Landing Page Dynamic Testing

[Graybeard]

Dynamically loaded (1 URL Only --'one page') landing page with;
Test completed in 4 hours run time
3247 ../clicklog/traffic.log <=raw landing page 'in [pops]'

however the coding to do this took maybe 20 hrs and and was a real headache!
The programing can be repurposed though ...

• 10 adult webcam sub-niche links
• 6 randomized cam videos
• 4 color scheme versions
• All counts are CTR from the landing page

RESULTS:

Top video -- pretty girl but not hard core like some were --surprise!

Runner up was 'very' graphic lolz

top version color scheme:

Not telling --but they are not in the order of the image above

 

[Graybeard]

I am still trying to figure this one out

 

[Graybeard]

Bottom line:
$22/222 clicks [site stats]
.0990
About $0.10 per landed 'click' to the money site --no results (yet) usually takes time
so the funnel:

3247 IN
222 at the end of the funnel
222/3247
.0683
or 7%| 93% garbage ....

 

[PureLander]

Very interesting!

however the coding to do this took maybe 20 hrs and and was a real headache!

Impressive! It'd definitely take much more for another one

- Samira

 

[Graybeard]

No it wont, it's copy and paste code I have developed and now own
This code is based in part on previous click tracking logs and the URL referral parameters I have used before.
Had I stared from scratch, without that prior experience, it would have taken 3 times as long maybe.

It's like a ladder --where the hell is the top ???

 

[Graybeard]

there's an error apparently ...
database failed there was a burst of traffic
grep -c ',a,' ../clicklog/clicklog.log
95
grep -c ',b,' ../clicklog/clicklog.log
81
grep -c ',c,' ../clicklog/clicklog.log
78
grep -c ',d,' ../clicklog/clicklog.log
101
grep -c ',f,' ../clicklog/clicklog.log
88
grep -c ',g,' ../clicklog/clicklog.log
75
wc -l ../clicklog/clicklog.log
843 ../clicklog/clicklog.log

/*******/added
/*********************/
grep -c '[a-z],[1],NULL' ../clicklog/clicklog.log
145
grep -c '[a-z],[2],NULL' ../clicklog/clicklog.log
136
grep -c '[a-z],[3],NULL' ../clicklog/clicklog.log
145
grep -c '[a-z],[4],NULL' ../clicklog/clicklog.log
180

well backup logs work too

Looks like some of the pages got refreshed maybe? to see the next video the page version stays the same a lot ...
weird

 

[Graybeard]

I fixed it
LOAD LOCAL FILE > the clicklog.log flat file backup

Now I will just cron the local file backup and add it to the database every night.
I got blasted with 14,000 pops in a few minuted --the server choked on all the mysql in real time
forensics but i new that would be a real possibility (<rolls eyes>)

Ugliest Page Colors win Version 4
Best video was a 3 way FFM (@_@) typical lolz

 

[Graybeard]

I had;
23,380 ../clicklog/traffic.log-temp pops served
( (118+296) /23380)*100
1.7700 % made it to the cam sites
-0- return so far
spent $40 or so testing
so about $0.10 landed per referral -- not bad so far

This pop traffic is just a big numbers game
my real cost as far as referral acquisition is
1.55/.017700 =$80.7570/CPM
if 3 in 1000 end up buying (in the end) I make $270-$340
so that being doable it's worth doing

Last time I did this I broke even in the end (if you recall)

 

[Graybeard]

This explains a lot --GA has some value
/***added***/
wc -l ../clicklog/traffic.log-temp
23380 ../clicklog/traffic.log-temp Gross received traffic +
../../logs/access.log|grep 'GET /1000/?a=' |grep -c ' 200'
23403 Allowed traffic (200 OK)
../../logs/access.log|grep 'GET /1000/?a=' |grep -c ' 302'
1549 Booted off the server as garbage
(5083/23380)-1
-.7826
78.26% bots or users that immediately dismissed the pop (hard to tell which)
still @ near $0.10 per referral acquisition --acceptable --that is what matters to me
/****end add ***/

(118+296)-25 (25 is my testing??)
389

389/931
.4178

• or a 41.78% 'engaged user' CTR

Engaged user meaning a person that looked at my pop under ad instead of just closing it as a nuisance
931/4411
.2110

• or 21.10% were engaged --that's pretty good for a trash ad

ANY MORE EXPENSIVE ADS NEED TO BE TESTED LIKE THIS FIRST
Before you buy Native, Facebook (other social media) or SEM (like Google or Bing search ads).
You want to know the ad and its design and copy have decent user affinity.
You don't just do what you like and hope the other person will like it.

What's next?
Add 2 more ad networks traffic on this test (I used 3 --1 toolbar search --2 pop ad networks) 5 total
Pivot on the ugly blue #4 design and make some variations for the next tests.
This is an ad serving subdomain --no SEO interest at all ...

The www.'sexcam'.com (parent domain) is the affiliate money ploy and the SEO (hopefully) domain.

The version test is really for that domain to design for better user engagement
There is a master plan here ...

 

[Graybeard]

20 min and a headache ... 0:00 query execution

I got 10 links/ and 4 versions to make some sense of --who said understanding affinity was easy

Shown is the percentage of link position clicks by the design version ...

 

[Graybeard]

now I can see can you

Affinity by version color^^^

overall link `niche` affinity

 

[Graybeard]

better graphic 'tool' representation

The LEAST green wins heat mapping

However, because of the diversity of the niches version 2 and 3 show better results when aimed at smaller affinity niches.
total sample size 978 recorded click outs why were there ONLY about 400 recorded at the offers?
is there some session time to count or is there a massive shave?

