Know-How Exchange : Strategy

Firstly, my apologies for not being able to respond to this posting over the last week – I’m afraid that I’ve had some time off work feeling under the weather.

Thanks for the answers though and for the off-forum emails and comments. As I anticipated they are as varied as there are people to proffer an example, so that’s lesson one nailed to the study door:

“Never assume that there is a common definition of what is meant by forecasting” and a further salutary lesson is that most of the on-forum comments refer to forecasting which broadly fits into the projection and planning categories rather than the type which is employed if you want to know where your own business, or the business of most of your clients is going over the next year

The term seems to be used for anything from the following list:

The sales projections contained within a new product marketing plan, a new company sales plan.

The expected sales revenue of an existing product or established company which has been based on the targets they have set.

The sales forecasts based on a variety of econometric measures.

The figures suggested by market or consumer research multiplied or divided by something to reflect the fact that the company can only address x% of the market.

The forecast based on known market and calendar trends.


What is notable is that the type of forecasting which applies to about 80% of all businesses is not represented at all. That is forecasting which uses the sum total of all forecasts which are projected by individuals who have “Ownership” of a prospective sale – ie everyone with a responsibility for making sales and managing the progress of prospective sales through to actual orders from both new and existing customers.

If anyone has examples of this, I’d be pleased to hear them, but I guess that I shouldn’t be too surprised that it has not figured. Whilst most small; and medium sized businesses can’t apply econometrics or complex operational and market research to their business models, they don’t seem to have much of a definition of the methods they do or could employ. If they have set themselves a sales target for the year or have a sales and marketing plan, then the next 3 months forecast tends to be 3 times the monthly sales target, regardless of what is happening on planet reality!

I’ll comment on the individual responses later this afternoon and thanks again for the answers so far.

Steve Alker
Xspirt

Dear Colleagues

Firstly, my apologies for not responding as frequently to your answers as you deserve – I’m scheduled for eye surgery this tomorrow and alternating between wetting myself and looking forward to having 20/20 vision restored – thanks for your astonishing patience and the emails wishing me a successful outcome.

All the points deserve some comment, so here goes:

Linda: That’s a fantastic set of observations from a notoriously fickle industry, where the sales of new designs must, by their very nature, be guessed at. No designer would ever accept you working on a figure which would make the assumption that his clothes will be a sales flop. And no marketer responsible for a brand would be prepared to admit that the volume is going to be lower than desired else, why the hell are they promoting it?

On top of that, you described forecasting by diktat. That’s OK for setting targets as a company needs figures to aspire to – if they didn’t, they would be utterly unable to plan buying, marketing spend, manufacturing and cash-flow.

The failure to accept the feedback though is fairly typical across a range of industries and most sales and marketing people will insist on figures which are larger than research or reporting indicates because they would be, at best exhibiting negative attitudes and at worst are forecasting their own failure several months before it happens. They cling on to unachievable figures on the basis that something might come up and in order to preserve their jobs – even if it is only for a matter of months. Once the booked figures come in, the marketing people are toast anyhow, so why bring the axe down earlier than absolutely necessary.

The problems you highlighted about the time consuming nature of analysis and forecasting and the delay in getting meaningful figures in time to do something about the manufacturing and supply chain is an example of companies not using the tools at their disposal, even if the consequences are substantial losses and waste of valuable time. Every hour you spend compiling reports which should; to be frank, compile themselves is an hour you can’t spend of doing something which earns money.

The time lags associated with the fashion business are well defined by the lag and latency model developed by Harvard Business School (I think) in the 1960’s. Their example was to split their students into three rooms about 100 yards apart. They were to play the role of executives in the drinks trade. One room would host the brewers, one the distribution system and one the retail or bar sales outfits. They could only communicate their needs, orders and instruction on a paper card which had to be hand delivered – hence introducing a time lag.

Everything started off well with the pubs estimating their likely consumption and ordering the booze from the distributor. The distributors would then place an order on the brewery which would in turn gear up to meet demand. There had to be a lot of guesswork in this process because no-one had accurately defined the market size and how popular their individual liquor stores and pubs would be.

Most pubs under ordered to avoid having stock which would spoil. As a consequence they ran out of booze and had to lay on a larger order than before in a shorter time frame than was envisaged. Meanwhile, the pubs which had run out started to lose custom as there was nothing to drink!

Being hit by many larger orders than expected meant that the distribution company could not meet demand from stock and had to order more from the brewery, even if there was an inherent delay in gearing up the brewer to greater production.

When a set of earlier and larger than expected orders arrived at the brewery, they would try to rack up production to meet the demand as the previous forecast was plainly crap.

The problem now is the lag in the system and the loss of custom from bars which have run out. By the time the brewery can fulfil the order, the demand has dropped considerably. As the bars and the distribution chain have entered into a contract. they have to take what is in effect excess stock.

Surprise, surprise, when the next order is due, it is both late and smaller than expected. Beer is now stacking up at the brewery and in the distribution warehouses, so everyone trims their forecasts.

