Apple Privacy & Facebook Advertising

[Paul Carmen]

If you want to see keywords from google.co.uk, connect Google Search Console with Google Analytics

It's saying things like this that are part of the problem, a little knowledge and all that... What you've said is completely true, but we are talking about paid advertising in this thread. But, all traffic channels matter, so, in fact you can connect Google Ads, Analytics & Search Console together, you can even pull them all into Google Data Studio and build great analysis of that data if you wish.

The problem is Google now shows less than 10% of your actual organic keywords in GSC accurately since they started anonymising user data, and about 1% in GA, other than the biggest queries its an AI driven fudge. Now do you think you can make good statistically relevant data driven choices missing in excess of 90% of your true keyword data?

The answer is no, you need other tools and data/keyword processes to do this, if you don't you will not be getting a true picture or making the right choices.

It's the same with using Google Ads or Facebook Ads solely to track your campaigns; can you track other traffic sources, can you track what page they landed on, can you track what page they converted from, can you track all your phone calls, form fills, sales, can you tie this back to the campaign, ad, keyword?

If you try to glue this together most people give up, or they say my analytics are never right, or the sales numbers never match the campaigns etc. The majority soldier on like this, with every data driven change by the platform making things worse for them, and having to blindly trust Google or Facebook!

 

[SEODEV#338055]

Now do you think you can make good statistically relevant data driven choices missing in excess of 90% of your true keyword data?

The answer is yes

If your data set is 10,000 daily visitors you can make very statistically relevant data driven choices from 1,000 keywords, in addition to 50,000 daily keyword impressions

10% is more than enough

 

[gpietersz]

That's the problem though, 99% of people can't do what you say, unless you have the buying power and budget to work in this way, as the big platforms control the placement, not the content provider/sites. Even the biggest websites tend to show Google display ads now, with less and less doing their own thing or running homepage/category takeover type ads.

Agreed, but if their current system became ineffective because of greater privacy controls, either they would have to adapt and develop the systems to do this, or other ad networks would do so.

I have not used Google ads for a while, but when I did it used to both allow advertisers to target particular sites and allow sites to pick which advertisers they allowed.

While I agree with you that it should be better, and the web isn't great as the wild west, the direction now is making it worse, most people just ignore/shut the cookie popup as they know the site needs it to function, so its just become an intrusion.

Which is why I use a browser extension (on my main browser, I use several) that automatically agrees to cookie popups - together with others to block third party cookies and delete cookies after I close all tabs on a site.

The problem that needs addressing is the biggest corporations in the world should not be making the decisions and regulating themselves.

I agree entirely. I think part of the problem is that regulators and governments do not have a good understanding of the technology and have not yet caught up with the complexities of regulating this.

 

[SEODEV#338055]

Also 10% of keywords is not accurate according to the data I have

It's more like 30%

 

[Paul Carmen]

The answer is yes

If your data set is 10,000 daily visitors you can make very statistically relevant data driven choices from 1,000 keywords, in addition to 50,000 daily keyword impressions

10% is more than enough

Great, please share how you'd undertake that analysis and the decision making process, plus use a realistic example where that level of data would be available to help most SME businesses make meaningful statistical decisions on pages, conversions etc?

 

[SEODEV#338055]

Great, please share how you'd undertake that analysis and the decision making process, plus use a realistic example where that level of data would be available to help most SME businesses make meaningful statistical decisions on pages, conversions etc?

Are you saying that you can't make statistical decisions based on knowing 30% of the keywords customers are using to find a business?

Or are you trying to find out my SEO strategy?

 

[fisicx]

Are you saying that you can't make statistical decisions based on knowing 30% of the keywords customers are using to find a business?

But what if those 30% aren't the converting keywords. What if the one that could earn you a bucket load of cash is in the 70% you know nothing about?

 

[Paul Carmen]

Are you saying that you can't make statistical decisions based on knowing 30% of the keywords customers are using to find a business?

You can make statistical decisions based on any data. However, what I'm talking about is "Statistical Significance".

Now, you may be able to make decisions based on statistical significance with only 30% of the data available, but in most circumstances its pretty unlikely, and if you're not careful you'll just be making a guess.

