads in the adgroup and then a week later you run this and didn't account for that change that might now be fixing or messing up the bad metric.

OR, you have a 25 day conversion lag ( because your product isn't a knee jerk purchase ) and you don't add that in...

^^ I know mentally you can try to prevent this, but this does happen.

2. Also, lets say you pull a ngram or bigram out and you think.. man this is a slam dunk to exclude.

What you will find is issues in it being valid.

Meaning, a lot of times if you take that ngram or bigram and actually go into your search query report and paste that ngram in or bigram in, the data in the Google UX is different slightly than the data in your CSV report.

I know that sounds messed up, but it happens a ton for various reasons that include whitespace, how google interpets the term, if that term is part of a longer word ( string in a string, etc ). Most of this can be solved with coding, but man.. it will mess you up.

I don't think you really need the extra metrics. Just pick 1 and stick with it.

For example, if your client grades everything on ROAS, or CPA, or downloads, or something else... just only calculate that metric and grade against it. There isn't any real valid reason to also now include impressions, clicks, etc. - I would include COST though so you can safely say, "I havent given this kw enough spend data to show statistical relevance yet"

You could grade it on impressions, but COST is a better metric ( I feel ).

Then that way you are only really looking at data in the end where you have at least spent X dollars, over X time frame, where Xgrams are over X CPA/ROAS/Conversion Event

I keep it really simple.

For a lot of mine, I'm looking at multiple time frames ( at once ) to ensure seasonality and LP issues/tracking issues are weeded out... where I have spent at least $500 ( thats just a number I use ) in that time frame... and the Xgram is doing less than 2.5 ROAS ( again a number I use ) or is 2-3x over CPA goal ( again a number I use ).

Once I have those narrowed down, I then go into the Search Query Report ON GOOGLE, with the time frame I wanna use, and I paste in BY HAND each of those X-grams and I make absolutely sure the data is correct in the results. Sometimes I have to add a space before and ALSO after the X-gram.

You want to triple check this shit.

I can't tell you how many times this last step has saved my ass.

For example, lets say the X-gram is "rush" as in rush shipping. You find out thats poor performance because your company/client doesnt offer that.

You put this in the SQR and notice depending on match type, you might be negative-ing out the word terms like "crush" or "brush", and your main product is a brush you sell.

You'd be fucked up my friend.

This actually is a lot larger issue that would be too much for me type here.. but just know you need to check this shit before you do the negative and plan accordingly once you get the data back from the Google UX in the SQR.

Also with the way Google does matching now, you just never know if the word you are putting like "wood paintings" is going to negative out your "brush" term. I swear I have seen this happen in real life. You just want to take this step delicately and really think it out which terms you negative and the match type for it.

 

[LarryATX]

thank you very much eliquid. that reply is just fantastic!!!!!!
I love the details and explanations, tips. It screams "been there, done that"
thank you for sharing!

this whole thread has been awesome.

a question: you mentioned, your looking at terms over multiple time periods at once.
Have you seen a difference in grouping them to compare the groupings?
In my head, it seems if you have more months to look at or use, then pick a longer, wider time period. I'm not understanding the benefit of looking at 6 individual months, if you can look at one grouping for 6 months.

Thinking, new accounts, they need more time (and data). BUT if you have an old account, with lots of data... isn't more time always better?

oh, never mind... I think I just answered my own question because a "search term" could be seasonal. for example: IF you ONLY look at long time periods, you would never see it. The average would be okay, but looking at many, smaller time periods... you would THEN see it.