Sales Forecasting with Prophecy - FAQ
Our FAQ is divided into 4 sections.
- Questions about Prophecy functionality
- Commercial questions
- Technical (IT) questions
- Questions about sales forecasting techniques
If there is a question we haven't answered here, please contact us, either by telephone or our web contact form.
Questions about Prophecy functionality
In addition, you can copy and paste data through the Windows clipboard. Prophecy lets you copy non-contiguous cell selections from the Prophecy report into the target Excel or other application, passing only the cells and labels which apply to the original selection.
Prophecy integrates seamlessly with the open source R Statistical Computing environment and allows expert and non-experts in R to generate statistical forecasts using the leading-edge forecasting routines in the R forecast library. Prophecy offers fully automatic forecasting from R, requiring no R or statistics knowledge, or 'to the metal' access to all the power of R, right from within Prophecy. This feature was added in February 2019.
Prophecy's original built-in statistical forecasting engine supports the following subset of statistical forecasting algorithms:
- linear regression, with and without seasonality
- exponential regression, with and without seasonality
- last year + previous year over year percent
- Batch mode: apply statistical forecasting to a large number of product/customer combinations. (This mode is closest to what the majority of specialist statistical forecasting tools offer.)
- Wizard mode: select the target range of cells in your Prophecy report, choose a method from a simple dialog box and either paste the forecasts in individually or to all target cells with a single click.
- On demand view: generates a new statistical forecast view for the product, customer and measure at the current cell position in your Prophecy report. The view shows the time series as an interactive table, a graph, and also generates a commented description of the statistical control information from the forecast.
- Accuracy analysis: the final, 'human' forecast is compared to the forecast the statistics engine would have generated at t he time the forecast was made.
Finally, bear in mind that increased statistical sophistication does not necessarily equate to increased forecast accuracy. In fact, there is some evidence to suggest that there is often a low correlation between statistical sophistication and increased accuracy. Sales forecasting software should empower the forecaster with the right tools for business forecasting, rather than overpower the forecaster with unnecessary statistical and mathematical complexity.
Business forecasting software should have the following properties:
- It must support multi-user sales forecasting because business forecasting is a multi-user activity.
- It must enable a proper forecast/reforecast process. That is, it must support the ability to review / change an existing forecast in the light of the latest period's actuals. It must support forecast versioning through time so that you can track changes made. It must extend through to money as well as quantities, so that it can be integrated into the overall process of business planning and budgeting. 'Money', meaning sales revenue, gross margin etc..
- It must allow you to generate budgets by product, customer and time and to compare the progress of the current sales plan (actuals to date plus forecasts thereafter) versus the budget, as well as previous year(s) at all the varying levels of product, customer and time detail.
- Accuracy analysis. It must be simple for forecasters to analyse their forecast accuracy. If they cannot identify their mistakes they cannot learn from them and therefore improve next time! The start point for a forecasting cycle should be an analysis of the previous period's forecast accuracy. How well did the forecaster use the information they had to make the forecast and what can be learned from the forecast errors?
- The data must be viewable and capable of being forecasted across varying levels of product, customer and time detail. Bottom-up forecasting is generally the starting point, but top-down helps to give an overall perspective and 'sanity check' on the overall numbers, and allows review by exception.
- Business forecasting software must support a review and sign-off process. Therefore the forecaster must be able to submit his/her forecasts for review. A reviewer must then review and either request/make changes or approve. Multiple review levels may be necessary - e.g. sales manager to regional manager to sales director etc.. This feature ensures forecasts have ownership and visibility.
- Software for business sales forecasting must be easy to use, because business forecasters are not necessarily IT or statistics gurus.
- Mainly concerned with generating a forecast mathematically, given a time series history.
- Not multi-user.
- Forecasts the measure you provide it with, but does not (automatically) extend to revenue, margin etc..
- Limited ability to analyse or report on the data it generates.
- Limited or no ability to compare forecasts with budgets.
- Limited or no accuracy analysis on previous forecasts.
- Does not support a forecast review workflow process.
