My Journey from scratch (17.08.16)

I have been a member for a short while and in the affiliate game for a number of years. My day job is heading up the Clickwork7 Network that until now has done very nicely being a very traditional affiliate network with traditional pubs and a good range of direct finance, survey, mobile and gambling clients.

The owner of the company has been media buying for around 7 months as he was obsessed with the potential of this side of the business. Until now, I have been pushing back on his desires for me to get involved as my team needed me day to day and I knew this would consume me!

I have now got them to a point where my time is a lot more free and could push back no more. As a learning experience, I knew that working with him on his journey and adding my opinions would be invaluable.

In just one week, I have taken so much information away and have actually become a little obsessed with the detail. This thread is simply a daily update of what we are doing and the results. It will undoubtedly bore some people as I am a complete novice however I wanted to document our journey somewhere other than my evernote!

Day 1 - 17.08.16

Following a sit down with my boss, we have decided to start with a few key steps...

1) Login to Adplexity and filter by

Geo
Longest running
Popup

2) Get a feel for whats out there and look for sweep landers to try. (Choose 3)

3) Go through a couple of his popular networks and see what is available in...

India
Thailand
Vietnam

Obviously at some point we are going to have enough data to optimise the campaign. A key thing within this process is to look at the traffic sources and blacklist the ones spending your cash with no return. (I have since established there are other things to tweak first and not to be too eager with this)

We are using Voluum for our tracking and for our initial tests, popads for traffic. Both platforms seem pretty straight forward but I'm sure they both have their complications that I will discover later!

ITS GAME TIME!!

There's me thinking I would just dip my toes today and get to know the system...Noooo, that approach is for pussies so I'm told and we jump straight in with a campaign my boss has been looking at for a while.

THORS HAMMER - (You can guess where this is going)

Mens health is a vertical that I have been lead to believe has legs if you get the right geo and test, test, test with your ads and LP's.

So we choose that as our vertical, Thailand as our geo, popads and adult as our traffic source and Adcombo as our network. As a newb, Adcombo is good as its easy to test ad and lander with a very simple interface.

With a few clicks, a top up of budget and lots of vigorous notes from me, our offer is live with the 2 pre landers and 1 lander that the dorris at Adcombo suggested were the highest converting. You have to start somewhere right!

What have I leaned today?

1) Split testing initially is vital. 2 - 3 variations of a pop lander
2) Tier 3 countries can be harder to optimise as they are often phone conversions rather than C/card. This means there is a delay on conversions by maybe 24 hours
3) US, UK, AU are all tier 1 geos - very expensive with a huge amount of competition
4) India do not seem to run many sweeps
5) Most of the tier 3 geos are heavy in dating (or porn if we are being honest)
6) Most networks in this space are pretty easy about what you are running and few need to check your ad copy
7) A lander in pop ads is the initial pop that a user sees. (A lander in traditional marketing is the final offer page) This can be ripped and hosted to include the various styles and angles you want to test
8) Testing early is key, stop reading and get playing!
9) Accept you will lose money and set small goals
10) Voluum can be used like Cake (Who I use for C7) to track all campaigns as long as there is a global post back with that network in place
11) Much of the offers in tier 3 geos are the same. Lander and offer page rarely have anything that ties them together like a typical affiliate campaign

And breath.....

If you are laughing at some of my assumptions, that's cool, I understand I am a novice but one day, with a little help from my friends, I will know my shit!

 

Day 2 - 18.08.16

Now have access to Adcombo, Popads and Voluum so I can go over the data from the first 24 hours from the comfort of my own desk.

So what do I take from the 2 landers we have tested?

1) Has a very 'Naughty' image on the front with a guy in the background in just a pair of pants. The image alone looks like 2 people just about to have sex or just finishing which could be classed as mild porn.

2) Has a different top image. This time the girl is a native Thai and looks like she is simply laying on the beach in a sexual but not naughty way. At a glance it could be a photo shoot for a summer wear line.

From the stats, I can see that the first LP which I have labeled 'Naughty' above has a lot more click throughs from the same number of impressions. To be honest this is not surprising given the nature of the sites the ad is appearing on and that men are almost always lead with their dicks! However...The conversions are terrible bringing us to the conclusion that the campaign is mis leading and does not carry a consistent message from pop to LP, hence a large drop off.

