Modelling should not be an elitist technique; virtually every organisation that uses data as a marketing tool will be able to benefit from an exploration of data modelling as a means of new customer acquisition, and as a way of identifying high value contacts within an existing customer base. However, accurate and up to date data must underpin any modelling or analytical activity of this nature. Without pristine data results will be worthless, because the constructed model is based on inaccurate data.

Also important is to set clear objectives when embarking on modelling activity. Most obvious is to determine the desired outcome; for example new prospect acquisition or a means of making marketing spend more efficient when communicating with existing customers. This clear focus can then better inform the model’s construction and refinement, with the objective is constant focus.

In the case of customer acquisition, building a model based on the organisation’s most profitable customers can ensure new prospects are targeted based on the potential value they can offer the business. Through this approach marketing activity can be created to acquire “more of the same”, perhaps using loss-leading opening offers or promotions, targeted at these new customers who offer positive returns in the long term.

As well as the traditional methodologies of "look-alike" modelling, there are alternate offerings such as those based on Standard Occupational Codes (SOC). This methodology, which is independent of Industrial Classification, seeks to define the relevance of products to job functions in the workplace, and ultimately can be used as a tool to identify a prospect in a business universe.

Often SOC modelling redresses assumptions made over time with the evolution of a customer base by defining the potential users of a product rather than simply relying on the existing patterns and falling foul of the self-fulfilling prophecy "we've always targeted Finance so we target Financial companies" scenario. It also works well in new markets without existing customers to make inferences from, and can be used in conjunction with "look-alike" modelling to create more complete prospecting lists.

Creating a usable and successful model is as much about the data and setting of clear objectives as it is the analytical expertise behind it, so b2b marketers should take full advantage of the skilled analysts available in the market to support the development and implementation of modelling as a powerful tool to drive their marketing efforts.

Why wouldn’t any business with many customers model its customers? Its like owning a car but not learning to drive. How can you make the most of anything if you don’t know much about it?

No longer does the scatter gun approach work with marketing. Customers are being treated to more in-depth, detailed and relevant messages coming from multiple channels. They expect relevance, and relevance is tricky to deliver until you understand the customer in terms of what drives their decision making process.

Understanding is the key, in which Modelling is ideally suited.
Modelling allows us to identify the good, the bad and the ugly of our customers and prospects.

It would be very easy to go straight onto the “lets append further information” bandwagon such as SIC code, number of employees, etc however this is jumping the gun. This gives information relating to the organisation, which is of some use. However it doesn’t help understand the driving factors.

Therefore the first consideration is to utilise the most important information you have on the customer – i.e. your communication and response histories. This is by far the most pertinent information as it is a true representation of the relationship between you and your customer. You just can’t buy this valued knowledge!

Modelling should be performed with both an analytical and business head. An Analyst can provide what he believes are some “very valuable findings” often only to discover they have little or no relevance to the business requirement. So careful management is vital in order to tie up the two disciplines.

In summary marketing strategy and analytical worlds collide, its important to ensure all parties understand what they are trying to achieve. Utilise initially your own important data then append additional variables if required.
Always remember to review what you are trying to achieve.. this is likely to be an ongoing and evolving task.

…..obviously all of this is pointless if your data is not clean and accurate, so step 1 should always be a data quality exercise.

Data modelling offers an opportunity for organisations to target B2B prospects based on recent transactional information. The benefits of this far supersede that of simply using job title selects in an attempt to target the right prospects.

Firmographic information often lacks the reliability found in transactional data, as the way in which companies record their information can be inconsistent. Modelling enables you to effectively reach those people responsible for making the final purchasing decision for your products.

In today’s turbulent market, it is vital to stay ahead of the pack. With budgets being cut and great importance being placed on ROI, modelling on transactional data identifies who is spending now which is crucial for effective targeting.

Data modelling doesn't need to be complicated. Obviously the more information you have access to, the more insight you can glean. Simple modelling techniques should be a pre-requisite for every company - regardless of size - in my opinion. Given that client data is the backbone of everything that you do, the more insight you can leverage from it (be it by demographic, postcode, purchase type/value or frequency, for example), the more data can drive and benefit your sales and marketing activities.

