Promo
MySales Promo helps you:
Create promo;
View promo;
Edit promo;
Approve promo;
Plan promo;
Compare promos;
Model promo outcome;
Keep all promos in one place;
Automaticly replenish stores and warehouse based on promotional forecasts
Promo module is based on the latest achievements of machine learning technologies and lets you forecast, model and store all your promotional activities in one place. The system is taking in account size of discount, type of promo mechanics, TV campaigns, Facebook ads, outdoor advertising, email letters, additional equipment and many other influencing factors.
It is possible to have several scenarios and mechanics for one promo. The system automatically calculates the performance indicators under various promo conditions (discounts, mechanics, timing, etc.) and offers you a comparative analysis of the effectiveness of each option.
To go to the Promo module functionality, use the "Promo" tab on the main screen menu:
There are two ways to display promos:
List - displays all the promos in the form of a list with the name and basic information about the promo. Details on working with the promo list are described in the section Operating manual;
Calendar - displays all the promos in the form of a calendar with the ability to display information in the weekly and monthly formats. The promo calendar is described in detail in the Promo calendar section.
How we predict promo
For successful and accurate promo forecasts, the key factor is the complete promo history in your chain. For a high-quality forecasts, system needs to have information about previous promos. It is very important that the system has full information about the past promos, all communications and mechanics, discounts and compensations.
If not all past promotions, as well as incomplete information about them, are entered into the system, it will not be able to determine the influence of various factors on sales and find dependencies between past promotions and future ones. Therefore, the quality of the forecast of the promotion depends entirely on the quality of your historical data.
MySales has two approaches to forecasting promos:
Search for dependencies between previous promos and new ones
Using a neural network for forecast
By default, the system looks for dependencies between historical promos and new ones, and only if it doesn’t find them, it uses a neural network.
The search for dependencies lies in the fact that system analyzes how different factors influenced the promo before and looks for already held promos that look like a new promo. This allows the system to model a change in sales for a promotional period. Usually, if the algorithm finds three or more comparable promos, then the quality of the forecast will be high. For a satisfactory promo forecast you need to find at least one promo. If comparable promos cannot be found, then the forecast is built using the neural network. This forecast is usually worse than one that uses comparable promos, since the system does not have enough data to understand the influence of certain factors.
The selection of comparable promos and the neural network forecast both results in two promo growth factors:
The uplift rate of the promo to the forecast, which does not take into account the influence of the price / discount
The uplift rate of the promo to the forecast, which already takes into account the influence of the price / discount
The uplift factors are separated due to the fact that the forecast, which already takes into account the influence of price / discount on promotion periods will initially be higher than the regular forecast, respectively, the additional increasing coefficient that is applied to it should be lower than coefficient applied to the forecast that excludes price elasticity. Further, application of this coefficient to the basic forecast when calculating the promotional uplift and its further adaptation is described in the section Forecasts - Analysis - How we build a forecast. Thus, having a coefficient predicted at the SKU level, using adaptation of the promotion uplift, the system takes into account the individual sensitivity to the promotion at the level of each store.
Selecting comparable promos
When selecting comparable promos, the system takes into account a number of rules that can be divided into strict and soft. Both strict and soft rules in the system can be customized. If strict rules are not followed then the system does not select comparable promos, which can lead to the fact no comparable promos will be selected with an insufficient promo history. Soft rules work as a prioritization in the selection of comparable promos. Promos with similar factors are more significant for predicting promo uplift.
Next rules can be considered as strict (due to system settings):
Promo separation by types of communications. For example, for promotions with newspaper communication, only past promotions with newspaper communication will be selected as comparable.
Promo separation by discount level. For example, for a promo with a 30% discount, only past promos with a discount from 20 to 40% will be selected as comparable.
Promo separation by mechanics. For example, for promotions with casual discount, only promos with the same mechanics or promos with similar mechanics will be selected (discount only for cardholders).
Promo separation by duration. For example, for promotions lasting 3 weeks, only promos lasting from 2 to 4 weeks will be selected, or for one-day promotions only one-day promos in the past will be selected as comparable.
Promo separation by assortment range with the allocation of chain-wide promos.
Promo separation by number of stores where it is held. For example, when you want to avoid matching promotions held throughout the chain with promotions held in several stores.
Comparison limitation by time range. Taking into comparison only promos, which were held less than two years ago.
Promo separation by festive periods. For example, when for promotions held at Christmas, only past promotions that were held in the corresponding period will be selected.
