1. Why multidimensional?
Traditional costing models are mostly
based on a single dimension output like
‘products’ or ‘customers’. Because of this
single dimensionality, some cost
allocations might not reflect a true cause
and effect relation to the output and can
lead to wrong decisions.
A traditional example is the cost of the
sales force. If the cost model contains only
one dimension, say product, you might
want to allocate sales force costs as an
overhead to all products based on say
sales value; or worse on the costs of each
product. This can lead to high cost
products being ‘over-priced’ where
perhaps the sales force actually spends
more time on less costly products. This
‘cross subsidising’ effect can lead to
wrong decisions.
Of course cost of a product is hardly ever
the real driver for sales costs so in a
multidimensional approach you analyse
what a sales force does. It spends time
and money on existing clients, on specific
sales channels or countries and perhaps
also on specific products such as a
marketing campaign. In order to prevent
cross subsidising you want to allocate
costs with the right ‘driver’ to the right
‘dimension’. In an example from a
Telecoms business the time spent on the
UK market was first allocated to the
object ‘UK market’. Next the costs were
allocated to those transactions that
occurred in the UK, based on the best
available driver, say sales volume.
The same sales department also incurred
costs on a world wide marketing
campaign for ‘Smart phones’. In the
multidimensional cost model these costs
were first allocated to the object ‘Smart
phones’ of dimension ‘Product group’ and
then to only those transactions that
involved smart phones.
So instead of allocating sales costs with one
driver two steps were used as well as better
drivers to the transactions. This resulted in
a more accurate cost allocation with less
cross subsidising.
Why transaction based?
Sales transactions or order lines are facts.
The more facts in your cost model the
better accepted are the results of the
calculations which get used to improve
decision making. The transaction files are
the life blood of the organisation containing
valuable dimension information that is easy
to obtain to update the model at any time.
Data mining: better decisions.
Once the allocations to the transactions are
done, the user can analyse the ‘result cube’
in many ways using all dimensions, driver
values and cost-per-dimension result fields.
Interactive slicing and dicing of the result
cube leads to a better understanding of the
true causes of profit and loss from the
multidimensional perspective.
The output is automatically generated as a
pivot table within Costperform or can be
analysed via OLAP tools such as QlikView.
#1: CostPerform’s multidimensional costing
function leads to improved profitability
CostPerform has a great new function to handle large data sets of sales transactions or order lines. In
many organizations, these types of transaction files are not only easily available but they also contain
valuable cost driver information for analyzing multiple ‘dimensions’ such as Product, Customer, Sales
channel, Country and so forth. With this function you can pull or push costs via these dimensions to the
most granular level available - your individual transactions. The flexible pivot function within
CostPerform runs on the ‘result cubes’ so you can slice and dice to explore the true profitability from the
multidimensional perspective. Better information for better performance improvement decisions.
How does it work?
Via a scheduled task this function is
designed to work automatically to:
 Import transactions, update attributes
definitions and values.
 Calculate and update the cost model for
each dimension instance found in the
transaction file.
CostPerform’s USP
CostPerform is not competing with Big
Data analysis tools that are designed to
find hidden relationships in existing
mountains of data. CostPerform delivers a
new piece of ‘calculated’ information that
is missing from the data. CostPerform
provides profitability information at the
detailed granularity level found in the
transaction data. This new insight into
true profitability from multidimensional
perspectives leads to better decisions
about where to focus on improving
profitability.
Traditional one dimensional costing
methods are no longer fit-for-purpose. For
today’s more rigorous needs CostPerform
does the job……and more.
The new multidimensional functionality in
Costperform can take data feeds directly
from your transaction files and is also the
ideal partner to work hand in hand with
all the leading edge Big Data tools so that
‘profitability’ could be one of the most
valuable pieces of information added to
the data set.
Conclusion.
If you want to know the true cause of
profit & loss then simply feed your
transactions into CostPerform, run the
model and start slicing and dicing on all
dimensions until you find the right areas
to improve your profitability.
Info: +44 (0)3 303 305 307 www.costperform.co.uk enquiries@costperform.co.uk
Example pivot table: Margin by Country, Channel, Customer Group

2. How it works
The Meta model contains a special layer
called Transaction layer. This special layer
cannot contain objects but knows where
it’s ‘objects’ (transactions) are stored.
Transactions are never loaded into
memory. If a model contains a
transaction layer the calculation starts to
process the related transaction table.
As a result, all members of each
dimension are inputted or updated into
the cost model as an object in a layer
designated as the ‘Dimension layer’.
For all selected values in the transaction
table, the calculation sums each value per
member of each dimension. These values
are stored per object in (generated)
attributes and can be used as a cost or
distribution driver.
Once the driver values are updated, the
cost model is calculated. The end user has
to make sure the allocations or
(preferably) distribution to the members
of the dimensions are defined.
For all dimension members that receive
cost, the calculator distributes the costs
for each member from each dimension to
the transactions using any available
driver. As a result, each transaction
contains costs-per-dimension results as
well as total costs and margin.
The transaction import task takes care of:
 Creating the transaction table
 Importing the transactions
 Updating the meta model attributes
 Updating the cost model for each
dimension member found in the
transaction table.
Handling Big Data files
The capacity to process transactions is only limited by time and database capacity.
We’ve tested the multidimensional function with a transaction file of 100,000,000
lines containing the data describing:
 4 dimensions (product, client, channel and country)
 3 cost driver values (number of products, income and weight)
In just over 2 hours on a SQL server database the update process performed the
following steps:
 100,000,000 lines were imported into a CostPerform database table.
 Per each of the 4 dimensions, 3 values were summed for each member of each
dimension. This resulted in 3 x 4 =12 queries
 Each unique occurrence within each dimension (all product ids, client id etc) was
inserted as object or updated in the target CostPerform cost model.
 The cost model was calculated
 The results per transaction were stored in the database table.
 Each transaction contains the original transaction and the result-set
 The result set consists of the ‘Cost per dimension’, a total cost value and a
margin value
 In total 600,000,000 values were updated.
 In total, the result-set contains 1,300,000,000 fields
A big advantage of this function is that internal memory usage is stable, regardless
of the number of transactions to process.
#2: Putting CostPerform’s multidimensional
costing function through its paces
CostPerform has a great new function to handle large data sets of sales transactions or order lines. These
days the sheer volume of transactions in the business would be part of the data set that is now called ‘Big
Data’. In many organisations, this type of ‘transaction file’ is not only easily available but also contains
valuable driver information (volumes, sales value…) and multiple ‘dimensions’ such as Product, Customer,
Sales channel, Country of sale and so forth. With the ‘transaction layer’ option added to a model, CostPerform
can access the Big Data in the company and auto-update cost models straight from the transactions and
calculate the latest costs per dimension, total costs and margin. The transactions are stored in the
CostPerform database, ready to be analysed in the pivot function and be picked up by any OLAP tool for
analysis. This multidimensional costing approach is very well suited for organisations that want to:
 understand true costs and profitability on the highest possible level of granularity;
 auto-update their cost models by transactions and have year to date cost
information in order to make better operational improvement decisions.
Info: +44 (0)3 303 305 307 www.costperform.co.uk enquiries@costperform.co.uk
Analysing 1,300M results
A result-set of 1,300M fields requires
specialised OLAP tools. For example using
QlikView on top of the result-set gave the
following satisfactory outcomes:
 Loading and compressing took 45
minutes
 Opening without re-compressing took
only 30 seconds
 Slicing and dicing, selecting and clicking
with almost no delay
Example QlikView multidimensional analysis