'Use what's available to you in the most appropriate ways possible' - that's my general rule of thumb. The insight that comes from even basic data mining (like the types alluded to above), will allow you to better target and tailor your offering. Nothing beats making the right offer to the right people at the right time, after all. By analysing and increasing your understanding of what drives clients to purchase and which channels they prefer you'll be able to increase both your brand loyalty and ROI. A little bit of data mining knowledge certainly can go a very long way.

Usually a brand's audience is made up of more than one type of consumer and understanding these different groups is key to creating an engaging experience that encourages interaction, purchase and ultimately advocacy. Treating everyone the same is inefficient and is likely to discourage some members of the customer and prospect pool.

Most businesses have a wealth of data locked away and much of this is private to the organisation. While customers may well also be the customer of a competitor, the interactions with an organisation are a unique asset that should be exploited for competitive advantage. Modelling provides a way to do this.

There are many different techniques available, from basic clustering which seeks to identify unique subsets of customers who behave differently, to predictive models that aim to identify those most likely to interested in a particular product or service.

You can model data specific to product or service areas, but solutions work best when you consider the individual audience member as the starting point - then it is possible to segment the customer and to ensure that you are offering the most relevant message when you communicate with them. Modelling and targeting based on models helps to significantly decrease the cost of communications as well as proving relevance which is likely to enhance the brand's image in the eyes of the customer.

There are challenges with modelling however, it is important to test out models and maintain them over time otherwise errors and unlikely connections may be created that don't fully map to reality - this can cause poor and costly decisions to be made. Business users are probably best using one of the packages that seeks to identify and qualify data before inclusion in a model as these steps are often involved and error prone with traditional statistical packages.

When a number of models have been created, significant advantages can be realised by utilising a contact optimization package that helps to create an optimal contact strategy plan including practical business scenarios such as stock levels, targets and channel capacities. Such packages ease the "model maze" suffered by some companies.

Using external 3rd party data such as SIC code and corporation size metrics in models helps significantly when targeting prospective new customers - by modelling the behaviours and preferences of existing customers against information that is available in advance or a relationship can help to shape messaging and reduce wastage when acquiring new customers.

Building models is one thing, but putting them to work is another. Organisations must be capable of changing the way in which marketing messages are targeted to take advantage of the new knowledge; in its simplest for this would just be reducing the volume of communications made by targeting only those most likely to respond. Advanced technique provides for a more engaging and relevant experience by allowing different messages, products, channels or tones of voice to be used when marketing.

Modelling provides a practical way to benefit for the unique assets of an organisation, so long as care is taken over the construction, implementation and maintenance of models they can provide significant improvements to the bottom line.

Building data models is a critical part of any successful relationship marketing programme irrespective of whether these models are created for customer or prospect programmes. Segmenting customers to understand current and likely future value allows businesses to plan the level of investment needed to retain precious customers.

Beyond retention, models can also give you a better understanding of what a prospect is likely to be worth, which in turn helps to budget accurately for the acquisition of them. Data modelling has the ability to improve ROI and reduce costs; every company that is serious about their marketing strategy should be considering using data modelling techniques, provided they fully understand what models they need and how they will use them to deliver against business objectives.

However I think businesses must also recognise that modelling is only one element of any analytical or insight strategy. Models can be used to good effect tactically around campaigns, but all too often models are badly built, used once and discarded – or even considered as a bolt on to the strategy and never really embraced and truly integrated with delivery.

Far greater leverage can be realised through planning and implementing a strategic insight programme, using a combination of segmentation and models. Ultimately everyone is looking for consistent and reliable results for the short, medium and long term, which can be better achieved through making the most of your data at the earliest possible stage. Embracing a true insight strategy will require significant change within the organisation itself, in terms of appreciating that customer needs are at the heart of everything.

Before embarking on developing models I seriously counsel any company to consider the merits of working with external expert suppliers - against the cost of investing in internal resources to build and refresh models. There is a cavernous gap between the ability to build a model, and to get true leverage from it in the real World and recruiting a Team of Analysts doesn’t necessarily guarantee success.

