Inventory is the largest investment in the jewelry trade. A chain of jewelry stores needed to bring costs down by optimizing its inventory - what SKUs to stock, where, and in what quantities. In the process, the client also wanted insights on new product possibilities based on sales history.

We first collected data - what items sold in what quantities, and how long they sat on the shelves. Based on this, we divided stores into four types and used (Can we insert the relevant methodology here, and hyperlink it to the 'Common Analytics Terms' section?) to create inventory rules for each. This led to lower inventory levels in stores. Sales increased, as customers found more variants that appealed to their tastes and budgets.

Studying the clusters further also threw up buying patterns that led to insights on the likely sales of new products. So, new products could be selectively sold across stores, instead trying all stores and course correcting over time. Point-of-purchase customer data was also invaluable in creating better customer promotions, new merchandise and in picking merchandise mix for new stores in similar areas.

The benefits from this exercise-lower inventories, improved product mix, increased sales, fewer discounts and lesser stock-outs.

In the highly-competitive Asian mobile phone market, the store is the 'last chance' for a marketers to sell their brands. A large media and promotion company found that, while it spent large amounts on different in-store promotions, it didn't really know which elements of its promotional mix really worked.

RainMan first classified and the elements within a store such as shelf space, wall space, counter space, no. of sales staff by brand, no. of products on display, POP and other promotional displays, the outer signage, etc. It then collated the resulting data from over 2000 stores across India. This was done daily and weekly, as this was a dynamic category in a dynamic market.

Using generalized linear modeling (GLM), six months' data was organized in a stack. We found that the results were unique for sales of each brand. Drivers also differed depending on the size of the store and its location (both neighbourhood and city). RainMan also created an online dashboard that made it easy for field salesmen to report this data with minimum time and effort.

This gave our client a deep insight into what inputs to use for specific brands, in different-sized stores across different geographies. This, in turn, led to efficient budget deployment and greater market impact, both at a lower cost. The tool was so successful that it is now being extended to other markets.

Things go out of fashion quickly in the premium garment market. The resulting dead stock, most often, is sold at deep discounts, affecting profitability and a brand's premium image. A leading garment brand wondered whether it could buck this trend and supply only as much as the market could absorb.

Closely examining the process, RainMan found that, the brand had large amounts of data in its ERP and Point of Sale (POS) systems, but wasn't using it in its indenting process. Instead, it relied on the judgement of a few key people, whose departure could have an adverse impact.

RainMan first broke down SKUs into their component parts-type of cloth, color, size, style, findings type, etc. It then modeled this data with sales information derived from the POS system. The model used an algorithm to link each component with the drivers of. When tested with data outside the modeled set, the model showed an improvement of close to 12% on the actual indent placed for the season.

With a better match with sales, the brand benefits from lower stock outs and inventories, both leading to higher profitability and, more importantly, a premium brand image.

A computer hardware major's 20-strong Key Account Management team handled over 9,000 accounts, across different product groups such as Desktops, Laptops and Accessories. And, it had a classic problem - sub-optimal productivity. RainMan was tasked with increasing three variables; conversions to sales per visit, ticket value and the number of products sold to each client.

The approach used was to define the probability of a client buying his or her next product, using a statistical approach called the Cox's Proportional Hazards Model. The questions that the modeling answered were-when to sell what and to whom? The model took into account variables like past behavior, type of industry, seasonality, company size, etc. The most probable (and profitable) customers were assigned to individual salespeople, with a list of specific product/s to offer.

Over 90% of the subsequent sales calls were successful. Better still, nearly 70% of clients picked the product the model had predicted. This led to an overall improvement of sales of close to 20%, with no cost increases, a hefty increase in profitability.

A consumer goods brand and its variant were once growing across Asia, but were losing sales due aggressive competition. It also had a typical problem - not knowing clearly what marketing inputs work across market clusters. RainMan's task was to identify the brand's key sales drivers of sales and help effectively allocate resources to maximise ROI.

RainMan built an econometric model. by carefully choosing relevant marketing variables, including distribution, advertising, pricing, variants, marketing inputs and competition inputs. Multivariate regression, along with Bayesian shrinkage estimation process then estimated the short-term and long-term impact of various inputs.

The study found various drawbacks and, importantly, prioritized action points to correct them. Distribution quality was the first area, followed by a revamp of the brand's TV creative and media strategies, which were delivering low lift and high decay. Price sensitivity also indicated where prices could be safely raised, where they needed to stay stable and where other sales drivers could negate the decline in sales due to pricing.

The implementation of the new strategy resulted in an increase of marketing ROI by 20% across all market clusters, with a significant growth in sales numbers.

A consumer goods brand that had been a category leader for over 2 decades was suddenly buffeted by competition. The brand's communication had been a consistent mix of functional and emotional messages, but the competition aggressively used its inputs and pricing to grab market share. Pricing adjustments and consumer promotions partly improved the situation, but the company also felt that, to regain its market position, it needed to understand how consumers felt about its brand imagery.

