Abstract This paper explores the evolution of customer interaction management systems and their possible place in the support of customer relationships.

BACKGROUND

Since early 1999 there has been an explosion in the number of technology vendors touting ‘customer interaction management systems’. In many cases it is campaign management or call centre or web personalisation technology under another name. But there are a few who have taken a novel look at a growing business problem, coping with real time interactions across a wide range of customers points. For most organisations the idea that there should be consistent and/or relevant treatment of clients (customer and prospects) across the organisation’s touchpoints seems logical, but for most marketing managers it is just a dream. Yet again, technology is ahead of most organisations’ ability to deliver the other pillars of success, the people, process and technology changes. In some cases the technology being promoted is ‘slide ware’ with little real substance — buyer beware.

This paper explores some of the approaches being taken by the technology vendors to solve this growing business problem.

INTRODUCTION

In the last five years the introduction of high-end campaign management applications with scheduling functionality provided marketing teams with the ability significantly to increase the volume of communications (tenfold is not uncommon) while maintaining and in some cases reducing the size of the marketing team. Adding to this armoury the tighter integration of the novel real-time communication methods such as e-mail and short text messaging, (many with significantly reduced unit costs per communication) has resulted in a situation where the same marketing team could potentially deliver 100 times more communications. The issue now is that many of the other marketing processes cannot cope with the increasing power of the core marketing communication engine. Wider automation of the marketing function and the evolution of customer interaction management systems is therefore inevitable.

In an earlier paper the requirements for a marketing automation solution were examined,1 this paper identified the key components and the functions to be supported by such a solution. The functions to be supported are:

• managing client information
• managing communication decisioning: maintaining communication reference data; determining client communication eligibility; determining client communication priority; determining client communication delivery mechanism;determining client communication cell involvement
• managing communication execution: maintaining communication delivery reference data; transferring required data to communication delivery system; executing communication; maintaining contact history; automating and optimising communication execution
• managing communication channels
• monitoring communication performance
• managing communication content
• managing the marketing process: task management; workflow
• managing model development and deployment.

This paper expands on the requirements to manage communication channels.

The last few years have seen   an explosion in the communication methods available to the average marketing manager. With this has come a requirement to manage the resources utilisation better in these communication channels and to improve the integration between them. A comprehensive marketing automation solution should support these utilisation and integration requirements.

The primary focus of the integration is to ensure a consistent treatment of a client regardless of the organisation touch point.

Consistent treatment does not mean that the same data are available at all points of contact. Providing credit reference data to a teller operator in a bank may be valid but the bank may not want to make these data available to a client as part of a personalised view on a website. The key is to disseminate the corporate understanding of the client resulting from the analysis of the data in the client repository to the rest of the organisation in a form relevant to the recipient.

The technologies that have supported this component of the marketing automation solution have tended to be independent of the communication decision process, but there has been a move to tighter integration of these systems to make more effective use of the channels. A whole class of technology solutions is being developed to meet this business requirement generically called customer interaction systems.

The following section explores the requirements for a customer interaction management system (CIMS).

OVERVIEW

The customer interaction management system (CIMS) plays a central role in the execution of client dialogue. It receives information from the organisation’s touchpoints, processes these through policies or business rules that determine how to respond, and sends appropriate content back to the touchpoint for delivery to the client. The CIMS should support the following key processes:

• obtaining data from the touchpoint
• processing data through a communication decision engine
• communicating the outcome to the touchpoint system
• maintaining a record of communication
• monitoring communication effectiveness.

Meeting these requirements raises a number of technology issues as follows.

INTEGRATING CIMS WITH THE TOUCHPOINT

Most organisations have a large number of possible customer touchpoints. The CIMS would therefore have to interface all the possible touchpoint systems.

Where this is the case, the CIMS would be able consistently to manage the customer dialogue across the whole enterprise. This is the core objective of this type of system.

Before a customer interaction can be managed the touchpoint system has to collect and pass the necessary data to the CIMS.

The following technical solutions have been proposed.

