Best use of FastStats 2013 Award
Using FastStats to start building a greater understanding of web buyer behaviour
Lands’ End UK are a multi-channel mail order company, delivering a wide range of own brand clothing, footwear and outerwear. Established in Chicago in 1963 (selling yachting equipment, duffle bags and a few clothes), they have since grown substantially and internationally.
Lands’ End established themselves in the UK in 1993, based in Rutland. The UK base is the European head-office and comprises a state-of-the-art distribution centre, call centre and outlet store. Lands’ End UK currently employ over 400 people - buyers, sales advisors, pickers, packers, hemmers, catalogue production teams, and marketing teams. In the UK, Lands’ End clothes can be ordered by phone, mail or online. Their range is also available at Debenhams and House of Fraser. They recently successfully launched a range of high quality children’s clothes.
Their loyal customer base continues to grow, with increasing retention rates year on year. A significant shift has been seen in recent years towards online ordering, with over 50% of all orders now being placed via the web site.
Results far exceeded expectations in the first 6 months alone. 60% reduction in marketing costs to these buyers, 10% saving in the Lands’ End’s UK total catalogue budget and increased contribution of 23% from these online customers
Project Background / Requirements
A significant % of Lands’ End customers now order online, with approximately 50% of customers now favouring this order channel. Given Lands’ End heritage as a classic mail order clothing company, much of their marketing effort focussed on sending regular catalogues to their customer base. Obviously catalogues are expensive to print and post, so key questions were being asked:-
- Are all customers really being influenced by the catalogues?
- What do web buyers need? Are they all the same?
- Are we servicing them correctly?
- Is marketing spend being used appropriately?
It was their belief that web buyers could be split into three ‘types’
Key objectives for this project were to
- Harness all available data to enable us to successfully identify these types of web buying customers
- Adapt the communications plan to optimise ROI from these customer groups
The critical role of FastStats in this project
FastStats has been firmly embedded at Lands’ End UK for over 3 years, and is central to their planning, targeting and analytical work. It not only harnesses core database elements, but more recently, has harnessed data from the web.
For Lands’ End, like many organisations, the rich data generated from web analytics was largely focused on the optimisation of their online conversion. But this web data held valuable unharnessed insight about the influences of other channels on buying behaviour.
The conversion of this raw session level behavioural data (such as pages browsed, incoming / destination URLs, items searched, elapsed times) into a full history of customer level journeys, was an absolutely critical first step. Fusing this vast online data with the existing customer view, and made available in FastStats, created a powerful multi-channel perspective putting all DM, Call Centre, Email and Website activities for each customer in one place.
So now, in addition to all order history data, Lands’ End could see:-
- No item numbers used / non specific search terms
- Pages browsed per item purchased
- Web visits prior to purchase visit
- Source of web visit (direct URL,affiliate)
- Frequency & level of activity per visit
- Use of other channels
- Email responsiveness
Having this data hub in place was critical to the project, allowing analysis be undertaken to understand behavioural differences between web buyer customers, allowing Lands' End to identify & segment all key behavioural groups.
Once identified and coded onto FastStats, the new communications plan could be executed, delivering significant gains for Lands’ End.
Project Process / Key Steps
The project comprised 5 key steps:
Step 1: Test
All web buying customers were identified and a test was put in place to withhold catalogues for a proportion of these customers.
The objective of this test was to identify how customers perform when NOT mailed – as the hypothesis regarding web buyers stated that they would NOT be influenced by a catalogue mailing.
Step 2: Create the Data Bank
Having brought the online and offline data together in FS, they needed to ensure all relevant attributes were created and available for post campaign analysis. Typically this data included the following types of attributes:-
|What they bought before||What they looked at, when, pages|
|Who they are: age, lifestyle, demographics||How did they search: names, terms, product codes|
|What we think they might buy next||How they came to the site|
Step 3 Analyse
During post-test analysis was undertaken to identify which of the many offline & online variables helped distinguish whether a customer was driven online by the catalogue, and which identified a ‘true’ online browser-shopper not influenced by the catalogue. Some behaviour such as searching for catalogue terms or item numbers using onsite search were obvious, but other equally powerful indicators were only revealed through predictive modelling.
Having identified the key attributes, rulesets were developed to fully identify web browser-shoppers vs. book influenced customers. Using the initial test data, is was possible to quantify the level of incremental revenue that catalogue mailing was actually having for each of the model groups prior to adjusting investment.
- Low Incremental: These are the Web Browser Buyers, not influenced by the catalogue. While they showed a slight uplift in terms of orders when mailed, this was insufficient to justify the cost of the mailing. Objective: cut marketing spend
- Medium Incremental: These we believed to be the Catalogue Prompters. They are high web browsers, but appear to be prompted on-line by the receipt of the catalogue. Unfortunately their lower incremental order rate vs. others has limited their ROI. Objective: switch to lower cost mailing tactics, to prompt online visit.
- High Incremental: The catalogue browsers. If you do not mail them, they do not order! Objective: no change!
Step 4 Prove
All web buying customers were flagged on FastStats according to their type and a new communication plan devised. Once this was understood a careful testing strategy for each segment was then devised: for some customers, catalogues would be stopped immediately; and for others reduced frequency or alternative media was tested. This crucial activity maximised learning opportunities without putting revenue from these valuable customers at risk.
Results: this testing phase validated the initial analysis, allowing the full communications plan to be rolled out.
Step 5 Roll Out
The following chart shows the predicted impact of the new communication plan:
Anticipated Savings: 40% saves in marketing spend to this group of customers.
Results far exceeded expectations:
In the first 6 months alone...
- 60% reduction in marketing costs to these buyers
- 10% saving in the Lands’ End’s UK total catalogue budget
- Increased contribution of 23% from these online customers
For more information Contact Us now.