Marketers sometimes struggle to bridge the void between the theoretical concepts and the practical real-world limitations.
Unlike financial statements where dollars and cents must add up, marketing research/analytics reports intended to impart an understanding of market dynamics, do not require complete, 100% coverage.
Moreover, research firms and their clients must make compromises between the costs of expanding services and the benefits of doing so. This remains a business decision — clients have budget limitations.
Like other forms of market intelligence, retail tracking data which meets acceptable accuracy standards, irrespective of coverage levels, remains useful, provided one is aware of what is covered and what is not covered, and one appreciates the strengths and limitations of the data. Utility, of course, does increase with coverage.
The retail measurement service (RMS) tracks the sales of goods from retailers to consumers, at specific outlets in a predefined geographical area referred to as the retail universe. This universe is usually not the same as the supplier’s sales territory. For example in China (Exhibit .5), NielsenIQ tracks the densely populated urban provinces of China, leaving out the sparsely populated regions in the West.
Moreover, as depicted in Exhibit .15, within the same geographical territory covered by a retail tracking service, the supplier may sell to outlets at locations such as schools and military establishments that are not accessible to retail auditors.
There are several such inaccessible or non-participating outlets, and many of them are important contributors to sale of some product categories. For instance, stalls at enclosed construction sites sell a lot of cigarettes. Or for instance, non-participating outlets at tourist locations sell large quantities of impulse products like ice cream and chocolates.
The difference between the supplier’s shipments and the agency’s estimate of store purchases is the coverage gap (Exhibit .16). This gap arises due to the following reasons:
As mentioned above, the build-up or depletion of pipeline inventory affects coverage. Some of the many factors that cause these fluctuations are listed below:
When a product is introduced, stocks start to fill the pipeline as they move from the supplier to retail outlets. While the pipeline expands, retailer purchases grow at a slower pace than shipments, and coverage is below norm. When the pipeline fills up and the expansion of the distribution network starts to level off, coverage starts to stabilize. (Refer illustration in Exhibit .17).
At a later stage, when the product is in decline, the pipeline starts to shrink, and pipeline inventories decline sharply. During this stage, shipments are lower than retailer purchases, and this raises the coverage level.
The accuracy of the purchases and sales estimates is affected by the level of distribution of a product. When the distribution is low, the number of stores stocking the product in the agency’s sample of retail outlets is small. As distribution expands, the product is likely to be found in more stores in the retail sample. The resulting increase in the effective sample size improves the accuracy of the purchases and sales estimates. As can be seen from Exhibit .18, this reduces the volatility in coverage estimates, which is markedly high when the product’s distribution is low.
Differences between a supplier’s shipments and the agency’s estimate of store purchases can be difficult to comprehend due to the complexities of the distribution networks, pipeline effects, practices such as investment buying, impact of promotions and the peculiarities of a retail tracking service.
A coverage analysis can impart an understanding of the factors contributing to the discrepancies. And, as can be seen from the following case examples, the analysis can also reveal very useful insights into the dynamics of the supplier’s distribution network.
In the last quarter of 2007, a few years after Kraft Indonesia took over Danone’s biscuits business, the company transformed its distribution network. The previous network relied on one national distributor, encompassing 26 branches and 28 sub-districts. The new distribution network, named Builder, relied on 15 territory level distributors, encompassing 39 branches and 13 sub-districts.
Kraft estimated that retailer coverage increased from 87,000 (pre-Builder) to 165,000 stores, by end 2007.
The company’s shipments of biscuits surged by roughly 40% in 2008. This achievement was one of the reasons CEO, Irene Rosenfeld, made it a point to meet the Indonesian team on her visit to the Asia Pacific region.
The only dampener was that estimates by Nielsen Indonesia showed substantially lower sales. This discrepancy led to a coverage analysis and an in-depth investigation into the Indonesian retail tracking service.
One year later, in 2009, Kraft’s biscuits shipments plummeted, the decline largely compensating for the surge in 2008. This confirmed what was inferred from the earlier investigation.
