Fewer Visits Leads to Lower Rankings?

[Dymero]

Recently I've discovered a peculiar phenomenon, and I'm wondering if anyone else, particularly in the e-commerce space, has noticed this?

Over the past few years, for a two to three month period around late April to early July, the overall rankings for the site will go down. Many individual keywords might stay right where they are, but collectively, rankings will go down. As expected in this situation, there are fewer visits.

I used to think this was due to a penalty of some sort, because it happened right around when a new Google update (usually Panda) was released. Now I think differently, as it turns out the entire site sees fewer visits (and sales) during this period, so it appears to be less a penalty and more a sales trend thing. People just aren't going to this niche's sites as much during the period.

That got me to thinking: Could fewer visits to the site lead to lower rankings? If Google notices that a site is getting fewer visits, the triggers something in the algorithm to lower rankings as a whole? And if the site gets more visits later on, it'll be triggered to increase rankings.

I've always waved aside speculation that Google is using site statistics to determine rankings, especially because G themselves say this isn't the case, but now I'm not so sure. What does everyone here think?

[Pjman]

One thing that I have noticed is that google referrals stay at pretty much constant ratio with my direct traffic. I only notice this because when I send out an email newsletter google real time referrals climb to keep the ratio. I send my emails pretty randomly, but they hold the ratio like clockwork.

[Dymero]

As far as I can tell, it's been the same here as well.

[denisl]

Pjman, I have noticed that ratio between G trafic and "direct". Have come to the conclussion that most of the "direct" trafic is infact Google trafic but with the referer missing.

[Pjman]

@denisl, I thought that for a long time too, but seeing the ratio stick when I send a newsletter that goes to 250K+ active users; really throws me for a loop.

Why do I get 40% more Google referrals during the 48 hours that my newsletter is going out? This is regardless of the day/time (Sun, Tues, or Fri) it's sent. The only thing that goes up by 40% at that time is direct traffic.

[Dymero]

That is really interesting, Pjman. Has anyone else seen this happen?

[Ralph_Slate]

Yes, I have been tracking this phenomenon for nearly a year. I noticed that my Organic/Direct/Referal numbers are relatively constant despite my traffic levels. The only thing that seems to vary them are mobile hits - on weekends, when I get more mobile traffic, I get more "direct" than on a weekday.

I also noticed a slight shift in these numbers on October 13, 2012 - the day my lost-traffic-from-April-24 returned - the percentage shifted away from referrals and toward organic.

The pattern I saw with organic was that from Jul 1 2012 until about Sep 21 2012, it was always between 60% and 65%, and usually at about 63%. Then, in late September, my site's season niche started to increase. By early October my organic number shifted down to about 57% to 60%.

Then my recovery happened, and the organic number moved up to the 63% to 68% range. It cycled more, due to mobile traffic on weekends related to my site's niche.

In mid-December, the pattern was disrupted. I'm not sure if that's related to general Christmas traffic (my site isn't related to Christmas though) or maybe something to do with college students going home on break (my site isn't particularly targeted to college students though).

Then, in January, my niche had an event which caused traffic to rise back to historic levels - the widgets in my niche had been locked out by management and they returned in early January. This changed the pattern too as interest in the niche picked up. The range shifted to the 61% to 65% range.

On May 1 2013, my site's niche had its peak interest day due to an external event. Following that day the numbers shifted again, and are now in the 56% to 63% range.

I noticed no major shift after the Penguin 2.0 event, although my "referral" category is showing a bit more variability than it had been prior to that date (spikier).

I feel like these numbers are important indicators, though I can't pin down exactly why.

I'm not sure if it's appropriate to post a graph here, but I'd be willing to do that if people are interested in seeing the numbers.

[RP_Joe]

I have suspected Pjman observation for some time.

It explains many things.

[Pjman]

I must note that I only notice this on my older sites 8+ years.

My newer sites don't experience this, I have several new
(<3 years old) sites where this phenomenon doesn't hold true. The Direct Traffic vs. G referrals ratio on those three fluctuate all the time. Newsletters don't affect the ratio at all.

It's as if G is saying, "Old sites: We gave you traffic for years. If you were really that good, people would still come to you regardless of the traffic we send to you!"

[Shaddows]

Back when Zombie wasn't a dirty word, I noticed that not only did G traffic vary with Direct, but if Direct Conversions (quantity, not rate) spiked, then G traffic declined in quality for the rest of the day (became zombified)

The thing is, it's pretty darn hard to explain this. If you talk about it too much, people think you're crazy. And you might well be, because it MAKES NO SENSE that G knows either your Direct visitor numbers, or any type of conversion quantity.

Back when there was a serious attempt to explain the Zombie phenomenon, people were actively encouraged to strip out non-G traffic before looking at the numbers. Which makes sense, but makes it impossible to notice co-occuring patterns like this.

Anyway, my rational mind tells me there is no way for Direct and Google traffic to be tied in any way, apart from the prosaic pan-niche explanations like:
- Seasonality
- Trendiness
- Newsworthiness
- Fad / buzz / hype (and lack thereof)

ETA - Site is 13 year old ecom (so matching the age-pattern), with no problems and strong rev growth, particularly in last 18-20 months (so not having a moan)

[Dymero]

If you talk about it too much, people think you're crazy.

This is why I created this thread. It just didn't make sense to me why rankings seemed to be tracked to industry sales trends (or visits, which is also related). It seemed crazy, but there it is.

On a side note, we're having one of our best day-to-day weeks (meaning rev is up every day, compared to just some really good single day) in a month or so. And guess what? Traffic to the site as a whole, including organic, started going back up a couple weeks ago. Around the same time, rankings went back up. Coincidence?