- Limited or no ability to report on or forecast at varying levels of product or customer detail.
- Offers numerous statistical options, which operate in a 'black box' manner.
- Not necessarily easy for non-statisticians to use or understand.
Therefore Prophecy will never change forecasts except at the forecaster's instigation. That way the forecaster is always responsible and retains full ownership of all their numbers.
You can display any dimension (i.e. product, customers, periods, years, measures or versions) on any report axis, just like (but better than!) an Excel pivot table.
You can drill up and down through the product and customer hierarchies.
You can conditionally format cells based on their numeric contents, to highlight particular conditions.
You can create your own, user-defined subtotals of products, customers or time period.
The data from any Prophecy report can be graphed and the graph always updates in real time as you change the forecasts or navigate the hierarchies.
All this functionality is implemented in a very visual way that non-technical users will find easy to use productively.
We have found that a lot of Prophecy users get a substantial part of their sales reporting from Prophecy. Sales managers in particular like the fact that all their data is in one place, is instantly available and that they only have one system to master.
Click here for a self-running demo of some of the features of Prophecy reports (opens in a new window).
Alternatively, we can run a Prophecy demo over the internet using our TeamViewer remote meeting software.
Then, if Prophecy looks like the potential solution to your sales forecasting software requirement we would often agree to develop a trial system for you, containing a subset of your products and customers. Typically there may or may not be a cost associated with this activity, dependent primarily on how 'elegantly' you are able to supply the data required to build the trial database. Normally any cost incurred would be subtracted from the eventual cost of your Prophecy solution in the happy event that you decide to proceed.
We are happy to put you in touch with an existing Prophecy customer in the event that you decide Prophecy is the solution of choice, subject to the reassurance of talking to an existing user. We are confident that this will provide full reassurance of the quality of our software and the strong commitment we have to providing the highest quality of support to our customers.
Please see here for a fuller description of the demo options or here for our full contact details or to contact us via our web form.
Prophecy is licenced on a per-named-user basis. Therefore any individual, whether as a forecaster or a viewer, who uses the Prophecy client application is deemed a named user. Volume discounts apply from 5 users upwards.
Please contact us for a detailed pricing quotation.
Having said that, Data Perceptions would like to be 100% customer-focussed in all areas of our business. If you absolutely have to have a concurrent licensing model then we can discuss!
Overall, Data Perceptions takes support very seriously indeed. If you pay us the compliment of using our software to help run your business we feel you are entitled to the very best quality of support that we can provide. We know that our existing customers speak very highly indeed of the quality of support they receive.
- SQL Server hosted databases store their data in a Microsoft SQL Server database. The advantages of SQL Server databases are scalability and the ability to access Prophecy data from any SQL reporting tool. Performance is outstanding. Sparse data is fully handled, minimising unnecessary storage and network traffic. Most new implementations of Prophecy are SQL Server based these days.
- 'Native mode' databases are proprietary, file-based databases. They reside in a shared folder on a server. Prophecy handles all contention issues associated with shared files via protected rights, queues etc.. The advantage of 'native mode' is simplicity. All the database files are held in a single, shared folder, which can be backed up or zipped easily and there are no third party software dependencies. 'Native mode' is more suitable for standalone use - e.g. on laptops.
A 'native-mode' database can easily be migrated to SQL Server at any time by running a Prophecy database maintenance script in our Prophecy Database Manager application. Both formats share the same real time consolidation and allow multiple users to concurrently 'hit' the database.
The IT elements of a Prophecy implementation are described in more detail here.
You only need to import data at the base level of the product and customer hierarchies
A Prophecy Script automates running one or several import filters sequentially, solving any dependent measures and aggregating from the base levels in the product and customer hierarchy.
The typical Prophecy refresh is from one, pre-made script, which is simply run from the Prophecy Database Manager.
Also, you can choose the product and customer levels in the export. So, for instance, to export the forecasts into a production planning system you would export the base level from products (i.e. skus) and the top level from customers (i.e. Total Customers). You can either use period/year fields in the export, or your own, custom-defined time labels which the export script extracts from a lookup table in the Prophecy database.
