Thread by @darth_na

1/🧵
::: Google do NOT use CTR as a Ranking Factor! :::
:::But does that mean they don't use Clicks? :::

This is a long running subject,
and one that get's muddied by poor phrasing,
almost every time :(

🧵>>>

#SEO #CTR #DarthAutocrat

https://twitter.com/rustybrick/status/1456587346816626691

2/*

Let's be clear here:

* GA BR (Google Analytics Bounce Rate)
* GA ToP (Google Analytics Time on Page)
* GA ToS (Google Analytics Time on Site)
* SERP CTR (Search Engine Result Page Click Through Rate)

are NOT (direct/indirect) ranking factors!

>>>

3/*

For starters, only a % of sites use Google Analytics,
so there'd be a Huge data hole.

And each of them are ambiguous/noisy,
with various reasons for whatever value,
(a page with the weather - quick visit, leave - does not mean the page sucks or is irrelevant!)

>>>

4/*

"But, but ... correlation!?"

Yes.
There are various studies that point to things like high Bounce Rates and low rankings.
But that is a casual correlation - not a causative one!

People may "bounce" from those pages for various reasons (some are ranking factors).

>>>

5/*

Metrics like BR or short "Dwell times", (time on page/site), may be indicative of an issue, or not -
so are not themselves actual signals!

Further, as they are non-descriptive, using them on their own would be pointless.

>>>

6/*

So we have a bunch of useless metrics,
and half baked questions/logic.

But to make matters worse, we have semantics and the fact that google doesn't really want to tell us some things.

>>>

7/*

When "we" say "ranking factor" - we usually mean anything that can cause/contribute to a positional change of a listing in the SERPs.

But to a Googler, it can mean that,
or it can mean a specific type of signal, with a certain influence, at a set stage in the algo!

>>>

8/*

This means that they can honestly answer such questions, but give a number of potential answers.

(This can be further compounded by changes, and even not knowing (not everyone is told everything!))

So asking questions and getting answers can be problematic!

>>>

9/*

Now, the concept isn't new.
It's been floating around since around 2008.
(@ajkohn was the first I remember on it?)
blindfiveyearold.com/search-pogosti…

No stupid metrics - instead using SERP data (which G has 100% of, unlike GA Bounce Rate!).

>>>

10/*

Pogo stick, Short click and Long click.

Pogo sticking covers Bounce Rate,
but a specific sub-type of bounce:
bouncing back to the SERP!

The user searches, clicks, returns to the search.

>>>

11/*

Then we have Short and Long clicks.
This covers "Dwell time" - how fast the pogo stick action is, if at all (some people search, click and don't return).

But even these aren't that well formed concepts,
and miss things out!

>>>

12/*

We also have people like @CyrusShepard and his piece looking at First, Long and Last Click;
moz.com/blog/click-bas…

13/*

Now, unfortunately G+ has gone - which sucks,
as I beat this topic to death (several times) on that platform.

There are multiple potential data points that could be used, and the more of them, the more informative the result.

>>>

14/*

User session
Search session
Query
SERP type
What gets clicked
Whether they return
If so, how fast
Do they click another
Do they refine/alter the Query
Do they do a different search
Time
Quantity

>>>

15/*

Now, I now the majority of you that are reading this (thank you for getting this far :D)
are looking at that list and going either "hmmmm" or "ahhh".

Stop

Look at the list again, and start thinking "negative",
not just "positive".

16/*

Almost everyone jumps to the idea that a higher CTR and no return to SERP could mean a positive ranking gain.

It may be more viable to work it inversely!

Further, there are several things we may be able to gain from SERP interactions!

>>>

17/*

Appeal?
We know that, in most cases, there's a natural decline in CTR by descending position (Pos. 3 gets a lower CTR than Pos. 2, which is lower than Pos. 1 etc.).

What if people skip the first one or two, and a larger % of users click on lower listings?

>>>

18/*

Relevance?
We know that some search sessions are more than single-shot, as we search ourselves, and we sometimes refine the search.

What if the user does their search, clicks on a listing, comes back and tweaks the query a little?

What about what they click on then?

>>>

19/*

Suitability?
G don't just use the Search box data - that attach additional info to queries, such as language and location etc.

What if the click behaviours differ by device/resolution etc.?
(note the difference between Desktop/Mobile at times?)

>>>

20/*

So rather than leaping to the assumption that clicks improve rankings,
consider the potential of what non-clicks could mean, or clicks and then changes.

Think of all those:
* Search Suggestions
* People Also Ask
* those weird keywords you drop after algo updates

>>>

21/*

(And yes - "non-clicks" ... and Googlers saying "we don't use clicks" ... see how that could work? :D)

Consider Ambiguous queries (ones that could cover several meanings (jaguar), or several intents (pizza)),
and how G can decide what to show (or not), and where.

>>>

22/*

Then we have the question of data and processing.
But G don't have to do this for all queries.

They can do it for high-volume ones (surface the most satisfactory results).
They can do it for ambiguous ones (identify the most common sense being sought).
etc.

>>>

23/*

But ... what about Click Farms and faking it?

Remember that list (tweet 14)?
At the bottom were Time and Volume.

If "last click" was a factor, you'd want to see that as the case a set %, and see it supported over time.

>>>

24/*

Now imagine the number of clicks you'd need for popular search terms to make a significant difference to the "normal" click distribution figures.

And, if they were using "non-clicks", that scales more!

And you'd have to do it over a period of time.

>>>

25/*

And, if you wanted to retain whatever ranking influence you might obtain - you'd have to sustain those (non-) clicks (else the rankings could revert, or you could potentially fall further!).

See how far more complicated it could be than just
CTR
?

>>>

26/*

Now, I was fortunate enough to have @methode answer a few questions when he did a Reddit AMA 3 years ago (was it really that long ago?)

Rank Brain may use SERP behaviours:
reddit.com/r/TechSEO/comm…

And

G may use SERP behaviours for evaluations:
reddit.com/r/TechSEO/comm…

>>>

27/*

NOTE:

This is Not Gary saying
"Yes, CTR is a Ranking Factor".

That is Gary saying
"Yes, we use some User Interactions for some things".

Not the same!

The fact that RB may have included/surfaced your content isn't technically the same as "ranking influence".

>>>

28/*

There are other potential use-cases for User SERP Behaviours too (how would you decide which of 2+ pages to show for a query, if they are identical?).

So it's Not CTR for Ranking.

It could be several User Interactions, in sequence, for some searches, to do some things.

29/*

Now for the fun part!

Personally ...
... I believe that G use User SERP Behaviour for Ranking.

As in - I think that they have stuff in place that can alter what position things may appear in,
based on what Searchers do in the SERP.

>>>

30/*

But what I think doesn't matter.

What we need (want!) is to get Googlers to answer questions about it,
rather than silly questions about Click Through Rate :D