Modify touchpoint solution to collect and transfer the required data

In this case the touchpoint system is modified so that it generates a message that is consistent with the application program interface (API) built into the CIMS.

This approach provides the implementer with the greatest flexibility, since it means the CIMS can work with any touchpoint system that can generate an appropriate transaction. It also means, however, that a significant amount of work is required by the customer organisation to add each new touchpoint system to the overall solution.

CIMS manages the communication with the touchpoint

In this case the CIMS provides an application that itself manages communications with the touchpoint. This approach moves the development effort to the CIMS vendor but does facilitate integration with any of the common touchpoint technologies.

For example, one could provide HTML snippets that can be embedded in a Web page; these call the customer interaction management system when the page is requested, transfer information such as the customer ID and context, and display as part of the Web page whatever message the CIMS determines is appropriate. These applications are called ‘adapters’ or ‘connectors’.

Middleware technology translates data between the touchpoint system and the CIMS APIs

In this case a middleware application would be used to translate data between the different system APIs, without building a point-to-point connection between each touchpoint and the CIMS. These technologies do exist, although this strategy has not been adapted by any of the main interaction management vendors. This generic approach may not be attractive to the vendors who want to own this key part of the process. It does, however, allow the customer organisation to deliver a flexible architecture that allows each of the components to be changed with ease.

Monitor data in non-touchpoint system

Some CIMSs avoid any touchpoint integration at all by scanning for   significant transactions outside the context of a particular touchpoint process.

For example, a system might scan a stream of bank deposits, regardless of which touchpoint hosted them, and select those over $10,000 for closer examination. This could be done by reading inputs to the deposit system, which might actually be linked to a number of different touchpoint systems.

In many cases some degree of transaction filtering is deployed into the touchpoint integration mechanism, so the CIMS is only called when a decision is truly required. This could be accomplished by building screening rules into either the touchpoint logic that calls the CIMS’s   API, or the CIMS application embedded in the touchpoint.

TYPE OF DATA PROCESSED BY CIMS

The type of data passed from the customer touchpoint will vary depending on the nature of the customer interaction. As a minimum the following types of data are implicated:

• client reference data, eg client ID (unique customer or prospect reference)
• channel reference data, eg channel ID (unique reference for the touchpoint)
• interaction transaction

In many cases the touchpoint system will collect significant amounts of data on the interaction including contextual information such as the path the client took to get to the present location, and specific information such as replies to questions.

If integration is achieved using a CIMS API, then the design of the API will determine how many data can be captured. Systems that integrate via connectors built for specific touchpoints are governed by whatever types of data the connector has been designed to process. The maximum data available is usually determined by the touchpoint system’s API, but fewer data may be transferred if the connector accepts only a subset of what the API presents.

A business rules engine is used in the CIMS to determine what the most appropriate response is to the current dialogue request. In its simplest form this could be a lookup of the client ID in a   list of eligible campaigns or communications. This campaign list could be held in the CIMS (most likely) or locally in the touchpoint sytem. The latter results in control issues when campaign involvement is changed in the CIMS.

The touchpoint system must continue to operate even if the CIMS fails to respond. Such a failure could occur for any number of reasons. If the CIMS is functional it should first test other content or campaigns to see if they are available; if not, it should deliver a default content or campaign.

DATA CREATED AS PART OF THE CUSTOMER INTERACTION

The CIMS accepts information from the touchpoint and generates additional information about its own activities, such as movement of a client through a campaign sequence or which messages a client has been sent. Persistent storage is important in campaigns that walk a client through a sequence of messages or that use changes in a customer’s behaviour or status to trigger interactions.

This has implications for:

• the types of data stored
• user control over data
• storage format
• accessibility
• data transfer to other

all in real time. Accessing these data sources represents a significant technical problem. This is further compounded by the potential need to execute complex decisioning processes including channel optimisation.