The second creative has a lower click through which again is pretty predictable. However, the people clicking this one are more than likely reading the heading and therefore have more of an understanding of what they are about to land on. This in turn means there are less surprises when a campaign for dick pills ends up on their phone.

With the volume of data we have after 24 hours, this is certainly not an exact science yet and could change over the next 48 hours but early signs suggest I am right with these assumptions.

 

Day 3 - 22.08.16

After a nice 3 day weekend with the family, I return to find out that on the Friday we had made some changes to the current test. Following my assumptions on the 18.08, the lander with the higher conversion rate continued to pull away, meaning that lander 2 was paused and eggs were placed in the 'Less Naughty' basket.

At the same time, based on our returns, we dropped the bid rate from $0.0007 to $0.0005 to see how it would effect margins vs traffic quality.

We now know without doubt that the lander with the sexy chic in the bedroom and a guy in his pants was more of an instant hit with the viewers, however not for the reasons we wanted, hence the drop off on the offer page.

With 2 full days of data, our ROI went from -77% to -49% following the change to go with only 1 lander. At this point we knew that we had to double our conversions just to break even but we were not expecting miracles. There was a hell of a lot of optimisation to be done before we could think about break even but we were showing signs of promise.

We left this running but reduced the bid to $0.0002, just to see if it would do anything.

I WASN'T HAPPY WITH SOMETHING!

I know the affiliate manager at Adcombo had recommended this lander and at first I was happy to go with that but I was not convinced. Yes, I am new to this but I do know user journeys, I understand how people think and I am very good at understanding why a user would convert on an LP.

This lander just didn't match the traffic source in my eyes and we could not determine anything with this test until the other landers had been tested.

So we did!

The 5 landers in Adcombo that had not yet seen any action were set up in rotation, going to the SAME offer page as the other 2 tests. (There are 8 others to test). Setting this up was a piece of piss however...Because we were testing 5 landers and direct linking through Adcombo rather than ripping and hosting, we could not track individual clicks on each lander which was an issue for me.

That said, we set is up and hot GO!

 

Thanks Tony, appreciate it

 

Day 4 23.08.16

After initial fears that the rotation campaign was not returning well, today has seen a total of 7 leads validate, leaving us on a - 34% ROI in Voluum with an extra lead in Adcombo that has not yet been validated

Today in isolation is seeing us at + 27% ROI which is great but will drop and the leads that are pending stay in pending for 24 hours (ish)

Our original test, dropping the bid from $0.0005 to $0.0002 has proved a step too far as it was not high enough to actually get any traction and therefore spent nothing. This was no problem as I had what I needed from this test to move forward.

Onwards and upwards and my eyes are now firmly fixed on the Rotation Test we started yesterday...

With just 24 hours of data, we had one lander outperforming the others however we are still talking small numbers and at this stage I was not happy to make a call on cutting the other 4. I needed another 24 hours to see if this one would pull away. This particular lander is very in your face, graphic and 'Selfie' styled which some may say looks more authentic. I did get a few looks though when people in the office saw Tarzan with his massive cock out all over my 22" screen!

In the mean time I wanted to get a head start on the offer pages that we have available and determine the differences. Here are a couple of examples...I was in for a big shock!

#1 - OMFG my eyes! This is hardcore from the first pain. Initial images are pure porn and pretty serious porn at that. It has a real 'Home Movie' feel about it which makes it feel a little more genuine that a photoshopped image of a supermodel with her bangers out. Its extremely seedy and gets no better as you scroll. Rude bits flying everywhere!

They are also utilising the 'Comments' section like many other campaigns of this nature. Giving the impression of genuine 'Customer' feedback about this product which we obviously know is bullshit.

#2 - This is a lot more professional with HD images of extremely attractive men and women and a clear CTA box high up. This looks like a much more traditional campaign and one you might see for Durex or Love Honey. It stays consistent throughout and does not use any of the tricks such as a fake social media feed. Is it right for our winning pre lander? I don't think so

#3 - Is extremely tame and looks like any other ad for health pills. The comments from previous buyers looks fake but to an untrained eye could be a good validation method. It is simple to follow and takes you through the order process step by step which could be a good sales aid. Does it match the theme of our pre lander? Hell no!

#4 - Seriously, am I on RedTube? This is probably the worst one and like the first, has a very hard core porn theme. There are testimonials from previous users but no social feed. I think this one has gone too far with some of the imagery and could just cause a distraction to potential buyers rather than enticing them to buy.