To do so, RainMan created an architecture that linked sales with mind measures (top of mind awareness, spontaneous awareness, brand consideration levels, etc.) and then linked those mind measures with brand imagery. It showed which mind measures best increased sales, and then suggested imagery and media metrics that could best meet them. We then applied this architecture to an extensive amount of consumer brand track data, using factor analysis to avoid redundancy.

Multivariate regression analysis estimate relationships between sales with mind measures and mind measure with imagery statements and, finally, the relationship with TARPs. RainMan's marketing domain and modeling expertise helped in identifying the right imagery drivers of consideration.

The brand was relaunched using the modeling results. The outcome was 12% improvement in the sales, which continue to grow.The brand team continues to monitor the relevant variables out of consumer track research, keeping the brand healthy and enhancing media ROI.

A mature consumer products brand was losing market share and wanted to regain its past glory. RainMan was asked to identify the key share gain drivers and recommend an effective marketing mix allocation strategy.

Consumer household purchase panel databases were used to generate insights. By using analytics on these databases, RainMan got insights on market structure, and consumer brand switching patterns. Next, we defined consumer profile groups and competitive brands that met their same needs as our brand (for example, a health drink may also compete in the breakfast food space). Thus, a need based competitive construct was prepared to get a complete understanding and estimation of market size.

Using cluster analysis , four consumer groups emerged. Using Cox's Regression Technique, we estimated the probability of consumers switching between our brand and competition, taking into account household characteristics, exposure to advertising, and other marketing variables. Market mix modeling then generated insights on what could work as inputs for brand switching. These were used to devise a strategy for market mix allocation.

As a result, resources were optimally deployed across selected marketing inputs. The total cost was kept the same as the past period. This allocation led to an increase in share by 4 points over three months.

With the regulatory regime changing, an insurance company needs to have sufficient capital to cover risk. By having a well-specified model to estimate the probability of Surrender, Lapsation and Claims, it can optimize capital based on the risk of its portfolio. The same model can also be used to identify vulnerable customers to reduce Lapsation and Surrender, and target less risky customers

Using a logistic regression model, RainMan scored customers based on a number of variables - demographics, environmental variables like interest rates and stock exchange indices, as well as behavior data like purchase and payment modes.Regulatory oversight meant setting high levels of significance, and easily available data was used to keep the model practical and usable.

Using the model, the company found that 90% of the lapsers were in the top three deciles of customers organized according to their probability of lapse (anything closer to 1). This helped the company identify vulnerable consumers and device customer retention strategies. The upshot - a good ROI and reduced risk for the Insurer.

With online penetration increasing across Asian markets, the medium has been witnessing an increased amount of clutter. A consumer products company needed to know online's role as a sales driver, so that it could optimally use the digital space in its brand Brand strategy.

RainMan created an econometric model based on carefully chosen marketing variables. These included physical variables distribution, TV advertising, price, quality of distribution, marketing inputs and competition, as well as online inputs like media impressions, search engine marketing and social media mix used by the brand in the past. Multivariate regression and Bayesian shrinkage estimation were used to estimate the short-term and long-term impact of various inputs.

We found that, of all online inputs, only media impressions had a direct impact on sales- a 1% increase in inputs would result in 0.02 percent increase in cases sold. Also, media impressions along with search engine marketing and social media impacted sales by enhancing spontaneous awareness of the brand in consideration.

Using these insights, the client implemented a new strategy, resulting in marketing ROI increasing by 16%. It also developed a focused strategy for online media by clearly mapping its role and performance.

A consumer products brand had a presence across multiple connect points: paid media, owned media and earned media; but did not know the relationship between them, and what worked across market clusters. It asked RainMan give it a clear insight as to what influenced paid, earned and owned media, and how they affected brand sales and mind measure.

RainMan conducted an econometric modeling exercise. We found that there was a clear relationship between "paid", "owned" and "earned" media and brand awareness. Similarly, media clearly influenced brand sales. We could also quantify the contribution of each media category on sales - 70% through "paid" media versus 70%, and 30% through "earned" and "owned" media - and brand awareness ( 60% through "paid" 60% and 40% through "earned" and "owned" media.

By clearly understanding these interrelationships and their influence on brand performance, the client was able to better manage connect points in terms of content and layering. This Improved its ROI of connect points by 12%, with significant growth in mind measure and sales.

A consumer brand had been airing multiple TV commercials on television. However, it didn't know when to stop a particular creative and replace it with a new one, so as to keep sales increasing. RainMan was asked to measure the copy wear out for each copy and recommend an effective strategy for creative portfolio deployment.

