[Journey] Bro, do you even build links?

[brocrates]

This thread will record my efforts to build links, rank sites, and make bank.

I am starting this thread with five money sites. I am leaving one money site untouched as a control group for the time being. The site is also by far the most profitable. I don't fix what isn't broke.

The other four are the experiment sites. They all get a little bit of traffic and generate a little revenue, but I would lose little sleep if they all went belly up

Three of the experiment sites are good old fashion product sites. You give me money. I give you a product. Two physical. One digital.

One experiment site is a long tail traffic site. It'll get about 1k organic from google this month for long tail.

My immediate goals are to rank the product sites for about 5 keywords each. I'm letting google and the long tail site get to know each other better for now since they've already hit off quite well.

For the product sites I am setting up and tweaking SER campaigns using article, web 2.0, and comment engines.

I am setting up this journey because I've had mixed success with personally ranking sites. I've contracted out link build in the past. And it works. But it's also expensive and a total black box. I figure I'm better off in the long-run putting up sites I can re-use for link building over and over again instead of paying somebody else to build their network.

 

[speed jet]

Bro,
Subscribed your journey thread with blessings..

Good Luck on your journey brocrates..

 

[brocrates]

Now for something completely different. People talk about 'testing' SEO all the time. But, how does a person actually 'test'?

About 70 years ago some super cool dudes laid down the law for how to test stuff. Like this bro. And this bro. In particular, I'm going to give you the 101 on a very simple way to use diff-in-diff to test your SEO juju.

The main idea is simple. Suppose you've got two pages you want to rank for different searches. You build links to one of them (A). You don't build links to the other (C). Wait a month. Now, compare the change in the ranking of (A) to the change in the ranking of (B). You've just done an adhoc diff-in-diff. And you're probably already doing something like this. But, what if you've got pages C, D, E, x100 and they're all in different niches and they've already got a lot of link growth and decay going on and some searches are more competitive than others and blah blah blah. You need you some diff-in-diff action. And, you don't want to build links to half your sites. That's half the number of sites you want ranked!

Here's a simple experiment design you can use. There's more souped up designs, too to that give you more info and control for more stuff, but this will suffice for the general BHW public.

1. Record the ranks for all your keywords at date t=1.

2. Plop these ranks into excel along with a column to keep track of the keywords. Add another column, call it 'date' and fill it with zeros. You can also use google sheets if you can't afford excel. We're calling the day you do this date t=1.

3. Start your new experimental link building campaign on all the keywords. Pick a date in the future to check in on how your sites are ranking.

4. Wait.

5. Wait.

6. Seriously, wait like a month.

7. Now we're at date t=2 a month or more into the future! Repeat step number two. But, first, you're going to place your new rows of data directly below the rows of data the spreadsheet you made for number two. Second, we need to tell the t=1 and t=2 ranking results apart. So, for the new rows of data you've got, instead of filling the 'date' column with zeros, put ones in the column instead.

8. We're going to add some more columns to the spreadsheet now. Copy and transpose the keyword column into a bunch of new column headers.

9. Write a short excel formula to set each cell to 1 if the column header and 'keyword name' column from step 2 match. Zero otherwise. So each column should have two 1's in it and a lot more zeros.

10. Power up the data analysis tool pack for excel. Choose the linear regression tool. For the Y column, select the column of the spreadsheet with all your keyword ranks in it. For the X matrix, select the 'date' column and all of the keyword columns from step 9.

11. Choose some place to output the results that won't overwrite your data.

12. Click the 'next' button or 'start' or 'regress' or whatever the button is called.

13. a. Now look at the results for the row labeled 'date'. Is the number in the 'lower 95%' column positive? Then I've got good news for you. Your new experimental link building method likely bumped you up in the serps!

13. b. Alternatively, is the 'upper 95%' column negative. Then I've got bad news for you, whatever you did likely hurt your rankings.

14. Now when will this procedure not work? The most important way this procedure can go south is if google rolls out an update the moves all your sites one direction in the rankings. There's ways to change up this procedure to control for this sort of error and others, but that's for later.

If you want to read more do some googling on linear regression and diff-in-diff in particular. Edit: Also 'average treatment effects'

 

[W9go]

good luck .... googled "diff-in-diff" not really knowing if this will help for my set

 

[Trained Wolf]

Good luck on your Journey OP

 

[brocrates]

This is a short update.

The physical sites are up in the rankings slightly. But, I'm still waiting for them to get in the top 100 for most of the searches I'm targeting. I've got a long ways to go for my goal of ranking the sites #1 for all the keywords I'm targeting.

The guy finishing up the digital site bungled the analytics. Thankfully, the cool dude taking care of the paid campaigns caught this and fixed it up. I think the guy is a BHW visitor, too. Advertising has not started yet.