I will be back tomorrow with day 5 and the latest from our rotation test and BIG TARZAN!

 

Day 5 - 24.08.16

Today we have seen 9 conversions from our rotation test and 2 still pending. ROI currently sits at -40% which is actually the best of the 3 tests and could get better with the 2 pending. There is a stand out pre lander currently with 4 conversions but for me its all about the LP now.

In isolation, the pre lander with 4 conversions is in profit based on the impressions split 5 ways across the different pre landers

BIG PROFFFF!

We have now taken the 3 highest performing pre landers and created a new campaign (Our Third) that rotates through the 3 to the same LP as before. This was started at 10.30 today and at 15.00 we are yet to see a conversion and have spend $24. This may well mature later into the evening because if we look at the conversions from the previous campaigns, evenings have been stronger.

Scrap That!!!!

I have just checked the original campaign we set up on the 17.08 that had 20 conversions prior to any optimisation and the bulk of those were delivered between 6am and 7pm which throws out what I just said! Bollocks!

Basically people in Thailand are watching grot in the middle of the day, what a life!

Another 24 hours for this test will hopefully see some results we can work with. These are the best 3 landers from the second test so I would expect to see something to analyse in the morning

 

Day 6 - 25.08.16

The Rotation Test 2 (Third test overall) has now been live for 24 hours and after an extremely slow start, we did start to see a couple of conversions coming through and are now sitting on 4 with 6 pending. The average conversion rate on this offer from pending to conversions is 33%. We have 6 pending with 3 already canceled giving us a total of 9 pending so far today which on that above average means that 3 should convert giving us 7 leads.

We have enough budget and still not enough data to start cutting sources so we are letting this run again today to see if my prediction is right and to get to an optimisation point.

One thing that has not been tested and I am keen to see if we can is a LP rotation. For me, this is the most important part of the journey as its the bit before a user says YES! So far throughout every test, we have only tested one LP which Adcombo said was the best. Personally, I am not convinced!

What am I seeing when looking round Adcombo?

In traditional affiliate marketing, the link from ad to LP has to be consistent if you want high conversions. If people click a link and the LP looks nothing like the ad they have just left, you can guarantee a high drop off.

What I have seen so far in my very limited time on this side of the fence is that consistency between ad and LP means absolutely nothing as there isn't any. Very often, affiliates are ripping and adapting landers to create an angle for a click with little or no regard for the LP and how the user will flow.

Maybe this is due to the nature of the offers we are talking about here. Dating, skin care, pills for this and creams for that all capture the user as point of initial ad, then work on the cycology of that user and the inevitable feeling of 'I must have this in my life because I am vain and insecure'

Another thing I notice is the amount of landers now utilising the appearance of social comments in there pops, giving the impression of real consumer validation. We all know this is bullshit but would I think that way if I was not in the industry? Probably not!

We are leaving the test running over night into Friday and will make some big decisions on it for the weekend tomorrow...

 

yes mate. Using Adcombo they are already in Thai but if we create our own they are all translated and written by someone that understands the native way of speaking.

 

Day 7 - 26.08.16

So my prediction of 7 validated leads was spot on and as of 8.34 this morning the campaign ran out of budget. We have 4 still pending approval so a couple may still drop but amazingly this test has performed worse that the one before with ALL landers on rotation We are down -54% ROI on this vs -40% on the last test

Looking at the data for this test in isolation, the temptation would be to cut some sources now however before I do I wanted to compare the same sources on the previous test to see if my decision is made easier...

Of course it F%$k*&g wasn't!!

Looking at the 2 tests, there are sources that performed on the first and not on this one and visa versa. I have a couple of big spenders that are consistently crap but not enough to cut today. I also have no clear winner on lander this time with a 2/3/2 split of conversions across the 3 landers.

We have decided (or my boss has because its his money!) to top up the budget and let it run over the weekend with the thought that this could be prime time for this type of offer.

I am still very much wanting to test the remaining offer pages and have got a commitment from him to do this when we return on Tuesday following a 3 day weekend and time for the pendings to convert.

I will be back on Tuesday, have a great weekend!

 

Day 8 - 30.08.16

So my long weekend expectations have not panned out and the campaign only saw a further 5 conversions, leaving us with a -59% ROI overall. If I drill down, the best lander still only saw a -50% ROI and there were 2 landers with the same number of conversions.

As much as the temptation would be to throw the towel in on this one, we still haven't tested any of the the other 8 offer pages and therefore do not know if our current lander works best for this traffic.