This process goes on, in this particular model, the brewery alternates between needing three times it’s capacity to supply followed by 3 weeks of no orders because everyone is over-stocked. A boom and bust model is no way to run a brewery. Likewise, distributors can’t hold more stock than they can afford or have space for, so they too alternate between being over stocked and under stocked. Finally the pubs find themselves giving away surplus beer or having to close because there is nothing to drink again. The model never settles down and is aperiodic – so forecasting is a fraught process.

To overcome this, the beer trade through to IT manufacturers and Automobile manufacturers use just in time production where only a minimum of stock is ever held. Kanban software controls inventory and forces the players to adopt flexible and intelligent strategies to overcome the lag.

If you can’t run to that type of analysis then the next best method is to sum the forecasts of all the people in the chain and establish some operating norms. These systems do boom and bust, or oscillate, because there is negative feedback in the system and the equations used are non linear. In other words, it is chaotic, so you can’t forecast months ahead at all. The solution is to reduce lag, agree stock levels and then try to iron out non-linear elements in the buying cycle and only by plonking a human being in the daisy chain of equations can they eliminate nonsensical predictions thrown up by flawed software models.

What amazed me from your post was the fact that mangers could be in denial of the facts when there are tools on the market to avoid such problems.


Best wishes


Steve Alker

More comments now that I can see again! Thanks again for patience and to Carrie for holding the question open.

Michael: That’s a good example of local detail being obscured – or potentially obscured by global level application of marketing. Actually these kind of examples from FMCG operations are superb exemplars of best practice which for some curious reason get totally forgotten about when the same people are asked to forecast from ongoing results.

An example of this is at P&G where these days a team of about 200 operational research guys, mostly with PhD’s in maths regularly shave billions off costs in logistics and add billions to profits in marketing and sales effectiveness.

The same team then used to be ignored when their results lay uncomfortably with the marketing or product pre-conceptions of senior managers – especially when a VP had a special interest in something. This attitude, the same team reported accounted for about 65% of all product marketing failures. It’s a good job that no one had a shoot the messenger policy there as the same team is now stronger and even more useful.

There one area of poor methodology that I am aware of (Along with every single chemical and pharmaceutical company) was to launch themselves into large scale mathematical modelling to link marketing, sales, forecasting and production. As any mythologist would have told you, from about 1987 onwards, iterative finite element analysis scaled up is OK as long as you don’t have non-linearity and negative feedback.

Marketing models and forecasting models have both in droves and manufacturing has it to a lesser amount. As a result, they tried to forecast what was essentially a chaotic system (Like the long-term weather forecast) where the ability to predict the future to plus or minus anything you can think of depends on your ability to know the starting conditions to an infinite accuracy. Even accountants can’t mange that!

More feedback in a moment – I need to recalibrate


Steve

Hi Everyone

Day 4 after the cataract operation and I can see well enough to close this down and award points to everyone who has contributed.

Last comments to follow in a couple of minutes


Steve

Dave (Telemoxie)

Thanks for your contribution. The difference between innovative products and standard ones is not a specific one which I have looked at, though the method of forecasting we employ and then automate does take into consideration the detail of the sales model the company thinks it should be using.

All too often, this is ill defined and as a consequence ends up with everyone very much doing their own thing without any check or management assistance until a sales target is not met. The causes of failure are usually then so far in the past that it is impossible to save either the sales person’s career or to rescue potentially promising sales which have gone astray.

With innovative products this is a case of triple jeopardy. Firstly it is unlikely that anyone has trialled the sales process to death and it is also unlikely that there will be any realistic measures to judge the sales performance of the salesman and thirdly, the target set will probably be both aspirational (Especially if the owner or inventor set it!) and largely based on assumptions and guesswork.

Having worked for over 10 years selling and marketing innovative products, I have quite a few ideas on how to overcome the problems but I have to thank you for making me think about it. As you said, SymVolli software is itself innovative, and a higher ticket item than a PC but where we get to present, we’ve been seeing a cycle of 3 months to close. Part of the reason is due to pre-qualification of prospects, but that said, there’s no way I’d refuse to see someone just because all the boxes we think should be ticked aren’t. My only imitation I set on it is the amount of time I allocate to the curious who don’t yet have a budget.

There’s some really interesting comparisons to be drawn out here and things from my experience which I can really relate to. That might be a problem in itself – just because I find them interesting and meaningful doesn’t mean that anyone else will – I tend to put my self and my own experiences too much into the centre of my analysis rather than being more dispassionate about it.

So that’s thanks X 3 I think!

Steve

Thanks everyoune - I've run out of catch up time!

Garry - sorry anout the unintentional pun - A Bloomer is Victorian for an embarassing mistake. You answers are far from embarrasing and I enjoy reading them in all posts.

I've got to develop a few points you all raised and will come back with my conclusions. Since I started to refer to forecasting cock-ups and losing control, we got 60 enquiires from our last email announcement and that was just for the small test database.

Thanks again so much and double thanks to Carrie for holding the question open until I could stop looking like Amiral Lord Nelson!