Lecture 1: Intro and Database Marketing

Lecture 1: Intro and Database Marketing

1. is about marketing decisions based upon data and analysing data

2. Direct Database Marketing

   • Methodologies: between statistical based and computer based, regarding data and how to analyse and use data
   • Sales Promotions (Campaigns): Analyse/ Plan and Prove that campaign is successful. Increase in sales (compare both pre- and post-campaign). Analyse conducted after to understand the campaign affected sales, awareness, customer feedback etc.
   • Trade Promotions: not direct to the customer but through the retailer (in-store promotions, displays, etc.)
     • Normally done by manufacturers rather than retailers. To do so, manufacturers provide trade promotion to retailers so that the latter can promote the products instead of the former. Some trade promotions work very close with retailers (e.g. P&G runs a trade promotion with Sainsbury), displaying the products on the best isles, whereas others just give a one-off discount to retailers.
   • Catalogue merchants: Ex; Argos: Sending catalogues to clients and letting them shop via the catalogues (different from retail stores, in the sense that they cannot touch or see the products physically before purchase)
     • Note: No real ‘marketing campaigns’, but rather catalogues are sent to customers or potential customers -> targeting customers with different catalogues based on their purchasing behaviours
     • Expensive
     • Modern Version: Email Marketing/ effective, if targeted and customised properly
       • (-) Spam, (+) Customized to past purchases and customer behaviour
       • Credit scoring: Large bulk of information is collected and then distilled into one single number - > allows us to rank the customers in an order (ex; 10-1)àcustomer information is mashed into a single number, and based on that information a decision is made
   • uses historical purchases, demo info.

3. Web Analytics

   • Clickstream Data (data collected online) Important data set: Contains every action that a customer does on a website; all clicks, IP addresses, etc. are recorded on the company database
     • Ex; from the collection of the IP address -> location -> demographic of location -> customer profile, etc.
     • The information it has is very detailed: IP address(where the customer searches from, etc) → enables company to create customer profile+ cookies( enable to track customer over time; not only on your website but across multiple sites)
       • Managerial Implications: - Real-time targeting (ads that are relevant to customer)  - Customer profile (detailed/ without having to know identity of customers)  -web analytics (developed personalised messages to customer on a web) - cross selling and up selling opp. - Amazong & Netflix - Ability to analyse marketing campaigns (keywords marketing/ SEO)
   • Multi-channel campaigns - Looking at the data and trying to understand what was worthwhile and what wasn't (from all selected channels)
     • Not easy to / have to allocate campaign budget over various channels then correlate sales with you message in every channel
   • Price Promotion: Looking at sales data and trying to understand the price sensitivity of the customer (demo , geo, channel etc.)
     • Model-based analysis: The initial price might be a problem in the market primarily due to inaccuracies of estimation and therefore companies use price promotion to sell more of a product. However, it is very difficult to estimate the value of price promotion primarily due to price sensitivity of customers across different regions.
   • Scanner Data: richer than sales data/ all the data that supermarkets, retail stores, point-of-sale terminals collect whenever products are checked out. information provided about the shopping basket, information on loyalty cards, etc.
     • Potential Analysis: Price sensitivity/ Effects of promotions (expected sales)/ Which products do customers buy together/ Which stores have what buying patterns
   • Panel Data
     • (Similar to loyalty cards, but a lot more controlled) Panel = Randomly selected group of customers that are representative of the market
       • (+) Information is also provided from other stores; vs. Scanner data that only provides the information on your store. (+) When panel data is provided from company (ex, Nielsen), information on demographics and other variables is provided as well.
   • Uses real-time information (likes/ clicks/ searches/ purchases etc.) triggering marketing actions (i.e. recommendations)

4. Analysis: RFM Analysis = (Recency, Frequency, Monetary) ->Very traditional and very basic; widely used in DB marketing

   • Ranking/categorising customers by the following:
     • The recency of the purchase
     • The frequency of their purchases
     • The amount spent on their purchases (avg. purchase amount/purchase)
     • The higher along these three dimensions, the better the purchase probability
   • Methodology; Take a historic customer list/ Rank them across the recency dimension/ Group them into a small no. of group = Binning (statistical procedure - concentrating the recency)
   • Significance of RFM score: Ex; customer with a score of 555 = Good prospect that they will make a purchase during their next visit/receiving catalogue/email etc.
     • When attempting to target customers, 555 customers are those to approachàRFM provides an indication of future purchases
     • Experimentation: Picking up a sample and checking thei response = Stratified sampling.
       • Using RFM to predict response: within each group we are going to take a smaller sample (hence, stratified, group within group) and test the latest products/catalogues/emails on them and use their responses for future targeting
         • Generating more information, which is correct because it based on current sales and current information (vs. the historic information used initially)

Images and files in the map