Or are you trying to find out my SEO strategy?

I'm not interested in your SEO strategy.

I wanted you to explain how you'd come to this conclusion data wise and give an actual working example of how you'd actually do or model this. Just to clarify, your original example is vague and not really relevant in the real world, as most businesses on this forum won't have sites attracting 10,000 visits a day, and for those that do it will be across many tens, hundreds or even thousands of pages.

So it comes back to how do you make meaningful decisions based on statistical significance if you're missing 70-90% of the granular data, as the bulk of your data will be missing as you drill down into a site. Depending on what you can/can't see you could target the wrong keywords for a business, going after high volume short tail stuff that gets traffic but not conversions/sales, when longer tail might be much more fruitful.

Finally, and to the point, this is a thread about paid ads and the impact of iOS 14.5, specifically to Facebook tracking data, but Google Ads has been talked about too. If you're missing 70-90% of your data here, you'll make the wrong decisions on campaigns, ads, keywords, pages etc. For A/B tests and demographic data to be relevant you need to know that the data is all either the same/or truly random. If you have a 70% error rate, your data and tests are useless and will mislead you, which is very much the hot new issue with Facebook pixel and data attribution for their ads post iOS 14.5.

So I am very interested in how you've reached your conclusions, plus how you apply this analytics/research methodology in your field, and specifically how you'd apply it to help with Facebook Ads and the original question of missing data?

 

[SEODEV#338055]

So we're agreed it's 30% and not 10%?

 

[Paul Carmen]

So we're agreed it's 30% and not 10%?

Either number works for the sake of argument, as both are too low normally for statistical significance. However, you seem to be avoiding the point or unable to answer the question, which tells its own story...

 

[SEODEV#338055]

Thank you for flagging my factual statement as junk

Come on, let's have a mature debate without resorting to personal swipes

Google Search Console is an excellent tool for keyword data analysis, and I think we should promote its uses rather than focus on its limitations

Your statement was: "Either number works for the sake of argument" and I take issue with that because the correct data is important, that's not a personal attack it's merely pointing out the facts

10% isn't 30%, it's not even 50% of 30%

10% of a 100,000-keyword dataset is 10,000 keywords

30% of a 100,000-keyword dataset is 30,000 keywords

My original statement stands: "If your data set is 10,000 daily visitors you can make very statistically relevant data driven choices from 1,000 keywords, in addition to 50,000 daily keyword impressions."

We've subsequently established that 30% of keywords can be accessed through connecting Google Search Console to Google Analytics, and if forum members would like to set this up here are the instructions from Google https://support.google.com/analytics/answer/1308621

 

[SEODEV#338055]

You have a single site with 100,000 keywords?

Sure, why not?

But again, it was an example highlighting that keywords are available to access and that we can make informed business decisions on what is available from Google

 

[Paul Carmen]

Thank you for flagging my factual statement as junk

It's all junk mate, you took a point about Facebook Ads/Google Ads, on a thread about Facebook Ads, and started talking about GSC and SEO in answer to a point I made about paid ads. That's spammy and seems to be about diverting the thread to talk about you.

You then failed to answer the question I asked you, coming back with questions about your SEO strategy and asking another question about statistics. I clarified what I was talking about and the question for you.

You then started arguing about a minor point, that's not relevant to the thread in any way, certainly not one that you've proven (unless I missed the proof, it was just a point score based on semantics from my post). This is the classic I'll argue a minor point to deflect from the answer, and It's not even relevant to what we are talking about.

You still don't seem to understand that for most scenarios, and particularly for this thread, your argument doesn't stand up at all. It might just do for a handful of main pages on a website (it most probably won't but, statistically it could), for most it will at best help guide your initial research only, for granular keyword research you need other tools. GSC is good for checking indexing, for flagging major changes in performance and direction, for broad brush website technical SEO work. What it aint any good for in most circumstances is granular keyword research.