 

[Graybeard]

found the tracking issue and fixed it now I am at better than 80%

100000-1-100 to 429496-7-290
$random_int = random_int(100000, 429496)
user + ad network + campaign --tracks the specific user referral/conversion

today's campaigns are tracking but no results --0-- again
the sample is way too small to tell really.

same as for the mobile v. desktop
more click-outs may change the averages or the format and the UX is very different.
this is why I am doing 2 LP one desktop one mobile (tablet is really either).

Not too bad for a pop ad but 90%+ are never seen real views maybe are costing $0.017 to $0.022 that's still cheap

7 hrs ttl of traffic?

 

[Graybeard]

Ad Network 1:
7.71/23 = $0.33521
Ad Network 2:
0.39/7 = $0.05571

$cost/Real CPC landed [recorded]
*appears in the offers stats (relative) to the same offer and campaign

---

Ad Network 1:
7/1080 = 0.00648
Ad Network 2:
23/11026 = 0.00208

clicks/my server access.log (gross requests)
*^ same comparative as above

---

The object is to determine the campaign efficiency between 2 traffic sources.

This has cost maybe $60 or so --only tangible here is assessing data before buying higher priced network ads
Plus a whole lot of my time ... that's the big money here to me ...

WELL ALMOST

 

[Graybeard]

We change the traffic and the GEO --it's another story now <==scratches head

Mobile affinity is a lot different in general and half the traffic was non-US
GB was the other part

Today;

17 hrs

1. =>7260 all traffic
2. ==>6374 my traffic allowed/received count
3. ===>2200 G-A Sessions
4. ======>110 my clicks out

^***10 clicks out were from one IP address that was omitted

---

2200/6374 ==>34.515% user affinity for a pop they couldn't close --the porn cam couple video lolz

110/2200 ==> 5% click out rate to the offer is good --most of them watched the cam couple getting down that keeps looping?

The video tactic does work I think.
My vids were cut to 30 sec or so --like a TV commercial is --attention span?

 

[Graybeard]

this will work better than

[GA4] Path exploration what a joke!​

then I can track individual users by URL actions & requests
I own the code and it cannot change --unless I change it.
Not arbitrary Google changes --sick of that.

 

[Graybeard]

Found an easier way

Network 1 mobile pops mostly GB

Network 2 all US

That's apples and oranges but that's a 100% difference US bid was twice as much
but I was 80% of both network's suggested bid price.

And and one cam site I can track the hits
tracking--------------time on LP sec------campaignID----------Cust ID--------------IP-------------------------------------------------------------------------link ID

 

[Graybeard]

$50** ad spend + $1000 of work --who said ads were cheap? lolz
**one campaign was desktop domain (keyword injection--toolbar) and was not included here.
Creatives, LP design, Tracking programing
Well, at least I own the code and the data ...

network_count​	traffic_count​	over/under ct​	campaign​	LP Out_count​
	My server Logs 100% control​
4292​	3087​	72%​	3201​	79​
17312​	23373​	135%​	3102​	969​
144​	37​	26%​	2110​	8​
4031​	125​	3%​	1201​	41​
21,487 ​	26,622 ​			1,097 ​
		Landing Page CR​	4.12%​

LLL
Tables

4292​	3087​	72%​	3201​
144​	37​	26%​	2110​
4,436 ​	3,124 ​	49%​	bot-rate​

LLL
Tables

campaign​	Cost USD​	CPC-LP Out​	Affinity​	Description​
		human/bot​
3201​	$ 27.39​	$   0.35​	2.56%​	Desktop-var-test POP US​
3102​	$   5.17​	$   0.01​	4.15%​	MobiPop US​
2101​	$1.97​	$   0.25​	21.62%​	Desktop-var-test POP US​
1201​	$7.41​	$   0.18​	32.80%​	MobileContextKeywordPop​
	$ 41.94 ​	$   0.19​		Average cpc cost​

 

[Graybeard]

At the end of the day net costs matter to me.

Most of this traffic does not create a session --so it's either click bot or bounced pops
No conversion results to free registrations or customer spends --so the traffic was a fail
Cam offer #1 451 clicks
Cam offer #2 163 clicks
Sample: 614 REAL REFERRALS COUNTED
Nothing but a failed experiment --possible reasons;

1. War --Russia invaded Ukraine (uncontrolled event)
2. Economics? That should not effect free registrations (more that 12 expected, max maybe 50?)
3. Low quality traffic --most likely reason.

Time out for now

What is gained?​

• new tracking server scripts that will be reused
• new traffic MySQL database to be used for tracking
• new solution for GA4 stats changes (@_@)

Every out click has a title
I cam out |$link_clicked|user ID (xxxxxx-adnet-campaign-code)
These values are gotten with dynamic linking --in that aspect this was a success

SELECT  SUBSTR(cid,7,10) = network,campaign
SELECT  SUBSTR(cid,7,8) = network
SELECT  SUBSTR(cid,7,10) = network,campaign

mysql> SELECT  
    -> SUBSTR('4000001101',1,6) AS click_out_ID,
    -> SUBSTR('4000001101',7,1) AS ad_network,  
    -> SUBSTR('4000001101',8,10) AS Campaign;
+--------------+------------+----------+
| click_out_ID | ad_network | Campaign |
+--------------+------------+----------+
| 400000       | 1          | 101      |
+--------------+------------+----------+