By the way, I don't think it might be so weird that both organic and direct traffic go up at the same time, or down, even. Direct is a bit of a hodgepodge of actual direct type-ins, copy-ins, or bookmarks, and whatever else isn't tracked, including no-referrer organic in some cases. Also, if you're getting some buzz, everything should go up, and fall down a bit once you're out of a news cycle.

But to track so well with each other is really strange. You'd expect some times organic goes up and direct goes down, or vice versa. That's not what happens, though, at least not in our case.

[treeline]

Maybe it's simple. They could be clicking on a link in your email from their traditional email program. Unlike webmail, it will show up as direct.

Lots of web users get to sites in unexpected ways. Instead of clicking your newsletter links, they could type them in google, cut and paste the link into their browser, or they use pre-existing bookmark, or create a new one and test it.

[indyank]

And you might well be, because it MAKES NO SENSE that G knows either your Direct visitor numbers, or any type of conversion quantity.

This thread interests me as I have been noticing these before. There are two things there.

1) How does G knows your Direct visitor numbers?

This might be possible if they get the data from their own and various third party sources. This includes

a) analytics software (their own and those owned by third parties),
b) browsers (their own chrome browser and other third party browsers).
c) all other sources.

Either they buy the data from third parties or they have means of getting such data from those browsers, say, thro. the integrated Google search tool. Almost all browsers would have the popular search tools like google, bing etc. integrated.

They then correlate data obtained from various sources and estimate the best possible direct traffic to a web page, which might be fairly accurate.

2) How does G know the conversion quantity?

Google might consider a frequently clicked link or button on a web page as the conversion unit on that page and keeps track of the number of clicks. They could do this for G (or search engine) traffic alone and then apply the direct to organic traffic ratio arrived at from 1) above to determine the overall conversion quantity or rate.

@shaddows and others, do you think this is possible?

[Shaddows]

Good point, they have aggregate data from Toolbar, Chrome and whatever else. They could benchmark your Organic exposure using Direct traffic from the aggregate data.

I suppose they could use the same aggregate data to turn the tap off, if you hit your Transaction limit for the day. But that seems a bit... evil.

"That's enough business for you, all you're getting now are info seekers"

If that was really happening, I suspect
a) Someone else would have noticed
b) Lots of noise would be made.

So, given that there is no noise, it can't be true (or at least not statistically significant in a substantial number of private datasets).

Anyway, I'm going to stop talking about the Transaction limit. It does sound crazy, and I have no wish to lose whatever remaining credibility I may still have.

[Pjman]

@Shaddows

I agree with you that it sounds crazy and that someone else would have noticed, but wouldn't it be ideal for Google's business model?

I know if I had a few hundred geeks at my disposal and I was them it is what I would be working on right now.

[RP_Joe]

I don't think its one size fits all. This may be a factor in low volume websites or low volume pages or highly competitive ares.

[martinibuster]

I agree with RP_Joe. Not that simple. Consistent clickthroughs are not a measure of quality. The reverse is true as well. Poor clickthroughs are not a measure of poor quality.

Fewer visits means fewer clickthroughs. Is it because the title tag & meta description are not good? Is it because the site positioned beneath it has better title tags?

Or... is it because the listed site is irrelevant? If that's the case then perhaps the algo would be looking for close to NO visitors, or better yet, a statistically low amount of visitors for the typical site listed on that SERP. Fewer visits can be attributed to many things. No visits or statistically improbably low visits might be a flag.

Now compare that metric with another one, time spent on a site. Time spent on a site could be indicative of user dissatisfaction, especially if followed by clicks on other sites listed on the SERPs. It could indicate satisfaction if the user ceases to click on the SERPs. Clicks on several SERP listings could be a flag that something's wrong with the rankings. Etcetera.

[indyank]

@martinibuster clickthroughs from SERPS is definitely another factor but what I feel is clicks on a conversion unit embedded in a web page might also be a factor. For Ecommerce sites this could be the Buy buttons etc., for blogs this could be the clicks on the submit button within a comment form, for download sites this conversion unit could be the download buttons or links and so on.

This would help them in arriving at the conversion rate which is yet another indicator of user satisfaction.

Don't you think this could at-least be true for transactional queries? But my guess is this might as well be used for informational queries as well.

[martinibuster]

indyank you are on to something. Justin Sanger touched on what happens pre and post transaction then took the topic way out there, way beyond the SERPs, in one of the best presentations I've seen in a long, long time. This was at the recent PubCon New Orleans. He talked about how there's the sales funnel we've all been talking about all these years. But he said that at the other end of that sales funnel there's a post transaction funnel that opens up on the other side.

Marketing, Facebook, Links, Website> Transaction < Post Transaction Actions

He then put up a slide dotted with the logos of companies that Google has purchased or has invested in that have something to do in one way or another with what happens in the post transaction period. Post transaction actions can be friend to friend recommendations, reviews, repeat transactions, coupons, emails, text messages, mobile/local business intersections etc.

He mentions some of that, in a different context, in this blog post on JustinSanger.com. [justinsanger.com]

Today, the marketing funnel opens right back up post transaction as the business has an opportunity to move the customer to a connection and potentially into an advocate.

[Dymero]

@indyank, interesting ideas. Unfortunately, I don't immediately see how this could be tested. But if it is ever proved, it'd fly in the face of just about every statement from Google, that says they don't use stats for rankings. I know few at this forum believe that, though. Just saying.

@Shaddows

I suppose they could use the same aggregate data to turn the tap off, if you hit your Transaction limit for the day. But that seems a bit... evil.

I've heard this a lot, but it hasn't been my experience that a tap is turned on or off at certain times. There are low times and high times throughout a day, but that's to be expected. Of course, I'm sure there are theories why some sites see the tap turned off and some don't, some more conspiratorial than others.