Jul 28, 2021 | Analytics, Management
Many if not most marketers, approach metrics that seek to increase their accountability with about the same enthusiasm they would approach a snake of unknown species in their backyard.
Warily.
The default has become a range of numbers that might look useful, are ‘saleable’ in the corner office, but usually do little to hold marketers genuinely accountable for the outcomes of the decisions they make.
The most common I have seen are:
- Vanity metrics. Typified by ‘likes’ or number of ‘friends’ on Facebook.
- Measuring what is easy to measure instead of measuring what is important, the drivers of outcomes.
- Measuring activity rather than results. This is endemic in publicly funded organisations.
- Measuring for efficiency rather than effectiveness. You can be highly efficient at doing exactly the wrong thing.
- Concentrating on cost rather than the return that the investment generates. This measure, as does the following one, infests organisations of all types.
- Measuring budget compliance.
Charles Goodhart, a professor at the London School of economics proposed what has become known as Goodhart’s law: ‘When a measure becomes a target, it ceases to be a good measure’
The implication is that you need two opposing measures that drive the outcome you are looking for to use as KPI’s.
For example: We all know that the best lead is one we get from a satisfied customer, a referral. Therefore, it is easy to set as an objective the number of referrals given. Unfortunately, this is very easy to ‘game’.
Sales people are able to just extract any old name from customers, to reach the number. Therefore, it follows that the KPI should be referrals that are converted into a sale. Better, that ensures that the referrals given are genuine. However, it is also flawed, by the simple fact that a conversion can happen for a number of reasons, including a below cost deal.
Therefore, the related KPI should be around the margin, or perhaps customer cash flow, something that reflects the profitability of converted referrals. This will ensure that the referrals are in fact worth having.
Developing KPI’s that are held across functions will improve the flow of information and resulting functional performance.
I refer to these as Tandem and Opposing KPI’s. For example:
- Sales people responsible for revenue should also be responsible for margin, but not for setting the prices beyond a proscribed band. Those who set the prices should also have margin as a KPI.
- Operations people responsible for efficient manufacturing should also be responsible for inventory levels and stock turn. This should connect manufacturing to market demand, and ensure some level of collaboration with sales to ensure stock availability.
- Those responsible for management accounting reporting and implementation, should also be responsible for reducing operational transaction costs.
Marketing is often accused of using garbage maths, fancy but meaningless clichés, and often they do. For credibility this must change.
It is not only marketing that overuses garbage metrics. It is just that marketing is an easier target than the accountants and engineers who have some numerical street cred and get away with it more often.
Having a simple set of cross functional metrics that go to the drivers of performance at any level, that are openly displayed, will be a huge step towards performance improvement.
Header cartoon credit: xkcd. https://xkcd.com/2295/
Jul 14, 2021 | Analytics, Management
The ‘Standard Error’ is another of those confusing statistical terms marketers need to understand. It is often confused with, and is as misunderstood as ‘Standard Deviation’. While they are related, and the Standard Deviation calculation is used in the calculation of the Standard Error, they tell entirely different stories.
The standard error calculates how accurate the mean of any sample from a population is likely to be, compared to the true mean of the total population.
An increase in the standard error means that the means of varying samples of data are spread out, so it becomes more unlikely that any mean of a sample will be an accurate reflection of the true population mean. The higher the standard error, the more spread out will be the population around the mean. Conversely, a low standard error indicates a closely distributed data set, and so is more likely to be representative of the population.
To continue the example in the earlier post explaining Standard Deviation. If you were planning to improve Sydney’s terrible road congestion, it would be valuable to know how representative of the total commuting population of Sydney the mean of your trips from Artarmon to the CBD of 30 minutes was.
To do that, you would do a wider study of the whole population, and calculate the mean, and standard deviation. You would then apply the Standard Error formula to calculate the standard error of the Artarmon sample, compared to the mean of the whole Sydney population.