The key reason for the surge in sales in 2008 was the creation of the parallel sales pipeline. The new Builder distribution network was larger, and importantly, the depletion of stocks in the old pipeline was occurring at a far slower pace than the build-up in the new one.
As a result, a large proportion of the 40% increase in sales was due to the increase in the pipeline inventory, which for a big country like Indonesia can be quite large.
When the expansion of the Builder distribution network levelled off, this coupled with depletion in stocks in previous network, resulted in the sharp drop in shipments in 2009. Shipments stabilized once stocks in the phased-out network depleted, and it ceased to exist.
During the transition, while shipments jumped up and down like a yoyo, the Nielsen sales estimates grew at a steady pace. The turbulence in the intermediary networks did not have an adverse effect on consumer offtake.
This case goes to show the importance of relying on a well-designed service to track consumer offtake. It also illustrates the use of coverage analysis to enhance our confidence in retail tracking data and improve our understanding of the dynamics in trade.
Some years back, when the lower trade was big in Singapore, Organics shampoo year-end shipments to distributors surged from an average of 17 thousand litres per month to 53 thousand litres. This was primarily the outcome of a trade promotion, and though retailer purchases reflected a relatively small increase, consumer purchases remained flat.
Enticed by the promotion, distributors merrily stocked up, inflating the pipeline. And because they did not pass the discounts to the retailers, there was no incentive for them or their customers to purchase more than their usual requirement.
In another example of investment buying, the Campbell Soup Company sold huge quantities of canned soup to a dominant supermarket chain by offering a large trade discount. Subsequently the manufacturer’s country head was disappointed that there was no discernible increase in scanned sales even several months after the promotion.
Abbott Laboratories, the manufacturers of infant formula, substantially increased their sales in Malaysia by expanding into newer channels of distribution. Unfortunately, for the Abbott team, these channels were not covered by Nielsen Malaysia, and so, it turned out that the Nielsen data reflected a decline in sales.
The decline was because, in addition to growing Abbott’s overall business, there was some cannibalization of the established channels covered by Nielsen Malaysia, by the new channels not covered by the agency.
Nielsen’s market share was a KPI for Abbott, and, despite explanations, the reported “decline” did not go down well with the regional/global bosses.
As can be seen from this example, an understanding of the strengths and weaknesses of a retail tracking service, particularly in the context of coverage, is of vital importance in the interpretation of the data.
Coverage analysis is an important exercise that helps the research agency and its clients assess coverage, and understand the reasons contributing to the gap. This understanding is used by the agency to prioritize improvements to their service. And as illustrated by the case examples, it can reveal very useful insights into the dynamics of clients’ distribution networks and guide them with their interpretation of the data.
Coverage analysis is based on moving annual totals (MAT) of reported retailer purchases and manufacturer’s shipments. It relies on three parameters:
It follows from the above definitions that: $$Coverage \,= \,Expected \,Coverage \,× \,Pick \,Up$$
Coverage reflects how the research estimates relate to the supplier’s total shipment and is dependent on the extent that the retail universe captures the manufacturer’s shipments (expected coverage) as well as the accuracy of the research methodology (pick up).
Pick up is a measure of the accuracy of the research methodology and is dependent on the sampling framework as well as non-sampling errors.
Example (see Exhibit .19) illustrating coverage, expected coverage and pick-up:
Since coverage analysis is conducted on moving annual total (MAT) sales, it moderates pipeline effects and smoothens out the impact of promotions and seasonality. It also evens out the variances in the manufacturer’s shipments which can be volatile with peaks and troughs over months.
The sales territory in a coverage analysis should represent the total country. This reduces errors due to intra-country movement of goods.
Note also that coverage analysis is appropriate for brands with numeric distribution of at least 80%. This conforms with Nielsen’s global standard for sampling error, which applies to products that are available in 80% of the universe. Estimates for products with lower levels of distribution may not meet the agency’s global norms.
Manufacturers should review coverage for major brands once in every two to three years, or more frequently if there are concerns regarding the quality of the data. At the agency, the measurement scientists (or data scientists as they are now called) should be conducting the analysis on a regular basis as it provides critical information on how their service can be enhanced.
Use the Search Bar to find content on MarketingMind.
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.
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.