Proposed Project Developments July 16, 2007Posted by alwilliams in Uncategorized.
Two key areas of development became apparent following the critique of my AI project.
1. Strategic Development:
As pointed out in the forum regarding the previous AI;
“…it [the previous AI] was missing some business-like data analysis. For example: an analysis of sales, incremental sales etc.”
(Jun 25, 2007 11:20 am; http://www.cemp.ac.uk/macmp/forum/viewtopic.php?t=342)
This is a fair point made for the past AI, but the intention was to set up the structure and tools to be able to begin data collection on a long-term basis and provide the data for such subsequent insight. As a result it is particularly apparent at this point in time that I need to start to look more closely at the sales data collected via the sales forms implemented in the last AI. The data collected needs to be deeply analysed to understand whether there are any underlying strategies that might help optimise each sales campaign to improve the ARPU metric employed. This is the most relevant development of the past project, which essentially facilitated this next step in developing our sales effectiveness at Habbo. Specifically this will allow me to begin to understand;
• Why the results occurred as they did
• The primary variables that influence Rare sales
• The optimum length of a sales release
• Any long-term trends that might result from an approach that was taken
Collectively these areas will allow future strategic development of our sales campaigns, and provide unique insight into the sales within and structure of a virtual economy, which is the primary feature of Habbo’s business.
As pointed out in the AI (www.cemp.ac.uk/macmp/forum/download.php?id=215), it is difficult to compare the sales effectiveness of individual items (e.g. comparing two sales campaign’s ‘performance’) as they are all inherently different to what they offer. However it is this comparison that will provide the insight and understanding for developing the sales strategy further. Sales campaigns differ in the nature of their release, as each need to be made relevant to the environment and context they are released into. For example the sales of the first item in a range (e.g. there are 10 dragons in different colours) has been largely seen to perform better than subsequent releases in the range as at that time it is unique. Similarly an item that has interactive functionality often performs better than a static item. The fact that there are numerous variables at work has been identified in previous cycles, but the extent that each has an impact and whether they can be manipulated using certain strategies needs to be investigated. As a result it may be necessary to make numerous comparisons on different bases in order to build a full picture. A future AI in this area should mobilise separate cycles that look at each of these variables, as well as combinations of different variables to look at whether and the degree they may interrelate.
A potential problem with this area of development however is that it is unknown how many sales forms are going to be needed before a comparison analysis can be undertaken. There may not be enough sales campaigns to make a useful comparison for the foreseeable future and so there is a level of uncertainty as to when this will be possible. Yet on a similar note the analysis may need to begin in order to establish whether more data is required.
This development is particularly significant in the context of change management, since it requires relatively little time and resource to achieve in itself. Certainly it requires less effort and resource than the previous AI and has few elements to organise/ manage. Scheduled meetings will need to be arranged for the sole purpose of analysing numerous sales forms. Initially this would have to take a wide look at the information and then scrutinise points of interest as they arise.
These will require members of the team to bounce ideas off and to maintain different perspectives from different functions in the business. However the emphasis is taken away from collaboration here since it ultimately requires a business analysis, which for Habbo comes under the remit of the marketing manager, who is also the author of the project and instigator of the change in the first place. Thus, issues of ownership and power will not be a problem and means that as a future project, this development will be relatively un-complicated to manage and run.
2. Metric Development:
Another area that warrants further investigation is the implication of using ARPU as a metric and whether this is of value when comparing our sales campaigns and their relative effectiveness.
“Are you going to use ARPU as a sales target in the future? This would allow you to compare and contrast campaign effectiveness.”
(see Jun 25, 2007 11:42 am http://www.cemp.ac.uk/macmp/forum/viewtopic.php?t=342)
Currently the adoption of the ARPU metric has not been tested fully in its ability to provide a useful comparison between the sales campaigns we have run. Whether it is of use or the most useful metric for all those in the company who need to refer to such data or be able to understand the impact the project is having on our bottom line will only become fully apparent when analysing data with this metric. This metric also needs to be valid in its measurement of our sales performance and would need to monitored for any follow up AI to ensure it is a useful metric. It is problematic if it is discovered later that ARPU is weaker than anticipated at meeting these needs and different data is needed for such an analysis. Yet this situation would have been unavoidable in adopting a new metric as there was nothing to compare the newly generated results with at the time of conception and so a risk needed taking. However as sales volumes and revenue information are being collected for both paying customers and in relation to the entire population, it is more likely that a different metric can be generated with this existing data. So, to an extent the first development (of the actual strategy) should perhaps be embarked upon in order to understand whether further development is required in this area. Otherwise there are methods for assessing metric reliability/validity, but this is not the practical approach that is most useful for AI.
As sales data is being captured as well as ARPU information it would also be interesting to begin to look at these two metrics objectively and individually to see whether the ARPU is a specific improvement on the sales volume metric. If sufficient evidence is found for this a case will need to be made to present these findings to the rest of the company and to attempt a wider degree of change within the organisation. A proposal would be put forward to adopt this metric as the default metric for future sales campaigns or at least to use it in addition to sales volume as is used currently.
“The beauty (!) of AR is that you can just go again and try tweaking something else…”
(see Mon Jul 09 2007 7:13pm; http://www.cemp.ac.uk/macmp/forum/viewtopic.php?t=342)
If the above scope of investigation was attempted, another full AI project would be necessary in order to effectively manage the workload. If this area did not warrant a full AI, in line with the values of AI (continuous development and analysis of all elements of the AI cycle) it is still important to be aware of the issue surrounding ARPU and its adoption in this context. It is important to be aware of whether another more transparent or useful metric might be identified and to test its integrity as it is being used. This might make up a smaller cycle within a larger project, particularly the one identified above, since both of these areas are interlinked with the same ultimate aim of improving our sales effectiveness. Essentially ARPU would need monitoring if this project were to moved forward top the next phase, but what level of inquiry it warrants would need to be considered in line with the resource and time available.
However the resource required to execute such a project development again is minimal compared to the implementation of the last project and would require less personnel and more reflective thought to work through the data. Again, it is important to include other colleagues to obtain a fresh perspective and validate assumptions that might be made. The time scale of conducting an investigation into ARPU is not particularly sensitive, since waiting for some time will only provide more data to analyse. Although granted if it was subsequently identified that the ARPU metric is poor, then this prolonging would translate to a waste of time and resource.
To conclude both of the above proposed developments are relevant and natural extensions of the previous AI, as well as being based on the same premise of; What to do (investigate how to improve sales strategy), What to measure (investigate usefulness of ARPU), How to do it. Yet as there is an overlap between the two, a combination might be attempted if the resource and time warranted. This might also be the case since the time consuming and complexity of introducing the structure and tools for the previous AI has been complete and provides the structure for these AI developments.