The standard error is the standard deviation divided by the square root of the sample size. It therefore tells you the accuracy of a sample mean by measuring its variability from the known mean of the total sample.
Header illustration courtesy Wikipedia.
PS. I guess the government could have done such a exercise in parking lots, swimming pools, women’s change rooms, and all the rest. Perhaps they do not understand real statistics when disconnected from political statistics?
Jul 12, 2021 | Analytics, Management, Marketing
Marketers often hear the term ‘Standard Deviation’ during research debriefs, and conversations with operational personnel managing quality. Many do not know what the term means, and in what context to use it.
Standard Deviation is a statistical term that measures the variation in a set of values from the mean, or average of that set of values. The greater the standard deviation, the greater will be the spread of the data from its mean. In effect, it gives you a level of confidence in the conclusions drawn from the data.
Those values can be anything, from the time it takes for you to travel to work each day, to the variation in the size or weight of a widget coming off a production line, or indeed, from any individual part of that production line.
Take your pre-covid commute as an example. You live in Artarmon, 10km’s from your Sydney CBD office. The drive can take anything from 15 minutes to 110 minutes. If you recorded the time taken for a period, say 3 months, assuming you worked every weekday, you would have 130 data points of the time it took to make the commute, to and from work. Assume you took an average of those times, and it was 30 minutes. The Standard Deviation calculation ‘translated’ means that in 68.2% of the commutes, your travel time would be within one standard deviation of the mean of 30 minutes, and 95.4% of the commutes would be within two standard deviations of the mean of 30 minutes.
Let us assume the distribution of the data points led to a calculation of one standard deviation being 7 minutes. In other words, 68.2% of the time you would complete the trip between 23 and 37 minutes. That calculation also results in 2 standard deviations being 17 minutes, meaning that 95.4% of the time you made the commute between 47 and 13 minutes.
Those ‘outliers’ falling outside two standard deviations will be unusual situations. You went into work at 2.00am for a conference call overseas, and you got to the office in 10 minutes, and one morning, there was a ‘prang’ on the bridge, and it took over 2 hours to make the journey. These would be the journeys that made up the very unusual data points in the set, out at 3 standard deviations, within which 99.7% of journeys fell, or further.
This might seem a bit quantitative for many marketers, but if you are to be taken seriously in the boardroom, you need to be able to speak ‘Data’ the language of the boardroom. The typical marketing type assurances based on opinion and theory must be at least partly replaced by the quantitative language of the boardroom.
For those looking for a bit more, there are plenty of resources on the web, and there is a SD formula in Excel which leads you through the steps to do the calculation. However, in principle, the calculation has a few steps:
- Calculate the square of the differences between all the data points, and the mean, then add them up.
- Divide that sum by the sample size minus1, which gives you the variance. The variance is a statistical picture of how spread out the data points in the set are.
- Calculate the square root of the variance, to give the Standard deviation.
As a marketer, you do not have to know the formula, but you absolutely must understand what the term ‘standard deviation’ means, and where it is best used. It might be useful to ‘fiddle’ with the formula in Excel.
Header graph from Wikipedia.
Jul 7, 2021 | Governance, Management
Small and medium businesses have many advantages over their larger rivals. They also have many disadvantages, which centre around the more limited resources they have to get the job done.
One of those disadvantages rarely spoken about, but sadly, present too often, is the lack of robust book-keeping procedures, which can and too often does, enable fraud.
This comes usually from the lack of internal controls, coupled with the ‘all hands in’ necessity of SME’s which means basic financial security measures are curtailed. For example, the recording of both debtors and creditors is in the hands of one person, or even simpler, there is little scrutiny on such items as credit card usage.
The propensity for fraud, and/or misleading financial reports and results can be looked at in a similar manner to fire.
For fire to occur, there needs to be 3 things in place. Fuel, oxygen, and heat. Take away any one of these 3 factors and fire cannot occur. If there is no fuel, it will not burn. Take away oxygen by covering the fire with water, or a fireproof sheet, and it will not burn, take away the heat source, and the fire cannot start.