If you want to make proper choices based on statistical significance it is just as likely to mislead you badly, especially if you use that as the only data source/tool. I've based my 10% on our analysis of a large number of websites we've worked on, and a large number of GSC accounts that give very poor keyword data versus our keyword tracking and monitoring tools. Plus, unless you set GSC up very specifically, it's not very accurate at all (often showing keyword data for the wrong pages, canonicals, averages that are way out compared to GA or site logs etc.). If you want to research it there are studies and data analysis available publicly:

• https://moz.com/blog/google-search-console-webmaster-tools-reliability
  
• https://www.searchenginejournal.com/google-search-console-google-analytics-data/
  
• https://www.portent.com/blog/analytics/google-webmaster-tools-query-data-is-worthless.htm

To return to the thread topic, it certainly doesn't work or help at all for paid ads work. What we need is statistical significance to make choices that work; e.g. an advert or landing page A/B test needs to be carried out with all data monitored to pick the right winner statistically. To know if a keyword performs you need to measure all clicks and all conversions to make the right choice on whether to bid up/down or stop it altogether.

However, you do you. If you want to argue semantics and clutter up the post about something else fine, I'd suggest you start your own thread. This isn't Facebook, I'm not looking for an argument, I'll leave my future comments to details about what's actually being discussed in the thread.

 

[Financial-Modeller]

I've pondered for years, that we will end up with a two-tier web; with those who pay for privacy and those who don't.

Some users will continue to use tools provided freely by companies that will recoup development costs etc by monetising user data and selling it to advertisers etc

Other users will buy a licence to use tools developed by companies that will recoup development costs etc by licensing the tool to users. Those users will effectively by able to buy privacy.

 

[OldWelshGuy]

Use the tor browser, nobody will track you then

I personally have no argument with being tracked beyond the fact that as a marketer I visit a massive range of sites, many of which are not in the least bit of interest to me beyond carrying out my work.

Facebook in very invasive in as much as it tracks users via its app even when the app isn't opened on the phone and that winds me up.

The Facebook tracking pixel was fun while it lasted, it was always going to fall foul of the new woke privacy beliefs and with Europe and the US having these laws, it had to happen.

I have to say though that seeing Apple post an advert with the headline 'We value your privacy' made me choke on my tea

 

[kulture]

I have to say though that seeing Apple post an advert with the headline 'We value your privacy' made me choke on my tea

But they do "Value your privacy" I bet that value is somewhere deep in their accounts logged as an asset.

 

[ctrlbrk]

Facebook says that it is working on new features to help make up for the data lost due to App Tracking Transparency, including “new advertising features that require less data to measure an ad’s success.” The company is also reportedly exploring ways to deliver ads based on data stored on the user’s device.

From Facebook and its advertisers are ‘panicking’ as the majority of iPhone users opt out of tracking

 

[Paul Carmen]

From Facebook and its advertisers are ‘panicking’ as the majority of iPhone users opt out of tracking

This goes back to unintended consequences, a lot of people think "I don't want to be tracked, so I'll opt out", but don't know what this actually means for them; e.g. instead of having demographically or personally targeted ads shown, they still get ads, but of far less relevance to them. Some may be happy with that, plenty won't be...

It still doesn't alter the fact that the Facebook pixel is not great from a tracking perspective anyway, and that most advertisers could have improved their performance significantly before Apple decided to shaft Facebook. They can still track and test/improve their ads/conversions now, just don't rely on the Facebook pixel!

 

[alan1302]

e.g. instead of having demographically or personally targeted ads shown, they still get ads, but of far less relevance to them. Some may be happy with that, plenty won't be...

People won't care in they are targeted ads or not

 

[gpietersz]

People won't care in they are targeted ads or not

Beat me to that. Very few people are going to care how well targetted ads are. How often does anyone think "I am so glad i saw that ad".

 

[DontAsk]

I don't give a flying **** whether as are targeted or not. I just ignore them all as even 'targeted' as are irrelevant, being based on data captured in the past. I've either already bought the product or service in question, or moved on completely.

Random ads would be much more likely to be advertising something I am interested in, like leading through a magazine or newspaper. Even they are 'targeted' to a certain demographic, but I might be reading something outside my demographic in the dentist waiting room, for example.

Who are all the airheads who fuel this sort of advertising by engaging with it?