An econometric model created by RainMan carefully chose relevant marketing variables. TARPs of different creative aired, along with quantified copy characteristics were modeled with sales in addition to normal marketing input variables. Estimated ad stock values were subjected to regression analysis and the most significant ad stock (for the overall campaign during analysis period) was picked up. This was then analysed using Bayesian shrinkage estimation by passing it through a Kalman filter algorithm. Each copy characteristic variable was regressed to measure copy impact and media impact separately. Out of eighteen copy variants aired, only four had a significant contribution to sales with the best creative's impact decaying by 12% per month.

The outcome of this exercise was an insight into effective creative portfolio management and implementation of optimum media weights as per the copy effectiveness and its wear out.

A fashion brand was launched in some Asian markets in Asia. Before its launch, however, the brand team wanted estimate likely demand, as the brand was supposed to cater to a certain mindset segment.

RainMan began the exercise using multiple sources of data. Secondary research data on consumer attitudes and mindset mapping with brands, media data, and macro economic variables used to estimate the demand. Using consumer mindset brand linkage data, cluster analysis mapped segments with life style brands consumption and estimated each cluster's purchase pattern. Econometric modeling estimated the likely size of each consumer clusters for the successive five years. Parallel models were generated by incorporating macro economic variables and life style product penetration to estimate the size of each products and market potential. The insights from the series of models developed were integrated to arrive at an estimate of the demand of the brand in consideration across markets.

Accurate estimation of demand helped the brand to cater to the core target segment's need with no wastage. This helped the brand to generate a healthy bottom line right its launch.

The salesmen of a large consumer product company were facing problems. They dealt with hundreds of SKUs, and found it difficult to scroll through hundreds of SKUs on their handheld devices, and then sell additional SKUs from the same list. RainMan was commissioned to identify a list of SKUs for each store with a high probability of purchase. The objective was to predict all the sales within a certain number, say 30, if a store typically purchased 15 SKUs.

RainMan used a probability scoring method from the insurance domain to list SKUs from high probability to low probability for each store. The model also eliminated SKUs with a low probability of being ordered. This allowed salesmen to focus on relatively few but high probability SKUs and get better conversions.

On validating the model, the algorithm showed a high predictive score -out of the SKUs bought, 90% were correctly predicted. Enabling better cross-sells and up-sells resulted in a significant lift for the additional number of new SKUs sold, which in turn had a high correlation with the improved morale of the sales force.

An Asian consumer products group, spent a lot on TV advertising. But, it didn't quite understand which channels were the most effective sales drivers. It also wanted to determine which components - national versus regional channels, genres and broadcast times - would drive sales the most.

RainMan built a market mix model between sales volume and various marketing inputs, such as distribution, price, overall TV weights, and competitions' factors, to measure their impact on sales. Having established the impact of TV on sales, TV impact was then analyzed at a micro level with unique statistical techniques. The contribution of TV was broken into the contribution of its components and was optimized with the existing budget as a constraint.

This analysis provided insights on key genres, day parts and channel mix that had a bigger impact on sales. It led to three optimization scenarios which, when translated to media buying guidelines, led to a scenario of 4% increase in sales without changing the overall TV budget.

A major food company had, till recently, measured price & distribution attribution at an overall brand level in their market mix, but not at a pack level. This led to inefficiencies and stock pile-ups, cutting into margins.

RainMan created a market mix model with drill-downs on price elasticity. The result was the discovery of price elasticity by pack, allowing for differentiated pricing. This allowed the marketer to take up the prices for the less price sensitive variants while protecting those more likely to be hit by price increases. This led to more optimal pricing adding as much as 6% to the margin, making the company as healthy as a freshly plucked apple.

Promotions are an important part of consumer goods marketing. The client - a consumer goods major - needed to identify the most effective consumer promotions - both value promotions (price discounts offered), and volume promotions (free gifts) - from the many it had deployed, so that a result bound promotion strategy could be drafted.

RainMan used all the client's consumer promotion variables on sales and market share to create a customized marketing mix model. Then, the contribution of promotions was measured by the kind and depth of discount offered, publicity the promotion received and its duration.

By understanding the drivers of success for consumer promotions, the client was able to create a more effective promotion strategy. The result was that it streamlined its promotional strategy, actually saving on promotional spends while delivering higher sales.

TV weight setting exercises usually stop at identifying the optimal TVRs a brand needs. But, this often ignores the frequency needed to optimize the brand's reach. Also weights are often set in terms of "effective frequency" or "effective reach". So, the only measure the exposure to be achieved beyond certain level of frequency and ignores how optimally exposure is delivered at discrete levels.

Giving importance to this core issue, consumer brand brand wanted to fine-tune its TV planning guideline by using optimal discrete frequency as a key input. RainMan identified the optimal TVRs required by the brand to grow its business metrics, then extended the process by bringing in TV frequency distribution as one of the inputs. This complex modeling exercise threw up at the ideal frequency band the brand needed to concentrate on to optimizing its reach.