The choice of expert support is also important though and diligence must be carefully exercised as the cost of working with the wrong supplier will be potentially very high. The right specialists should of course be able to build models, but must also be able to demonstrate a clear track record of implementing them, connecting insight to delivery against the businesses commercial objectives.

Much of the efforts of marketing have been solely focused on B2C scenarios while relegating B2B to adapt, change and reorganize B2C techniques.
Nevertheless, the level of complexity of B2B call for unique approaches that fulfill the ultimate goal of a Single-Customer View (SCV) in the relationship management landscape.

Data modeling benefits come through 3 main areas: segmentation, attribution and messaging each one being more granular in detail that the previous. Although segmentation is a defacto in marketing, new technologies can work with great amounts of data applying complex algorithms to define different segments.

The key is to allow for a segment-specific mindset. In other words, appealing to both large and small organizations can be somewhat like walking on the balance beam and, when done correctly, segmentation allows for a careful roadmap to managing perception and ultimately catering both segments.

As this information is gathered, attribution will lead to the creation of customer profiles with specific needs, wants and constraints . By executing search and mining campaigns, new prospective leads that match current profiles will result.

At the most granular level modeling of data can enhance the messaging efforts to engage a client; the correct message needs to be send out and measured to continue refining the marketing proposition across different channels. Nevertheless, we can only think of data modeling as a tool to reach an end-goal. The final outcome will be entirely directed, and dependent, on the objective which should be formulated prior. This will, in turn, provide us with the clues to the data needed to be retrieved both in terms of the type and the timeframe.

Addressing the last question, the simple answer is that in spite of marketing being transformed by the incursion of online campaigns, data modeling can be ubiquitous to both traditional direct marketing models and new digital solutions to which best practices can be equally applied.

As we all know the world of the customer has evolved, some consumers even expect us to provide them with information that they actually want – the cheek of it, how dare they!

Modelling is essential if a brand is attempting to change customer or prospect behaviour through data driven communications. It’s an over-used and under-implemented idea, but you really can’t get away from ‘right people, right message, right time, right place = right result’ – and that rule applies as much in B2B marketing as anywhere. Perhaps even more so, because business people are busy people, and interrupting them with irrelevant marketing is going to turn them off your brand.

Modelling can be as complex or as simple as any brand requires, and shouldn’t be seen purely as a tool of the ‘big boys. Smaller brands are often pleasantly surprised at the amount of information they have about their customers dotted around the business, when they actually look for it – really, there’s no company for which data modelling isn’t appropriate.

Once this data has been collated, audited and managed, the resulting accurate information and insights can be fused with creative at the very heart of all communications to increase personal relevance, deepen relationships and increase active engagement.

One of the most important factors to consider when using data modelling is the brief. As with creative development – a shoddy brief will result in a shoddy output, and an excellent output that doesn’t meet the brief is just as disappointing. Spending a little extra time on the brief at the outset will result in a much better end-result.

It is also vital that common sense is applied to all outputs and that the initial modelling results are not read as ‘black and white’. For example on a recent B2B modelling brief and campaign for Bang & Olufsen, we refined our modelling conclusions through call centre activity. These conversations highlighted additional opportunities, which lead to an expanded 3 stage direct mail campaign targeting architects and surveyors – the right solution for the real brief.

Data modelling can also be seen as expensive - but we must all fight the short-termism created by the current financial situation. Yes modelling costs, yes the money could be used to communicate to more people – BUT the best way to achieve a higher ROI is to focus on quality and relevancy rather than quantity.

Sophisticated data modelling is fundamental to all companies in order to build up detailed knowledge on their customers and provide an understanding of what their customers like, how they interact, their buying history and product preferences.

All companies should invest in data modelling and data analysis software, as it offers real insight into their customers’ identity. It can provide them with a thorough analysis of their entire customer base, across all channels in real-time.

Marketers can then utilise this information to effectively drive campaigns and communication strategies. The results can also be used to fine tune data for future analysis.

While the core concepts of data modelling are now well established, new tools and techniques continue to evolve, so if any company undertakes modelling properly they should continually challenge the model assumptions and update their model accordingly, this is particularly relevant in the current market where consumer behaviour is changing rapidly.