by James Lawson.

Analytics work across the UK takes many forms, but with the inaugural Database Marketing Awards debuting this month, let’s remind ourselves just how good the UK is at customer analytics.

Honing

“Database marketing used to be very much batch-based and focused on limited customer data,” says Andrew Campbell, Managing Director – Insight at 20:20 Dialogue. “The new model is more akin to the sort of real-time, online transaction processing environments found in operational systems. This is a very different data and technology challenge, and one we’re increasingly asked to help clients with.”

Using all the data available in innovative ways is what this is about. In the travel sector, it’s possible to use inventory availability data to personalise offers for ancillary cross sales.

“The My Holiday Countdown campaign we ran for Thomas Cook more than doubled sales of airport lounge facilities,” says Campbell. “Our recent work has taken that concept to far greater levels of sophistication and personalisation.”

G2 Data Dynamics’ ten-year relationship with Adobe has seen the company’s data-driven work steadily increase in effectiveness. With evolution and refinement of the various analytical and operational techniques key, the latest campaign for the launch of Adobe’s flagship CS5 software suite was the most profitable yet.

“We have now done a lot of work on managing relationships over long purchase cycles using customer data,” says G2 MD Kevin Slatter. “You have to nurture customers in between upgrades. It’s critical to understand who will skip upgrades and what the upgrade path is.”

Professional graphics companies will always upgrade to be compatible with their clients while one-man bands are more likely to wait till they are forced to spend money. “We have to target these different audiences in the right way,” says Slatter. “If they are ready to buy then they will get a hard sales message; if they might skip they’ll will get a softer treatment.”

G2 built the Adobe global customer database from scratch and now hosts and maintains it across three offices in the UK, America and Asia/Pacific. There are 60 feeds in Europe alone and analysis has benefited from a decade of data acquisition. For example, extensive historical data on lifetime value means that the agency can now predict future value very accurately while geocoding, channel preference and RFV values are only a few of the variables available. There are multiple propensity models covering various strands of behaviour, as well as being used to distinguish business buyer from consumers, and the level of penetration of the product into a company.

Five waves of communications went out through April to June 2010, breaking revenue records for the various regions as it progressed. Over eight million individuals were contacted and the campaign eventually generated around $119m from 152,000 product orders.

“There’s been a lot of test and learn,” says Slatter with a fair degree of understatement. “The CS5 launch revenue outperformed the previous launch by 45%.”

Sometimes the poor relation of commercial operations, charity marketing has become increasingly sophisticated in recent years. Kidney Research UK has multiple income streams but legacy marketing and fundraising is crucial. Typical legacy values are quite high and maximising this source was key for the charity. To this end, it worked with Tangible Data to identify and attract future legacy pledgers from its existing supporters.

This involved identifying how the profiles of legacy pledgers, considerers and legators differed from the overall supporter base by considering demographics, supporter behaviour and value, and communication patterns. First the charity had to select only those it had a relationship with from around 500,000 supporters on its database.

These included active and lapsed donors, event participants, shop volunteers, lottery participants, kidney health information contacts and patient survey respondents. Of those with some form of legacy relationship, around 1,300 were legacy givers with the remainder pledgers and considerers. These audiences were matched and processed to form the analysis dataset.

The clean, consistent data and a clear brief from the client helped greatly. “They used the information from analysis very intelligently, particularly in telemarketing,” comments Nigel Magson, MD at Tangible Data.

Tangible built a legacy targeting model in SPSS using logistic regression, scoring supporters according to their likelihood to pledge a legacy. The results were modelled in Excel, and the scores written back to the supporter database.

Top scorers were invited to a nationwide programme of legacy receptions to further increase pledger value and loyalty by providing opportunities for cross-selling other products.

Actual communications involved an initial mailing, a call by an external telemarketing agency and selective follow-up activity carried out by the charity’s legacy team based on the quality of respondents. Response was classed as: identifying a pledge the charity was unaware of or contact from those intending, considering or confirming a legacy pledge.

The result? An unprecedented 30% positive response. The charity identified around £785,000 worth of actual pledges, an ROI in excess of 50.