As a result, the brand reduced the over delivery of TV weights as seen in the past by 20% and increased the TV effectiveness by 12%. This process of weight setting helped the brand regain market share and to gain dominance in the market!

A leading Chinese consumer goods manufacturer had a goldmine of information on the various creatives it had aired for one of its flagship brands. However, it found that its TV weights were set without integrating its brand creative scores. Hence, it wanted to integrate its creative scores to scientifically arrive at a TV weight setting guidelines, especially since the brand was using multiple creative.

RainMan's unique approach involved creating a unique score that had factored media weights and creative score. This unique score was created across the creative executions and this score was used as the input for TV impact modeling.

As a result, the client could identify the media TVR needed to support creatives to get maximum response. Half-life analysis also determined the optimal time the creative needed to be aired. This brought big benefits to the brand, as TV advertising delivered higher response while avoiding overexposure. Campaign planning, therefore, was fine-tuned and RoI maximized.

A beverage brand had three main variants, all with high media support. It therefore wanted to optimize media inputs of all the three variants by capturing the interdependencies between them.

Appreciating the presence of cross relations between the variants, RainMan used vector auto regression (VAR) to reflect multiple outcomes for the same input - meaning that a variant's input could lead the benefit to other variants. VAR thus identified marketing inputs on these multiple outcomes by capturing the interdependencies between the variants.

What we found was that there was a high impact of halo advertising - ad stock of one variant had an impact on the other variants. This impact was increasing year on year. The long-term impact of advertising was seen quite prominent for one variant and was three times that of the other two variants

These insights were used to optimize media inputs for the entire brand portfolio and generate an optimum media input schedule across variants. As a result, by investing more on variant one in mass media, and keeping the total cost constant, overall brand sales be increased by 3%.

Pricing is one of the key decision for a brand or a retailer. A number of factors need to be taken into account while setting the price. One key resource available to the marketer is past data of sales at a daily level (from own showrooms), or weekly data from Multi-Brand Outlets (MBOs). As it happens given competitive pressures and stock clearances - prices for the merchandise will be varying over time. This is a gold mine of information, and can be used to optimize pricing and improve margins.

One brand wanted to understand if price increase can be taken on some of its categories, such as Shirts, T Shirts, Pants of different types, Blazers, Shoes etc. After understanding the data across own showrooms in 3 states and modelling the data to understand the price elasticity by category, RainMan suggested an optimum pricing by category (either increase by a certain price, keep the same or decrease by a certain price) such that the overall rupee sales went up while keeping the volume higher or constant.

The analysis clearly indicated the inefficiency of earlier pricing and suggested the need to use data available to improve the quality of price setting.

A Global Premium cosmetic brand with high presence on digital media reported decline in sales in Asia. The brand, wanted to identify its drivers of sales and optimize media deployment in order to arrest the fall in sales. The brand had presence on various platforms like TV, Print, OOH, mobile, facebook, iVideo, iMedia and Search, with facebook and mobile (non-facebook) having a lion�s share of investment. Investment was slightly skewed towards digital media.

Within digital media, facebook was one of the lead vehicles. RainMan used innovative technique to arrive at the quality impressions on facebook that consisted of not only paid impressions but also a quantification of consumer engagements through various metrics of facebook. RainMan thus used its proprietary weighted technique to derive �facebook quality measure factor� from customer interactions data. The total facebook Impressions weighted with quality measure was considered for modelling.

Linear Regression technique was used to build model which were used to identify key drivers of sales. It was discovered that, digital media was the highest contributor of sales with facebook being the leader. RainMan then did deep dive analytics on digital media and generated hidden gems that were to drive the sales through careful digital planning.

Using the insights derived from the modelling exercise, marketing budgets for various key drivers of sales was reallocated using mix optimization. This resulted in an additional 8% lift in sales!

Optimizing the in store visibility and consumer promotions to enhance the sales

Forecasting the demand of various stock keeping units (SKU) across modern trade chains is a key challenge for any brand. Instore visibility and consumer promotions play an important role in addition to other mass media in driving sales at the modern trade outlet. RainMan through its analytic expertise implemented a planner tool that helped the client to plan the consumer promotions and visibility investments for 500 SKUs across four categories by retail chains. An optimizer was created and embedded on to the planner software to optimize the consumer promotion and visibility spends to enhance the net turnover for the SKU by chain.

RainMan created the homogeneous stacks of data and used Hierarchical Linear Model technique to identify the drivers of sales at modern trade outlets. RainMan successfully forecasted the SKU volume demand at modern trade outlets with less than 10% Mean Absolute Percentage Error (MAPE) across all SKUs.

This forecast helped the client in better management of inventory, instore visibility, consumer and trade promotions thus enhancing the efficiency by 20%!