Back in November I wrote my first advice for entrepreneurs article and it’s taken me 6 months to write this second article. I’ve been busy trying to make money from my business and one mechanism I’ve explored is mobile advertising and I want to share some of those experiences here.

Firstly I have to say that I’ve never seen mobile advertising as a main stay income for my business.  It has been consistently disappointing at the levels of income it has delivered, but at the same time it has been a welcome background trickle of small amounts which can keep the business going whilst we search for a real effective business model.

Secondly as well as being consistently disappointing is also been consistently inconsistent.  From one day to another I see massive swings in the income coming in and I’ve struggled to understand why that is.  The problems are that it’s a complicated area.  My service has changed its geographic spread over the months and this has changed my income.  In general Europe and the US can pay 10 times more per click than India for example.

I quickly realised that using a single ad provider like Admob was non optimal.  The initial issue I had was dramatic changes in the fill rate from one day to another.  This led me to adding a 2nd provider to place ads when the 1st was unable to give me an ad.  As my service grew in volume I added a 3rd and a 4th provider because each was failing to provide sufficient inventory – even now 4 is still not enough. 

The next issue I saw was that as my ad impressions increased I was not seeing a consistent increase in revenue – it was not scaling.  What was happening as I requested more impressions I was being given lower quality ads which generated lower quality revenue.  The graph at the top of the article plots ads impressions per day against eCPM (the effective income per 1000 impressions).  This graph shows the variable nature of the income and at first sight there seems no pattern to it.  But there are conclusions I have drawn.  I think the algorithms are favouring companies with small number of impressions (eg 5K per day) and also there is an optimum number of impressions I should not go beyond.  Above 30K impressions the eCPM starts to fall off.  

Because of the fill rate issues I had taken a waterfall approach to allocating ad provider (ie if one failed then I allocated the next).  However, because of the 30K optimum impressions I then modified my allocation to only allow my main ad provider 30K impressions on adverage per day and then allocate impressions to other ad providers.  This has the effect of inceasing my revenue with all the providers including the primary one.  That is less page impressions generated more income.

Now I have to give a health warning with these findings.  They may well be somehow related to my own circumstances and/or a flaw in the Admob algorithms (my main ad provider) and this will not remain the same over time.  However, for my own circumstances I have found less can be more and anyone in a similar situation I’d encourage you to investigate their own figures.