It is the same in finance.
The 3 factors are opportunity, pressure, and rationalisation.
-
-
- Pressure. For an individual to perpetrate a fraud, there must be some pressure for them to do so. Some situation in their lives that makes them take what they know is a risk creates the pressure to take that step. Debt, divorce, a burning desire to own something they cannot afford, and so on. Then there are a few who are just opportunistic, and when the opportunity arises, will leverage it.
- Rationalisation. The individual must be able to rationalise the fraud, stealing by another name, in some way. They need the money more than the company, nobody will notice a bit going missing every month, ‘I have been underpaid for the value I add’ and so on.
- Opportunity. There must also be the opportunity for the fraud to be perpetrated. The easiest way is to gain access to cash before it is counted, as in retail environments, but the lack of controls in a bookkeeping function or warehouse can have the same impact. I have seen payments clerks set up an account for a phony supplier, generate invoices from that phony, and pay them. I have also seen an order for 10 pallets have 11 pallets loaded onto a truck for delivery and the extra pallet not accounted for. This may not be financial fraud, it is stealing, and the effect is the same. Involving more than one person to perpetrate a fraud makes many types more possible, while at the same time, offering greater detection opportunity.
Remove any one of the 3, and fraud is far harder. Not impossible, but harder.
There are many simple things that management of any enterprise can do to minimise the occurrence of fraud in their business.
-
-
- Separate the recording from the physical. The obvious example of this is the use of a till in retail that records the products and amounts bought, which then must be reconciled with the actuals in the cash drawer. This also records the products being sold which can be reconciled with stock records in a stocktake, further eliminating a source of opportunity.
- Separate the two sides of a transaction or create transactions as a data collection point. For example, the separation of the recording and payment of a debt should be separate to avoid the ‘phony’ customer situation. In a factory, you might institute a data collection point using bar coded pallet numbers on transfer from the factory production floor to inventory. This separates inventory from the waste stream, sometimes the source of an opportunity, as it was in the example noted above. This latter example also offers management some of the data necessary for improvement projects as an added benefit.
- Regular and close management of the debtors and creditors ledgers with the bank accounts. This will detect unauthorised payments made from the bank accounts.
- Restrict manual transactions, particularly where they involve cash.
- Control access to the books to ensure accountability of individuals for accuracy and completeness.
- Build in multi person authorisation into standard processes.
- Have clear transaction audit trails, that are monitored.
If some of these seem a bit ‘over the top’ for an SME, I understand. However, the volume of fraud that goes undetected and unreported is huge, and very damaging for an SME without the financial depth to easily navigate the losses.
Implementing some simple safeguards is just responsible business practice.
Get some help if you need it, and usually the return on a modest investment will be very quick.
Jun 25, 2021 | Governance, Management
Often when seeking advice, we set out to find those who have ‘Domain’ knowledge’. Those who know what we need to know because they have ‘been there done that’ or have studied the domain extensively for one reason or another.
When you break it down, there are three components to building valuable domain knowledge:
First hand experience.
There is no better way to gain a feel for a market than to be out there, in the weeds, dealing with the drivers of performance as well as the day-to-day challenges that arise. Understanding ‘why’ things happen is infinitely better than just being able to observe them happening, it gives you a sense, an instinct that cannot be easily defined. Hands on experience and exposure to a market and its drivers offers the opportunity for the nuanced understanding you may be looking for. Fingerspitzengefuhl‘ is a German word for it.
Helicopter view.
In a helicopter, you are high enough to see the whole domain, but still low enough to be able to see the features that make up the whole. Importantly, you can zoom in and out to investigate features that grab the attention in some way, to examine how they work, and the relationships they have to other features in the terrain.
Through others eyes.