The list above is not final. Each of the strict rules above is configured in accordance with the specifics of promo conduction. It is not recommended to severely limit the system with strict rules, because of this, the percentage of positions for which the system will be able to select comparable promos can be reduced.
The soft rules for matching promos include a large number of factors that can be seen in the menu MORE - PROMO - MATCH DATA. Soft matching rules comparison is based on similarity coefficients, which the system automatically calculates based on history, sorting out combinations of promo positions in the past and evaluating how the repeatability of a factor affects the repetition of a promotion uplift. When comparing promos according to soft rules, historical promos that do not comply with strict rules are cut off.
For all historical promos that comply with strict rules, the system calculates the similarity coefficient, which is used to sort historical promos to match those promos that have the highest similarity coefficient. Thus, for each position, from 1 to 6 historical comparable promotional positions are selected that have the greatest similarity of all available.
After selecting comparable promos, the system adjusts the final uplift rates to their statistical differences in terms of discounts and mechanics.
The criteria by which the system evaluates the similarity of historical promos for comparison according to soft rules:
The number of calendar dates of the month for which the predicted promotion matches the comparable. For example, if the historical promo was held on December 2-4, and the predicted on December 3-5, respectively, the coefficient of similarity by dates would be 50% (December 3-4)
The duration of the predicted and historical promo in days. For example, the duration of the predicted promo is 7 days, and the historical one is 5 days. The similarity coefficient will be 5/7 * 100% = 71.4% (a smaller number is always divided by a larger one)
The time period of the historical promo in relation to the predicted
Sales season match
Promo name match
By discount level
By mechanics, considering similar mechanics
According to the total set of communications
According to each communication separately
By similar stores
By total stores count
By similar SKUs
By total SKUs count
By the number of groups in the promo
By average sales of SKUs in a promo in volume and value
By average price of promo item
By coincidence of holidays falling into the promotional period
By coincidence of the current selling price
By terms of competitiveness of promo items within one group
By coincidence of product group
By coincidence of the item name
By coincidence of a specific SKU
By similarity of the historical promo item uplift
By similarity of historical discounts of promotional items
By average sales positions in pieces and in money
Predicting promos using neural network
If system can't find comparable promos, the uplift rates of the promos are predicted using a neural network. The neural network is trained using historical promos. It's trained both with all data and separated data for each group, determining the dependence of the promo uplift coefficients on the influencing factors and their synergy.
In the menu MORE - PROMO - BRAIN CORRELATION, you can see how the neural network has been trained on the whole system level and for each group separately. And how the input data affect 2 output neuron values, which are responsible for the uplift rates with and without price / discount effects.
Business processes for working with promo / Promo verification:
Promotions in MySales are managed by Promo administrator. This is a person who creates promo, calculates a promo outcome in system, and passes the results to category managers.
A promo administrator can work in a category/commercial department, a logistics department or a marketing department. In all cases, the processes of working with the promo will be approximately the same; main differences, will be in permissions for promo approvement.
The process of working with a promo is as follows:
Category managers come up with promotional initiatives which they send to the promotional administrator;
The promo administrator calculates these initiatives in the system and generates a sales forecast in MySales at the SKU-promo level for all stores and a promo margin plan;
The promo administrator verifies the promo forecast (described in more detail below) and sends it with the margin plan to category managers;
Category managers make their reasoned comments on the forecast (if any come up);
At the promo committee, the option of conducting the promo is selected and approved;
Promotional support is agreed with suppliers;
The chosen promo option is approved in the system - it automatically applies to the forecast, warehouse order and store orders;
If the category manager has reasoned comments on forecasts (with the consent of the promo administrator), an additional master forecast at the SKU-promo level is entered into the system, which is distributed to the system’s forecast at the store level and is overwrites promo forecast which was automatically calculated.
Promotional administrators can also offer their own promotional initiatives.
Promo administrator motivation:
The volume of promotions (in money and in margin);
The accuracy of the promo forecast, which is measured as % of the promo positions that fall within the allowable error range at the SKU-promo level (usually + -30%).
Promotional forecasts of the system must be verified. In the verification process, promo administrator must pay attention to the following:
Are comparable promotions matched or not?
If a comparable promo is selected, you need to pay attention to SKU and check if comparable SKU is similar to the predicted?
What are the sales and price / discount of selected comparable promos?
How long has a comparable promo been held?
Also, special attention should be paid to positions, for which:
Promo is the first one;
There is a pronounced seasonality - for example, ice cream and beer;
The system could not find an analog;
Promo forecasts, for which there is no sure answer for questions above must be checked in the Analysis form.