In the retail sector, Indicia worked with Sainsbury’s Finance to help re-position and grow the business by increasing cross sales and reducing attrition. Key to the work was Optimiser, Indicia’s campaign optimisation tool. This aims to calculate the most efficient and profitable overall mix of channels, budgets, contact density and so forth over an extended period, rather than campaign-by-campaign, minimising the risk of overcontact.

“The project was a ‘next best offer type of approach’, but there were conflicting metrics,” says Mike Fisher, Chief Futurist at Indicia. “We had to create a multiple metric able to cope with both response and ROI. That is, you may be more likely to respond to an insurance offer, but you are more valuable as a credit card customer.”

Indicia first built an analysis data set from sources across the company, including Nectar retail data, then embarked on building the many propensity models an optimisation approach requires. Both detailed and more general retail behaviour was key to this. For example, buying pet-related purchases would make an individual more likely to be interested in pet insurance, while other metrics such as “overall monthly spend”, “number of items in basket” and “range of items purchased” were also influential.

“We needed to test and back check many times to get the analytics correct,” says Fisher. “There were also a large number of business rules that had to be worked within the optimisation.”

The campaigns were delivered across multiple channels and overall the work improved sales by 13%, with the best campaign achieving over 25% more than the norm – far surpassing the target of 5% to 10%. Against the control, Optimiser uplifted overall campaign response from 0.71% to 0.97%, at the same time as making a significant reduction in the DM spend.

Work by Cancer Research UK (CRUK) to promote its annual Race For Life (RFL) events incorporates all the elements of best practice discussed above: data collection, friend-get-friend, personalisation and detailed tracking. As part of a gradual ramping up in sophistication, Rapp brought together content from the website with the marketing database to customise emails to the CRUK database; runner images were selected based on the age of the recipient for example. Adding in nearby race sites as options doubled the conversion rate but the big step was letting recipients forward the email invite, post it on Facebook or talk about it on Twitter with one click, instead of being linked to the RFL website where they would have to fill in various details.

“It was all about making it as easy as possible,” says Ryan Davies, Digital Product Consultant at Rapp. Clicking on one option popped up a fully personalised email inviting friends to join or sponsor them, while another link offered a Twitter message with similar, though summarised, content and a short URL for tracking purposes. Clicking on the Facebook link generated a custom web page based on personal data and dynamic content for use on Facebook (rather than re-using the original email content).

“We customised that Facebook page content to appeal to friends rather than just the individual,” says Davies. “When friends click through from the Facebook page then we can link that back to the original recipient. The email was opened by 21,000 people, shared by 3000 but potentially seen by 400,000 people.”

Specialist over-50s insurance provider RIAS exceeded its expectations for marketing efficiency improvement when it revamped its contact programme. Using IBM predictive analytics software, it built a new set of predictive models based on demographic and transactional data to select target groups and deliver personalised offers and content across direct mail, inbound / outbound telephony, and web campaigns.

Previously, RIAS contacted its customers through direct mail and telephone, a strategy that had worked well for some years. However, campaign selection relied on a marketer’s own knowledge allied to a spreadsheet system, so little in the way segmentation. Customers were being contacted multiple times with information that was not always relevant to them, resulting in lower returns.

When searching for a solution, RIAS’s main aim was to find a way of targeting the right customers with the right product at the right time whilst retaining the best value customers. By applying a predictive model to analyse customer demographics and known preferences, RIAS was able to cut out the contacts least likely to respond as well as better matching product offers to customers. The profitability of acquisition campaigns has grown by 20% in some cases.

“When we started, we hoped to cut spending in some campaigns by 20%; to be able to achieve this and much more is extremely encouraging,” says Frank Abu, Modelling And Analytics Manager at RIAS. “ROI on the software investment was achieved in six months and we now have a multi-channel, segmented, contact strategy that is streamlined and more effective.”

Take a bow

The impressive results for the examples above are testament to the many years of development and learning that the suppliers and the client companies have put in to get this stage.

“There’s a real opportunity to understand customer behaviour using data analytics,” concludes 20:20’s Campbell. “Fully supported, you can create effective, micro-personalised CRM programmes, in real time across all channels.”