Posts Tagged ‘forecasting’


I spent most of last week in Boston at the Enterprise 2.0 conference where I was honored to be the sales and marketing track chairman.  Next year it will be called E2 Social and will bookend the other conference that has been held in Santa Clara and will become known as E2 Innovate.  There’s good symmetry here.  I can’t think of another purely social show or one focused on innovation.  Most shows today are vendor sponsored which is good but different.

Our track had some cool presentations on social marketing from IDC mavens Gerry Murray and Joe Farentino, revenue performance management from Phil Fernandez, CEO of Marketo, and an intriguing discussion from Pam Kostka, a fellow Crusader and CMO of VirtuOz, a company that makes virtual agents.

If you are wondering, a virtual agent is a software robot that you can talk to regarding sales, marketing or service issues just like a person.  These agents are a happening thing and promise to do away with wait times and improve service.

There were also two panels, one on M&A activity that we put together last minute with the able assistance of Sameer Patel, Josh Greenbaum and Louis Columbus.  As is so often the case with these things, serendipity played a role and caused more than a few people to walk away with the idea that this kind of thing ought to happen again.  Thanks guys, the panel was outstanding and a good example of the talent pool that lurks in the Enterprise Irregulars a group with a low profile (that ought to be greater) and an inversely proportional IQ factor.

The other panel, which I want to focus on, was illuminating to me for an unexpected reason.  I invited some of my brain trust including Thor Johnson, Cary Fulbright, Derek Peplau, Columbus and Murray mentioned above.  Toward the end we had a discussion of big data and someone mentioned a large company that had converted from one CRM system to another and had deleted many years of sales data in the process rather than bring it along and try to figure it out.

Initially I thought throwing away all that data was folly but I came to see it as smart but for reasons that I think are different from the consensus of the panel and audience.  One audience member got the analysis right, in my opinion, when he said simply, “There’s nothing in it,” by which he meant there was a great deal of data but that it was devoid of information content.  How could this be?

Very simply, most CRM systems either have fields or enable you to create them to capture important data like product interest, deal size, projected close date and much more.  All of this is valuable but CRM’s point of failure is that these fields can be overwritten and there is no provision for storing historical information.

Now, you’ve heard my sermon on historical data before most likely.  But at E2.0 I had an insight about the difference between sales and marketing that reflects the difference in the data we collect and analyze in each space.

In marketing we collect data once from a large sample.  If you run a program against a list you collect data from a large number of people one time.  You analyze the data and perhaps discover people who are interested in a product now or in the future and you process accordingly.

Sales is different.  The universe of data sources is smaller but the sources give off data constantly through a sales cycle.  Sales reports — pipelines and forecasts — show a single cross section of the data and they are equivalent to the individual frames of a movie.  Most of the time it’s hard to say much about how a film ends by examining a random frame.  Sometimes you get lucky and the random frame shows the butler with a knife in the in the library etc. and you can make a deduction.  But most of the time you aren’t that lucky.

Unlike the movie, which is a succession of stills projected in rapid succession to give the illusion of movement, the sales forecast is a one and done thing.  Worse, making the report necessarily destroys the old frames.  So, getting back to the company that threw away old data, I would throw it away too.  The old data was simply the last frame depicting the end state of a deal and usually the end state is a loss.

There’s almost nothing you can discover from the end state but if you have all the frames that led to the end then you can apply analytics to it and find out things you didn’t know.  Analytics lets you play the movie back and forth to find the aha! moment.  But you need to keep all the frames.  The point is that in marketing you can apply analytics to a single state of the market but if you try to do the same with sales you’re toast.  Sales data is different from marketing data and so are the ways we analyze it.

In the panel I moderated last week that idea was not in evidence and it shows how we need to re-think and maybe find new people who think differently about selling and sales data.  Without new thinking we’re liable to not be able to figure out the importance of social tools and selling will continue to be a hard nut to crack because it remains more art than science.  It doesn’t have to be that way.


Forecasting and pipeline management don’t get nearly the attention they deserve and that doesn’t make sense.  Of all the parts of CRM, the forecast is one of the few things many companies still leave to manual systems, i.e. spreadsheets.  Even sales compensation has a higher place in heaven as companies like Xactly have blazed a trail away from spreadsheets to a system with a database and analytics, with excellent results.  You’d think that sales people would be willing to invest as much in the forecast as they do in counting their commissions.

Part of the challenge with forecasting and pipeline management is that some professionals might resent conventional forecasting systems for the same reasons they like compensation systems.  Confused?  You shouldn’t be.  Both types of system reduce uncertainty to certainty as much as possible.  But while that’s a good thing when you are counting existing money (your commissions), it’s a problem when figuring out the future because the future is anything but certain.

That’s why last week’s Cloud 9 Analytics user meeting was so important.  At their third annual user conference, CEO Jim Burleigh, talked about the importance of understanding the probabilities when forecasting.  It’s no coincidence that Cloud 9 now boasts a forecasting user interface that uses probabilities but also acts like a sales manager.

If you’ve spent any part of your career in sales then you know there are deals and there are DEALS.  Some deals are like racehorses, they practically sprint from first call to closure while others plod along and maybe even stop.  That’s an extreme situation and it’s easy to spot the real winner.  But consider two deals at a 90 percent completion stage.  They might look the same numerically but each took a different path to that 90 percent mark.  One might have taken twice as long, one might not have enough money budgeted, one may be run by a C-level officer on the customer side the other might be managed by a director.