Operating manual
Promo table elements
Created promo are displayed in a form of a table with data, that displays basic information about the promo and allows user to select separate promo for further analysis.
Table with promos has the following fields:
ID – promo identifier;
Status – promo status (if promo is draft or approved);
Creator – the name of a user, who created promo;
Promo's name – the name of promo;
Start date – promo start date;
End date – promo end date;
Disc - discount in percent;
Type - the type of promo;
Uplift – manually set additional sales increase in a promo (in addition to the calculated system);
Sales LY – last year sales in promo period;
Base – sales forecast, excluding promotional communications;
Promo - sales forecast for promo;
Sales CY – promo period sales (they appear after the beginning of the promo and are updated every day);
Date saved - date of last promo parameters update.
For detailed display of an individual promo data, click on its name in the table.
Table menu
Table with promos has a functional that allows you to perform basic operations with the promo. This functional allows you to delete, copy, recalculate promos or upload one or more promos to the excel. To select promos with which you want to work, tick the box in row with promo.
Then press button, it opens promo menu.
"Recalculate" recalculates promo, taking into account the newest data available.
"Export to Excel" function uploads promo data to excel file.
"Copy" makes a copy of the promo.
"Delete" deletes promo from the system.
Filters
For mor convenient promo displaying and searching, table has following filters:
Search string;
Start date - only those promos will be shown that has a start date bigger than the selected one;
End date - only those promos will be shown that has an end date is less than the selected one;
In date - only those promos will be shown that are actual for the selected date.
Adding new Promo
To add a new promo, click on the button "Create new promo forecast".
To continue creating a promo, click the "Continue" button.
The following window will open for adding a promo, in which the user should choose which SKUs will be used for the promo.
The list displays the manufacturers of goods, to go to the SKU level, click on , groups will open, pressing
near groups opens SKU level. To add a manufacturer, group or SKU to the promo, click
button. You can also add all the products by clicking on the "Add all" button. To display already added items, click on "Added skus". This will open a window with already added items. To return to the selection window, click on "All skus".
To remove the position from the promo, click button. You can also remove all selected products by clicking on the "Remove All".
After finishing the selection of SKUs for the promo, click "Done, I've picked all the SKUs", to go to the selection of stores.
After SKUs for the promo, you have to select the stores in which the promo will be held. Selection of stores is similar to the choice of goods. Stores selection window displays a list of regions, you have to click to go to the store level.
To add a store to the promo, click in line with the store, to add all stores, click on the "Add all" button. To display already added stores, click on the "Added stores". This will open the window with the stores that have already been added. To return to the selection window, click on "All stores".
To remove store from the promo, click button. You can also delete all selected stores by clicking the "Remove all" button.
After finishing the selection of SKUs for the promo, click "Done, I've picked all the stores", to go to the selection of promo parameters.
After selecting SKU and stores, you must select the promo parameters. The window that allows you to select promo parameters automatically opens after you finish stores selection process.
You can set the next parameters of promo:
Name - the name of promo;
Discount - promo discounts in percents;
Type - the type of promo;
Mechanic - promo mechanic;
Compensation - supplier compensation in percents, if there is one;
Compensation type - the type of compensation;
Communication - a selection of how the promo will be promoted (for example, TV advertising, facebook, etc.);
Uplift - sales growth in the promo in percent. Set if the user is sure that the promo will receive a guaranteed increase;
Remarks - additional information about the promo (if the user finds it necessary to enter it).
After selecting the parameters, you can click the "Calculate" button to calculate and save the promo, or "Save" to save the entered promo data with the possibility to calculate in the future. After clicking "Calculate" the system starts the process of promo calculation.
When the calculation ends, a window with promo data opens. It contains all the information about the created promo, and the functionality to work with the promo.
Displaying promo
The promo view page displays all the details about the promo and consists of two blocks.
On this page, a user can view both new, not yet held promo, and those that have already ended to assess the economic effect and compare previously predicted sales with real promo sales.
The first block has information about promo parameters and is displayed on the left side:
The second block is located on the right and has information about promo-sales, as well as functionality for working with promo.
This block contains the following information about the promo:
Previous year sales for promo period;
Sales forecast for promo period, excluding promotional communications;
Promo sales forecast.
Information can be displayed in volume, value and margin.