I was hoping that following my analysis of the variables I would have a clear idea on who to cut. Surprise surprise, only one has more than one conversion and the ones on this test that have none, had one on the last test! I think I have picked out 2 that can definitely go but not sure whether to cut them before the LP test?

Out of budget now and the boss it out of office so tomorrow is another day and I am going to pin him down on the lander test. Beyond that, I think it's time for me to start testing and building my own campaigns

 

Day 9 - 07.09.16

After what seems like a year (actually about a week) my boss and I are back on it with a bang today. My first test still needs decisions made around landers, offer pages and variables so I am all over that today.

Beyond that, we are planning to launch a totally new campaign today in a different geo and vertical! This goes against what I said I would do as I only wanted to change 1 part whether it be geo or vertical but this test was time sensitive and we thought we would give it a go to learn some lessons.

Sweeps vertical,
In AU
On Pops
2 landers (ripped and optimised)
2 Offers (Both of which are running on my network)

Lander 1

Follows the 'Test' theme rather than the 'Win' theme. Copy on the original lander found on Adplexity was about as shit as it gets and despite possible previous success with it, I had to change it so it at least made sense and looked like a human wrote it!

I also wanted to strip out references to the brand in question that made it look like it had come from them. It's not necessary and could just land you in a world of shit.

Lander 2

Follows the traditional 'Win' path and goes down the road of gameification with a 'wheel of fortune' style spinner in the middle of the lander that the user gets 2 spins of. To us its all bollocks however I have been around long enough to know not to judge the average user on what I know.

It also has the increasingly popular social comments underneath (Also fake) for extra user validation that the campaign is genuine.

I have no idea which lander will come out on top and much will come down to the offer page and how engaging that is. My goal as ever is to test for long enough to find a successful combo and then scale. My expectation is this will be very tough and it will be about small tweaks and fine margins.

More to come later....

 

Day 10 - 08.09.16

Yesterday did not pan out as I had hoped and the campaign did not go live. Today we have used the beta version of DSlander to optimise our ripped landers, got our voluum links in place and can now put the damn offer live!

1 landers, 2 iphone7 offers from Clickwork7, pop traffic on Popads tracking in voluum. (Second lander will be added tomorrow)

Of course, the iphone7 has now been launched and my temptation was to change the images in my landers...Then I saw the phone and thought, why bother?

Initial budget has been set at $50 a day with a bid rate of $0.05 which compared to my last bid rate is high. Traffic in AU is however considerably more expensive and volume is significantly lower so this looked like a decent rate to get some worthwhile data

I am expecting to make money on sweeps in AU straight off the bat? Hell no! Will I learn a thing or 2 over the next 2-3 days? Most deffinitely.

We are live!

 

I am not mate but thanks for the heads up, appreciate it

 

Day 11 - 09.09.16

So our test now has a little data and as I expected, the results were shocking! However, I have now spread the widest net possible and only spent $102 so I can now close the net piece by piece and see if I can improve on the -100% ROI!

7281 pops
37 clicks (0.005%. My previous offer saw 0.001% but that was a mens health vertical)
1 Conversion

Second lander is now live which is a 'Test' themed lander. This was supposed to be live yesterday but my tech guy decided to go for his 15th meeting of the day! Techies!!!

Duplicate campaign set up in Voluum and new offer set up in Pop Ads with a few tweaks on the original test that include

General traffic only, excluding adult
IOS and Apple only
2 Landers going to 2 offer pages evenly

I have kept the daily budget the same and starting at this time may show a slight uplift as its 7.45pm in AU. Bid rate is still at $0.05 as I did not want too change to much at once.

What am I hoping for here?

An improved CTR with the release of the second lander. Conversions are in the hands of the gods at this stage however I will look at the consistency between lander and offer if I have no decent data to optimise with after this.

 

Day 11 continued...

As I have no idea where the bid rate should be for AU sweeps on pops, today has seen me spend $100 in less than 4 hours!

$50 spent at $0.05 in less than an hours (Zero Conversions)

$50 spent at $0.01 in less than 3 hours (1 Conversion)

Both campaigns have now been paused as I am not taking anything away from these in terms of data to work with. New campaign has been set up as follows

2 Landers
2 offers

IOS only
Apple device only
$0.002 Bid rate
$50 a day budget
$200 total budget

I will update you on Monday