These differences in the history of the deal add up and a seasoned sales pro knows they are important.  But conventional pipeline and forecasting tools (e.g. spreadsheets) make no use of history, which might help explain why only nine percent of organizations we’ve surveyed have a 0.9 correlation between the forecast and reality.  The rest?  Foregtaboutit.  When it comes to forecasting these deals, the sales pro might favor one over the other for reasons that add up to gut instinct.  So, it’s no surprise that the pros create three flavors of forecast — the best case, worst case and the most probable.

The genius of Cloud 9 today is that they’ve found a way to take the best of what analytics can do to track history and spot trends and combined it with a forecasting user interface that enables a professional to apply common sense to arrive at best, worst and most probable scenarios.  Some people call it gut instinct and I suppose that’s as good a term as any, but really, it’s not gut — it’s applied intelligence and experience that just happen to be hard to put into words.  At any rate, the new forecasting UI is straightforward and looks easy to use and it will remind professionals of their beloved spreadsheets, but with a lot more intelligence behind it.

Getting sales people to put aside the pure spreadsheet approach and go with something with more rigor behind it may still be a challenge.  But Cloud 9 has demonstrated that it both understands the challenge in all its dimensions and that it can turn its knowledge into very serviceable product.  Like the compensation managers before them, Cloud 9 has replaced the spreadsheet with something that makes more sense, is easier to use and should result in better results all around.


I have been studying sales forecasting and forecasting tools a lot recently and I have come to the conclusion that we need better tools as well as better ways of using them.

There is a lot that can be said about forecasting, its current state and how to improve it and I don’t want to leave anything out but I will try to be brief.  First off, how we forecast says a lot about our views on economics.  Given that most of us are not economists, our views of economy are most likely derived from what we see and hear on a daily basis, much like our view of the weather.

For over thirty years our view of economics has been increasingly colored by the ascendant views of the New- or Neo-Classical school of economics.  To over simplify, it is a view that goes back to Adam Smith, of supply and demand and a belief that economics is a hard science governed by equations as rigorous as Newtonian physics — wishful thinking I’m afraid.

The most germane idea for our purposes is Say’s Law.  Say was a French economist, very much in the Classical school who said that “production creates its own demand” and from that we derive the famous supply side economics of the last thirty years.  Supply side economics corresponded nicely with another phenomenon in our world, the introduction of the CPU chip in 1968 and the cascade of new products that ensued over the coming forty years, roughly the high-tech era.

Increasing CPU power followed Gordon Moore’s famous dictum, now Moore’s Law, of increasing CPU power and decreasing cost, and it created a special circumstance that governed supply and demand for technology goods.  Moore’s Law made Say’s Law work like a charm.  A corollary to Say is that all markets clear, i.e. all supply is eventually absorbed at some price — but maybe not a premium price.

Moore’s Law ensured that a fresh supply of technology goods that superseded the earlier generation would arrive and drive demand thus ensuring Say’s Law would operate as advertised.  But if Say’s Law requires something like Moore’s Law to operate smoothly, then it must be said that Say’s Law is a special case, not an iron clad law of economics.

What’s that got to do with sales forecasting?  Quite a bit.  In the special case of selling into a market with undiminished demand, sales forecasting need not be a lot more complicated than determining where we are in the sales cycle.  If we’re ninety percent through the cycle we ask for the order and there is a reasonable chance that we will get the business — no guarantee, but a reasonable chance.

It hardly matters that our ninety percent is not really a probability derived statistically but really just a milestone in a process.  In an expanding market there are enough deals percolating that reasonably diligent effort will result in on-quota performance.  But on-quota performance is not what it once was and forecasting is in disrepute in many places.

According to Jim Dickey and Barry Trailer at CSO Insights, only about fifty-eight percent of sales people manage to make or exceed quota.  Also, according to my research less than ten percent of sales forecasts have an accuracy of ninety percent; the rest aren’t worth the time and effort it takes to compile them.

What’s happening to sales forecasting is not surprising.  With Moore’s Law slowing down and with so many formerly new market niches filled with products, we are transitioning from an era of expanding markets to one of zero-sum conditions.  In a zero-sum situation, if you are going to win business you need to do it by displacing another product.  If you are a customer in a displacement game it is always easy to do nothing and wait for a better offer and continue using an existing product that might not have all the bells and whistles you want but fills the need nevertheless.

A zero-sum economic environment has a lot of uncertainty in it.  You might use the words uncertainty and risk interchangeably but they are not the same.  Risk is something that is unknown but knowable.  If a deal forecast is at risk a sales representative — frequently at the urging of the sales manager — can ask more questions, get more data, and piece together an answer.  There are many issues in sales that are simply unknowable or mostly unknowable, for example, the details of the bid your competition makes.

When uncertainty — not just risk — enters the picture, our forecasting paradigm that relies on milestones in the sales process becomes useless.  We need better tools if we are to forecast in the face of uncertainty and those tools exist but few of us have taken them up yet.  For example, prudent managers might start with the territory planning process.  How much white space is in the territory?  What percentage of that white space is likely to churn this year?  What is the overall economic forecast?  Given our market share what is the probable share of that white space that we can capture?  Is that enough to sustain quota for one or more people?  How should we incentivize them?

Sales forecasting will always be an inexact science but we can do better than we are currently.  We could persist in basing our forecasting ideas on Say’s Law but inevitably it is a race to the bottom, to pure competition on price.  The airlines do that but none of them makes any money.