Home > Questions and Answers > General Questions

---

How To Go From Data to Information? (2)

---

Hi,

I hear it all the time you need significant data. Invest in it to learn from it.

Gaining data is easy. Going from data to information is not as easy as I thought.

Deciding whether data is significant or not is a lot easier with a statisical significance calculator stackman provided us, but still I have some questions...

- When do you know a placement is not worth the time? To what data must someone look? it could also be you just sucking
- How to determine which factors influenced the significant difference in metrics from day to day? it is easy to blame the day but how do you know for 70% sure that it is
-When do you can conclude that an offer that converted previously is now shit? How to find the factors influencing the consistent conversion drop? to what data must one look? is it possible that because other affs are bidding higher and promoting the same offer canabalize your conversions? is it possible that emails from offer are i the spamfolder and never see daylight ?
-When do you know that a placement is likely not to become profitable at all? im always blaming the lander the offer or the banner ctr but how to determine if it is just the site or placement that is causing the jojo effect in stats?
-when do you decide to put a campaign in the trash and start a complete new one?
-how to determine that a placement is the property of a bigger aff/brand than you and just the fuck up? at what bidding price do you stop assuming an average ctr? 0.5$ ? 0.6 $ cpm?
-how much data is needed to assess whether a banner ad has an off day or is just done? what is a normal life time cycle of a banner ad generally in terms of page views?
-what looks like a solid bidding strategy? is there any? if you bid high you get traffic if not you get shit on for example exoclick. whats the usefulness of a strategy?

I mean data is easy to gather but sometimes the wrong decisions are made based on this data...

this are some questions Im wrestling with and some are certainly stupid and thats why i would like to ask you whar are the MAIN numbers to look at and take action on and what assumtions are generally safe to make.

---

My usual approach to this question is that it doesn't matter whether the placement's terrible in general, or just terrible for me. If I've tried hard to make it work (with a number of widely different angles) and failed, it's still not something I can make money from. There are so many places out there that I can make money that I'm not going to obsess on the ones I can't.

Try not to use data from a single day. If conversions sucked over one day, it's probably the day. If they sucked over a week, it's probably not the week (unless it's Christmas week or something similar).

I'd normally abandon a placement, testing combination, or similar after I've tried several rounds of optimisation on a number of different angles and offers. On any traffic source there are usually hundreds of different potential placements, targeting combos, etc. Mostly, the mistake that I see people making is being too narrow in their efforts and not testing broadly enough within the limits of their demo and their chosen traffic source.

By and large banners take longer than you'd think to burn out. One day's low stats is probably just an off day. One week, not so much.

---

Home > Questions and Answers > General Questions