Think of Prophecy as the forecasting application within Microsoft Office! Therefore the best way to deploy Prophecy to remote users is via a Remote Desktop connection to your server. This enables remote and office based Prophecy users to share the same database concurrently, with no additional consolidation or administration.
There are currently over 130,000 lines of hand-crafted C++ code in Prophecy. That's code written by Data Perceptions, and excludes comment lines (we have them too!), blank lines and code from the Microsoft class libraries that we use. It's tough, but we remain committed to C++ because it provides the best performance for our users.
However, Prophecy is not just a simple, single user spreadsheet. It is a multi-user business system. Therefore it should be implemented with the processes and disciplines you would apply to any other business system implementation. The hierarchies, data and database management processes must be robust and properly validated before users are allowed to start using the system.
The Prophecy implementation process is summarised here. The nature of the IT resource required to implement and support Prophecy is summarised here.
Please contact us for more information on the implementation process. We are happy to send you additional documentation on implementation and the data import/export processes.
- Reading in the latest product file (to add new products to the database).
- Reading in the latest customer file (to add new customers to the database).
- Reading in the latest sales actuals and (if included) cost etc. files to the database.
- Aggregating the new data over products and customers.
Typically, database management is performed by a person from "the business" who has good administrative skills but is not necessarily an IT 'guru'. Refreshing the database each period should take a lot less than an hour each time.
Server: Prophecy hosts its data in a Microsoft SQL Server database (2008R2 or later.) The Prophecy client software connects to SQL Server using the standard installed ODBC drivers. A Prophecy database on SQL Server is normally no more than ~4Gb, though obviously the number of products, customers, time periods, measures and forecast versions, plus the data sparsity, affects database size.
If forecasting the monthly peaks and troughs is the required output of your forecasting process then it is likely that statistics is going to be of limited use, other than possibly as a baseline forecast or a sanity check.
We believe in empowering your business managers to become better forecasters. These are the people who are most qualified to know and understand the effects of promotions, competitor activity etc. and when these factors are likely to apply. Prophecy is designed from the ground upwards to help business managers forecast productively and effectively by providing them with the flexible tools and graphical options which support 'judgemental' forecasting. It has statistical forecasting too, of course. Used carefully Prophecy's statistical forecasting algorithms can provide a start point or act as a sanity check on a forecast.
We believe that the forecaster should 'own' the sales forecast. This ownership comes from having carefully and objectively prepared it. Getting the overall (smoothed) medium or long term sales trend right is relatively easy - a year on year percent is likely to be virtually as accurate as a statistical forecast. What is hard, but extremely valuable to the business (and often the primary reason for forecasting), is predicting the individual peaks and troughs in product sales which is frequently typically very 'spiky'. That's where 'judgemental forecasting' and having the right people and tools to support 'judgemental forecasting' is key. Data Perceptions has the software and your business will have the right people!
One additional important point to underline. In the business process of sales forecasting the starting point for this month's forecast is normally last month's forecast for the out periods, updated with last month's actual sale. The process is then one of modifying the existing future forecasts in the light of the latest sale plus new information about promotional plans, competitor activity etc.. You will very rarely need to generate a completely new forecast from scratch, because this will overwrite all the promotional and other peaks or troughs that (should) have been crafted into the forecast. Therefore, again, sales forecasting software for business needs to offer features that help forecasters track and understand the patterns of sales and identify which forecasts they need to review rather than continually overwriting existing forecasts with new statistical ones.
At the very least, if you feel you want to use statistical forecasting, make sure the statisticians prove to you that their algorithms work on your data! We realise that there will be markets where statistical forecasts can be used - we just haven't found any yet! (Note: Prophecy does have statistical forecasting options! See above for more information.)
Please see our whitepapers section for more information about the thinking behind Prophecy's approach to business forecasting and the role of statistical algorithms.
Telephone 01494 785574 (UK) to talk to a Prophecy expert now!