I've still got about three weeks till the SER campaign experiment I'm running comes to an end and I use some diff-and-diff to decide what did and did not work.

The long tail site and google are very well acquainted. The traffic is up to ~80 new visitors / day. This is by far the most daily traffic I've gotten to a site so far in my life !!!

I'm completely finished with the SER related projects for now. In particular I coded up some extra special sauce for inspecting target sites. For example, do you think links on sites with compromised SSL certificates are going to be worth much? I don't. Let's get rid of those.

My next personal project is automating link building from my long tail site to other sites.

 

[brocrates]

Of course, pics or it didn't happen.

Physicals A

Physical B

Mind you the search volume is world wide. But the SERPs are local.

Long tail

I've set up long tail sites in the past. And they were all duds. In fact I've set up about 50 of them. I suspect the past ones were all flops because the on page content was a fraction of what I've got going on for this site. And, I optimized the bejesus out of google's on page user experience scores for this site. Like, bro do you even http2?

Edited a stay img tag.

 

[iBlackSky]

All the best to you on this Journey.

 

[RavageX]

Nice journey, useful and helpful information. Will stay tuned

 

[brocrates]

I made my first official amazon sale today for... $1.76! Or about $0.01/click! The earnings look a lot more impressive in the little plot amazon provides. No matter the size. It's still progress. I have to make $1.76 before I can make $2.00.

 

[TWdesigns69]

I made my first official amazon sale today for... $1.76! Or about $0.01/click! The earnings look a lot more impressive in the little plot amazon provides. No matter the size. It's still progress. I have to make $1.76 before I can make $2.00.

awesome. i too am writing a similarr traffic building journey thread. do u comment on forums? its an effective method imo

 

[Jiung]

Wow, rank 8 for 135k MS must be nice!

 

[brocrates]

awesome. i too am writing a similarr traffic building journey thread. do u comment on forums? its an effective method imo

I have not.

Wow, rank 8 for 135k MS must be nice!

The SERPs are local for the keyword are local. There's maybe 300-400 local searches for the keyword.

 

[brocrates]

Fun fact: If the dinky server your hosting a site on craps out, then google quickly yanks your site from SERPs. There's probably a lesson here too for which sites to build links on.

Traffic to the long tail site yesterday dropped by 50% because the wee instance I was running it on couldn't handle a tiny traffic spike.

Solution:

I'm moving the site to nginx and uwsgi. I expect this will substantially reduce the memory footprint directly. uWSGI has caching and process baby sitting built in, too. I expect both of these will let the dinky server handle more traffic.

I'm using this as a chance to write up proper ansible scripts for nginx/uwsgi provisioning, too. [BTW: ansible is legit. unless you like reinventing idempotent bash scripts. then keep on bashing]

I suppose this is a good problem to have.

 

[blackh4t]

How many links you build?

 

[brocrates]

Btw, I also looked at using circus from mozilla for process management. But, at this time, I think circus solves mostly problems I don't have and would require additional software to solve the problems I've got.

uWSGI solves a lot of problems I don't have. But I also don't need more software to solve the problems I've presently got. Plus the uWSGI docs should get an award with how much info is in them. And not like the brick wall of stupifying text you'd see on the man page for some bash program. The uWSGI docs contain material for people who don't dream of bash pipes and obscure kernel features.

 

[brocrates]

Physical sites

On balance my sites have moved up in rankings for keywords they're already ranked for over the past monthish. But, they have not shown up for new keywords.

Digital
The digital product site has been live for one week. 44 visitors from adwords. The email capture rate is ~5%. The sales rate is 0%. I'm going to try a different audience. Right now it's only US based out of habit. I figure anyone speaking the same language can buy a digital product though so why not try restricting by language instead?

Long Tail
The long tail site is now getting ~200 visits/day. And still no conversions from the amazon affiliate links. I'm still swapping out apache set up from something leaner and more robust to handle the traffic without needing to manually restart the server every few days.

Next Up
1. I'm finishing the long tail site server configuration.
2. I'm contracting out setting up a dropshipping site. This is not my skillset and not a skillset I want to develop when I've got too much to do already!
3. I'm finishing the code to use the long tail site for link indexing.
4. I'll make a few more sites like #2 based on the same data.
5. I'm adding #2 to my link building stack.
6. I'll set up an experiment to test #2. This will use the methods I talked about earlier.
7. I'll figure out what project I want to run next.

If the ranking tools are working, then I'll set up some money sites. Else, more link juice!

 

[PropertyGrow]

What kind of links are building? Do you have a PBN? web 2.0s?

 

[Fiesty0ne]

Thanks for the read about the testing. Never knew about the DnD

 

[brocrates]

What kind of links are building? Do you have a PBN? web 2.0s?

This experiment is web 2.0s, SER, and blogs.