Being able to see and objectively assess your value proposition from the perspective of your ideal customer is vital, a basic discipline of marketing and sales. Why should they buy yours, and not the offering of the opposition? You also need to be able to look at yourself through the eyes of your competitors, seeking the points of relative weakness and strength, the potential paths to a greater share of wallet, or attracting new customers. Others, not necessarily those with whom you are commercially engaged can also have an influence in the way you deploy your limited resources. Regulators, interest groups, research bodies, and others can all have an influence on your enterprise. Being attuned to the potential impacts of those views is an important component of domain knowledge. Just look at what a small group of animal rights activists did to the live cattle trade to Indonesia a few years back. Irrespective of your views on the rights and wrongs, they managed to totally change the face of a large industry almost overnight.
Most business owners find themselves short of the time and expertise to build a nuanced view of their domain. Confirmation bias also tends to rob them of breadth of view. Engaging an advisory group, or individual is the best way to fill in the holes and build long term success.
Jun 17, 2021 | Lean, Management, Operations
Chasing improvements in an enterprise comes down to doing the small things well, every time, and continuously improving, generating a compounding effect.
The best way to achieve this is for everyone involved to be engaged in the process, have a stake in outcomes, and understand how they impact on others.
At every level, this is achieved, not by memo, or strategic planning, but by consistent, focussed verbal communication backed by facts.
Best way to do this is to communicate often.
Not a lot at one time, but small bite-sized chunks regularly.
Daily, weekly, monthly, and so on.
At the ‘coalface’, it should be daily, which leads us to the daily stand-up, huddle, group chat, or as one of my clients call it, ‘toolbox’. Whatever you choose to call it in your workplace, it plays a crucial role in performance management.
This is a daily meeting at the beginning of a day, shift, or whatever the work cycle is, that reviews the day to come, in the light of what happened yesterday, with some acknowledgement of what will be coming tomorrow, and perhaps the next day.
Why it works
- Daily communication keeps everyone on the same page, enables problems to be surfaced and addressed before they really hurt, escalated as necessary, and contributes enormously to a culture of communication and collaboration.
- They replace the one-to-one conversations that need to happen many times, with a one-to-many conversation. This saves time and energy, while ensuring the communication is the same to all parties.
- It enables focus on the priority activities, removing some of the day-to-day firefighting and craziness that always occurs.
- It also enables quick updates to larger objectives and relevant projects to be delivered, which removes the always present rumour mill. This works equally well for the positive things as it does for the negative.
- They lead to significantly engaged employees, as not only are they heard, but they can see the outcomes of their ideas and suggestions.
In these days of increasingly remote workforces, this daily get together takes on a much wider role, in reminding everyone that they are a part of a team, and others are relying on them.
What makes them work
- Same time, same place. Having the huddle, stand-up, whatever you choose to call it at the same time, in the same place, every day creates a cadence that drives activity. Start on time, finish on time.
- Sitting down will elongate the meeting, so stand. It might be in the workplace, often it is just outside the workplace, which adds credibility to the process.
- As short as possible, no more than 15 minutes should be an iron rule.
- Everyone gets a say. Engagement comes with being heard, and the chairman must ensure everyone is explicitly given the chance to have a say.
- This can take many forms and will vary with the level of the huddle. At the coal face, a whiteboard is usually sufficient, with perhaps a photo or copy taken and kept for reference for a short time. At higher levels, the recording will vary.
- Be on time, do not ramble when it is your turn to speak. Take any follow up or extended conversation offline, the huddle is to identify problems, address the molehills, but the mountains are for another place.
- Be respectful of time, others and the process. Be attentive, with no side conversations, or banter.
- Meeting chair. Someone must lead the meeting, have control of the conversations and agenda. That may be the same person every day, or it might rotate, which in my experience is the better way.
Usually very quickly there is a sense of team effort, and even the small wins become evident and can be celebrated. It is an incremental process, which once the ball is rolling, picks up momentum that is very hard to stop, even if you wanted to.