Also this block has the following functional for working with promo:
Approve - by clicking this button, the user approves promo in the system;
Comparable - displays already held promos in the system, which can be compared with the displayed promo, the system displays the following information about comparable promos:
Promo ID - identificators of comparable promos;
Date - the period when promo was held;
Promo name - the name of promo;
Benefit - promo discount in percent;
SKU count - the number of SKUs that participated in the promo;
Avg. match - a coefficient that shows how in percent comparable promo is matched with the displayed promo;
Avg. uplift before - average increase before the promo in percent;
Avg. uplift after - average increase after the promo in percent.
Upload - upload files with the promo data. It is described in detail in the section "Downloading data files";
Excel - downloading promo data into excel file;
Analyze - go to the Analysis page for a detailed analysis of each of the promotional items;
Edit - go to edit editing promo.
Adding new promo variant
The promo module has a functional that allows you to create several variations of the promo for the same SKUs and stores, this allows you to compare the effect of promo which uses different strategies.
To add a new promo variant, click on "Create a new promo variant"
After clicking, the window opens in which you can set the parameters for the new promo variant, similar to the parameters that are set when creating a new promo.
After entering the parameters, you can press "Save" button - for saving promo variant, with the possibility to calculate it later, or "Calculate" button - for calculating the promo and getting the results of the forecast.
After clicking one of the two buttons a new promo variant will automatically appear in the promo display window.
The new promo variant is displayed first, previously created variants are displayed below, sorted by creation date. The left block of the window displays information about the last added promo variant. For previously created variants, the following functionality is provided:
- deleting promo variant;
Up - rises variant to the first place in the list of promo-variants;
Edit - change the parameters of the selected promo variant.
The user can add an unlimited number of promo versions for comparison.
Promo details
The Promo module functionality allows user to view not only the general parameters of each promo, but also a detailed level of each SKU participating in the promo. To proceed to a detailed analysis of each SKU, click the "Details" button:
After clicking, detailed promo information for each SKU opens:
On SKU level, data is displayed as a table that has the following fields:
SKU Name - identifier and name of each SKU in promo;
Uplift - field in which user can enter a promotional sales uplift factor for a particular SKU in percentages (you must click the "Save" button to accept the changes);
Disc - field in which user can enter discount for a particular SKU in percent (to accept the changes, you must click the "Save" button);
Communication - user can select individual communications for each SKU (to open the selection window, press the "Select" button):
Comparable SKUs - to open a list of comparable SKUs, press the "C" button. Table of comparable SKUs has the following fields:
Promo ID - identifier of promo, which has comparable SKU;
Dates - the period when promo was held;
Promo name - the name of promo with comparable SKU;
Benefit - promo discount in percent;
SKU ID - identifier of comparable SKU;
SKU name - the name of comparable SKU;
Match - a coefficient that shows how in percent comparable SKU is matched with displayed SKU;
Uplift before - increase in sales before the promo in percent;
Uplift after - increase in sales after the promo in percent.
Previous year sales for promo period;
Sales forecast for promo period, excluding promotional communications;
Promo sales forecast.
Stores - the number of promo stores in which SKU is in an assortment.
The detailed information can be displayed in volume, value or margin.
Promo editing
On the promo page, you can edit the promo. It is possible to edit the following promo data:
Promo dates. To start editing press button near "Dates" field. After clicking, the window for editing dates will open:
SKUs that participate in promo. To start editing press button near "SKUs" field. After clicking, a window will open for editing the list of skus (editing the skus is similar to their selection when creating a new promo):
Stores that participate in promo. To start editing press возле параметра "Магазины". After clicking, a window will open for editing the list of stores (editing the stores is similar to their selection when creating a new promo):
Promo parameters. The editing window is opened by pressing either the "Edit" button in the right block or the button near "Parameters" field on the left side. After clicking, the window for editing promo parameters will appears (parameters editing is similar to their selection when creating a new promo):
Downloading promo data to Excel
When you select the function of downloading promo to Excel, the system automatically creates an Excel file with information about the promo. The structure of this file is can be customized for user needs. An example of excel file: "Excel File Example".
Loading files with promo data
The user can load promo data from Excel files to simplify the process of adding large amounts of data to the promo. The system supports loading of the following data:
List of SKUs in promo with discount and communication data (if this data is not specified, then global promo parameters will be used for each SKU). The file must have the csv extension and the following fields (all fields must be separated by a semicolon):
SKU_ID/ITEM_CODE - the identifier of the product for which you want to load data. In the case of loading by MySales code, you must use SKU_ID, but if loading by the internal code, you must use ITEM_CODE;
BENEFIT_PERC - the benefit for SKU, must be expressed as a percentage with a sign of % at the end (for example 40%, 33%, 28%);
COMMUNICATIONS - communications for SKU. Enter either a communication number or its full name (the way it is saved in the system). Each new communication for SKU has to be added in through the semicolon after the previous one in the row with SKU.
