Circulation modelling is in vogue, it seems. A recent session at the PPA conference was packed to the gills. They certainly hadn’t been drawn by the star speaker (me), so I had to assume that the topic was of big interest….
Modelling isn’t a universal pastime for subscription marketers though. It seems only the larger companies apply the techniques with any scale. There could be a couple of reasons for that – one might be the time and cost of deploying a robust model. Another, I might suggest, is the absence of commercially available simple tools.
It also seems that there are significant variations in the depth of application – even within individual companies. My own employer, Emap, which uses Alan Weaver’s complex but comprehensive tool, varies significantly in its application. One company might only use the lifetime value projections to justify investment in circulation development with a predicted ROI – another will use the full model to plug into the budgeting and forecasting process.
So there’s no ‘industry standard’. Does that mean this is still a science in its infancy, or just that different companies have different needs?
What is modelling?
Most circulation managers have some sort of spreadsheet they use to provide numbers for annual budgets, or to predict the efficiency of a subs acquisition campaign. Early in my career I created my own in Lotus
(see how long ago that was?). However, self-built models tend to only predict accurately for the very short term. Mine was routinely 10% out in year three for example. Once I knew that of course I was able to adjust my forecasts and get them right, but that meant I had to do more work with the output of the model – not ideal.
Also, I had to adapt my tool every year – always ‘improving’ its accuracy (although not always making any real difference), and introducing levels of complexity which inevitably led to modelling errors, and more ‘maintenance’ work on the model.
Now I use a large and complicated tool which takes a series of inputs from the fulfilment system, asks a load of questions about market and economic trends, and allows me to simulate (for example) the effect of an extra 1% price increase in the Guatemalan market. I enjoy the sophistication – but not the time it takes to make it work properly. And sometimes you forget it’s just a fancy tool for guessing, and start believing the numbers – very dangerous indeed.
Of course, what happens in practice is that my team use only those elements of the model they can easily and quickly deploy – and get immediate use from. They all use the life time value predictor tool – which helps them assess the ROI for their promotions. Few use the full production model.
What’s missing from the market is a simple tool, commercially available and independent of the fulfilment or mailing sectors, to allow circ managers without time to spare, or who don’t work for FTSE 250 Plc’s to ditch their self-propelled spreadsheets (and the maintenance that goes with them) and still have some sense of where their publications are headed. Come on consultants, lets be having ye!
Degrees of modelling
As I said, different bits of Emap use different bits of the model we have access to. The individual ‘degrees of modelling’ are decided by the publication group, and usually chosen because of the resources available, and the forecasting needs of the group.
Given the absence of a simple ‘entry-level’ modelling tool, I’d advise a gradual progress up the ‘degrees of modelling’ road – start with the easy stuff – life time value projections – and find your feet first – LTV projections are particularly useful here for getting buy-in to investment in circulation development – and then move on gently up the levels.
The outputs of your model have no inherent value. They’re only useful if you can do something with them.
I tend to use them to explain vision and direction, to help win investment, and to provide checks against progress toward revenue or volume targets.
Models can provide output in various formats – or rather, the output can be put into various formats. I find these particularly useful for explaining my activities and goals to those very few around the place who don’t really understand subscription marketing. It’s exceptionally useful to be able to show a spreadsheet to an accountant, or a pie chart to an editor, or a big number to an ad sales rep to illustrate the point of a shift in circulation strategy. A good model will provide you with all the basics you need, short-cutting hours of calculation and effort.
I’ve mentioned LTV projections a few times – I’ve found these extremely useful for winning investment. Often when pitching circulation development work to a Board it’s hard to get over the value of spending a large sum of money in the current financial year when the return is at least half in the next. A decent LTV projection will give you an ROI the finance director can understand and value alongside other investments. Indeed, due to the recurring nature of subscription revenues, this tool, when applied to other projects that might be competing for the same investment pot usually shows subs investment in a very favourable light. Many other investments will only bring short term returns.
A good model – if regularly updated through the year – will also tell you if you’re on target for your annual budgets, three, or five year plans. It takes effort and time to do this of course, but the rewards are great. When asked ‘so are you still on target to deliver that revenue?’, you can just print out a graph and then get back to work. No tiresome construction of spreadsheets and graphs, just press a button and then do something more useful instead.
The Case against
There are things to know about modelling though – apart from the obvious commitment you need to have in terms of time and energy. The first is that models are just sophisticated guessing tools. They are a little like the form systems that professional gamblers use – just because the system said that Singspiel should have won the Diamond Stakes (1995) at a canter doesn’t mean he won’t get passed on the line by Swain. Bitter experience that one.
When you look at the output of your model – you need to run a reality check on it – does it in any way resemble something that might happen? Run some what-ifs, and try to be destructive. If by assuming that your subscribers will not be affected by doubling your price your output is unaffected, then your model is in trouble.
And just doing it once is just not enough. You need to set up a series of scenarios to show the variety of possible outcomes – otherwise one or two wrong or over-confident assumptions could send you to an early career grave.
Models don’t replace intuition either. You need that to operate one properly – where else would assumptions come from? So don’t assume you can replace team members with experience for a data monkey with a sophisticated spreadsheet.
At the risk of repeating myself – you need to choose the right degree of modelling for you, based on your experience of using such tools, and your needs in terms of output. Going ‘over the top’ will be a horrendous waste of time, and therefore money.
So, finally, modelling can help. Yes it can. It can help you understand your circulation base, it can help you explain what’s going on to the people you work with and for, and it can help you predict what might happen in the future.
But – it takes time, it takes effort, it takes understanding (for which you’ll need training). If you’re short of these things – don’t start. You won’t enjoy it.