The methodology commonly used for projecting data is called ratio estimation. The ratio projection factor is usually based on the all commodity value or ACV. ACV is the sales value of the product categories (i.e. those that are tracked by the research firm) sold by store. The ACV of sample stores reflects the relative size of these stores compared to the universe.
If the universe for a market breakdown is 1,000 outlets, and the sample size is 100 (i.e. 10%), the numeric projection factor would be 10 times. Furthermore, based on the census data, if the ACV of the sample stores is estimated to be 12.5% of the ACV of the universe (i.e. these are somewhat larger stores on average), in that case the ratio projection factor is 8 times. By factoring the size of the stores in the sample, ratio estimation provides for greater accuracy.
Ratio estimation also reduces erroneous fluctuations in sales estimates arising due to any churn of stores within the sample. This is important because some amount of churn is expected on account of attrition of the sample stores. If a relatively small store in the sample is replaced by a relatively large store; in that case the numeric projection method would tend to inflate sales. Since ratio estimation factors the store ACV into the projection, it is able to contain any swings in the data arising due to changes within the sample.
For the manual audit, from month to month, there tends to be some variation in the audit time interval. For instance the retailer purchases data collected on 25 March for the store in Exhibit 28.10 pertains to the 28-day interval between 25 February and 25 March. For consistency this time interval or lapsed days is adjusted to reflect a 30.5-day interval, so that each of the 12 monthly periods are equal in size. The 1000 units of sales estimated in Exhibit 28.10 would accordingly be adjusted to 1089 (=1000 × 30.5/28) units.
You may have noted that the data in our example pertains neither to February nor March; it pertains to the period 25 February to 25 March. In the retail index report, it falls under March. This is a peculiarity of manual audits — the actual time period varies from one store to another depending on the audit cycle for that store. Considering that stores are audited throughout the month of March, for stores that audit at the start of the month, the estimates for March reflect more of February sales than March. In interpreting manual audit data, one must bear in mind that the retail audit report for any month pertains partly to sales during that month and partly to sales over the previous month.
Typically for scan channels, the MBD estimates are based on those chains that collaborate with the agency. Estimates for the stores of non-collaborating chains are made based on a combination of stores within the sample that have similar characteristics. If a non-collaborator’s stores are distinctly different from the stores in the retail panel, in that case the non-collaborating chain should preferably be excluded from the universe definition.
The service relies on many external factors that lie outside the agency’s control, and time and again situations arise that need to be resolved. To manage foreseeable situations, the agencies maintain a prescribed range of standardized techniques that are used to improve the quality of the data.
For instance, occasionally data sourced from a collaborating retailer may contain unusable or missing data for some stores. When this occurs, data for the missing stores is imputed based on prior period data for the same stores and projected to reflect the trend (increase/decrease) in other similar stores, for each and every item.
Consider for instance the following sales data for a brand in stores X and Y:
Jan Feb
Store X 1200 missing data
Store Y 1000 1050
If store Y is similar to store X, one may estimate Feb sales of the brand in store X, by projecting the increase in sales in Y (1050/1000 = 1.05) to that of the sales of the brand in Jan:
This imputation technique which is commonly used for estimating missing sales data is called matrix projection.
Note: To find content on MarketingMind type the acronym ‘MM’ followed by your query into the search bar. For example, if you enter ‘mm consumer analytics’ into Chrome’s search bar, relevant pages from MarketingMind will appear in Google’s result pages.
In an analytics-driven business environment, this analytics-centred consumer marketing workshop is tailored to the needs of consumer analysts, marketing researchers, brand managers, category managers and seasoned marketing and retailing professionals.
Suite of interactive, online dashboards that seamlessly integrate retail and consumer data sources in a manner that makes it easier to glean insights.
Suite of dashboards to visualize/analyse retail scan data.
Is marketing education fluffy too?
Marketing simulators impart much needed combat experiences, equipping practitioners with the skills to succeed in the consumer market battleground. They combine theory with practice, linking the classroom with the consumer marketplace.
The Plannogrammer is an experiential learning facility for category managers, trade marketers, and retailers in consumer markets. Ideally suited for hybrid learning programmes, Plannogrammer imparts hands-on training in the planning and evaluation of promotions and merchandising.
It supports a collection of simulation and analysis platforms such as Promotions and Space Planner for optimizing space and promotions, Plannogram for populating shelves and merchandising, a Due To Analysis dashboard that decomposes brand sales into the factors driving sales, and a Promotion Evaluator to evaluate the volume, value and profit impact of promotion plans.