An example of csv file: "Example of csv file with data".
Example csv file is filled with discounts and communications for the three SKUs:
For SKU 360274 both benefit and communication are filled (using the name of the communication).
For SKU 360337 only benefit is filled. Communications need not be filled, but you need to put the required number of semicolons.
For SKU 1137890 both benefit and communication are filled (using the communication number).
It is important that the number of semicolons in each line with SKU should be equal.
Updated discounts on SKU level for already added SKUs in promos. It is used if the user needs to change the discounts for individual SKUs in the promo. В файле должно быть две колонки:
SKU_ID/ITEM_CODE - the identifier of the product for which you want to load data. In the case of loading by MySales code, you must use SKU_ID, but if loading by the internal code, you must use ITEM_CODE;
BENEFIT_PERC - the benefit for SKU, must be expressed as a percentage with a sign of % at the end (for example 40%, 33%, 28%);
The master forecast data. It is used if it is necessary to load data from the file about the master forecast for individual SKUs in the promo (master forecast is a manual forecast obtained by experts). The downloaded file must have the csv extension and the following fields (all fields must be separated by a semicolon):
SKU_ID/ITEM_CODE - the identifier of the product for which you want to load data. In the case of loading by MySales code, you must use SKU_ID, but if loading by the internal code, you must use ITEM_CODE;
FCST - master forecast value for SKU;
WEEK - week for which the master forecast is loaded (optional field);
An example of csv file: "Example of file for master forecast data load".
Promo calendar
A promo calendar is another convenient way to display a promo. It allows the user to track and control all the promos at specific time intervals. In contrast to the list display, the user can see which promos are being held, view past promos, and see their effectiveness in a convenient familiar calendar.
There are three options for displaying a calendar:
Week - displays in the context of the week, displays all the promos held on the selected week:
Month - display in the context of the month, displays all the promos held in the selected month:
3 months - display in the context of three months, displays all the promos held for the selected three months:
The user can not only display promos on the calendar, but also see general information about the promo simply by clicking on it. Similar to the events in regular calendars that everyone is used to. And if you need to see the whole analysis for the selected promo, you can open the promo analysis window by clicking on "Full promo information".
Use cases
Promotional activities in retail is a very important driver of sales and customer acquisition, as well as a tool for competition between retailers. According to a number of studies, the share of promotional hunters (customers - discount hunters) among customers exceeds 70% and continues to grow.
Modern retail offers customers a large selection of products in each category and in most product groups, and price is one of the key criteria for choosing a product on a shelf by a buyer. Also the width and regularity of promotions is an important criterion for a customer to choose a store.
At the same time, each retail chain, attracting a huge number of customers, has a powerful marketing resource that allows it to offer the supplier favorable conditions for promoting its product. Therefore, often, retailers conduct promos together with suppliers. Such cooperation is beneficial to everyone:
The customer receives favorable conditions for the acquisition of goods
The retailer attracts more customers and grows its sales
The supplier provides an opportunity for more customers to learn about their product and also increases sales
However, improper promotion planning carries a number of risks and dangers that can turn it from a profitable tool into a deeply unprofitable one. If the promo did not give the growth that was expected, everyone lost:
The customer did not know about the favorable conditions for the purchase of goods
The retailer did not attract customers and did not increase sales by providing a discount and generated an effect of sales cannibalization of more profitable goods. At the same time, having ordered more goods than necessary, after the promotion, the retailer was left with over stocks, which will not be sold soon, but if it is a perishable goods, then it will lead to write-off and direct losses
The supplier also received illiquid inventory and lost margin without effect on sales
It often also happens that both the retailer and the supplier offer discounts well below the psychological limit of the customers sensitivity, which may entail an unjustified loss of margin.
Accordingly, the main cases for using the promo module are:
Accumulation of structured information about historical promos and keeping a calendar of future promotions
Modeling various scenarios for conducting a promo and choosing the most optimal one. What if you give a 30% discount? What will be the sales? What will be the margin? Is it profitable? What if 40%? But what if you additionally hang a banner, an external outdoor, or carry out a promotion with the support of TV advertising?
Automatic promo sales forecasting with support of forecast verification tools
The use of promotional forecasts for the automatic ordering for both warehouses and stores