OVERVIEW OF CIMS VENDORS

The following is a sample list of vendors claiming to have technology that supports customer interaction management:

Allink Agent
Black Pearl
Data Distilleries
Epiphany
Intrinsic
Manna
MarketSwitch
NetPerceptions
Prime Response
Quadstone
Recognition systems
Revenio
Verbind
Xchange
Yellow Brick

Even though each vendor/product offers a different combination of functions, the systems appear to fall into two clusters. Products in the first group (Yellow   Brick, Verbind, Revenio) offer complex, multi-step marketing campaigns that are delivered through one or more touchpoints, but driven by the CIMS itself. The second group (Epiphany, Black Pearl, Manna, Data Distilleries, Intrinsic) delivers recommendations in response to touchpoint requests. In these, the touchpoint itself is the main driver. Note that all the products in this group provide integrated modelling, while none in the first group do.

The one product that falls into neither group, Allink Agent, is indeed an oddity, designed less for real-time interactions than to identify marketing opportunities and react through outbound messages.

Products in the recommendation group may find themselves competing less with the other CIMSs than with recommendation engines like NetPerceptions and modelling products like MarketSwitch and Quadstone. This assumes the modelling vendors deliver real-time scoring solutions that are easy to deploy.

Similarly, products in the multi-step campaign group may eventually find their primary competitors are conventional campaign managers and marketing automation systems, particularly as these re-engineer to integrate with multiple touchpoints in real time. However, the division between one-time and multi-step CIMSs itself may not last, if vendors in each group add the key capabilities provided by the other.

An analysis of the key vendors of customer interaction management technology clearly shows that there is:

• no best practice approach: no two vendors have the same strength and weaknesses. This variation in approach reflects the newness of this technology category and the lack of an accepted best practice approach in the industry
• no comprehensive solution: no single vendor has addressed all the key business requirements. This reflects the fact that few vendors have given this area a high priority. It is also technically a complex problem, particularly when it comes to handling real-time decision making. The scope of the current solutions tends to reflect the background of the vendor company or the design team
• mainly small specialist vendors: most of the products in the market are provided by small (early adopter) organisations, with many being less than two years old. The fragmentation of the market among multiple vendors suggests that any significant market expansion will be accompanied by consolidation, or more likely, acquisition of vendors by larger CRM firms
• immature market: no vendors appear to have more than a dozen sites up   and running. This would indicate that the market is still immature and in   the early adopter stage.

CONCLUSION

The desire by organisations to manage client interaction across all touchpoints will drive technology vendors to develop systems to meet this business requirement. There are significant technical issues surrounding access to data from multiple sources in real time and the complex decision process and these problems will slow the evolution of these systems, but they can be solved. The bigger issue is that few organisations have the ability to deliver the other pillars of success, ie the people, process and technology changes.

Abstract This paper identifies the system requirements and marketing process impacts   of trigger or event-based marketing in promoting products and services in a financial services organisation.

INTRODUCTION

A framework for classifying customer communications was described in a previous paper.1 This framework described five basic types of communication that need to be supported by a marketing database solution. These were:

• strategic communications
• tactical communications
• customer life cycle events
• product life cycle events
• customer trigger

In the case of strategic communications the marketing activities are part of an ongoing programme aimed at meeting a specific strategic objective such as:

• development of the brand
• growth of a new customer segment
• launch of a new product or service
• managing customer profitability
• migrating customers to the e-channel.

In the case of tactical communications the marketing activities are used to address a specific business issue, which is transient or short term in nature. Typical tactical communications are associated with:

• new branch opening or closure
• exploiting short-term competitive advantage
• spoiling competitive activities
• meeting business shortfalls
• changes in legal or government regulation
• changes in market environment.

The marketing activities in the case of customer life cycle events are focused around a customer or prospect life cycle event. Typical customer life cycle events are:

• birthdays
• changes in family status
• changes in employment status or type
• changes in wealth
• inheritance

For product life cycle events the marketing activities are aimed at product life cycle events for current or historical customers. These events are normally associated with key dates or product-based transactions. Typical product life cycle events include:

• account opening
• account closure
• anniversary dates
• maturity dates
• renewal dates
• acquisition of a particular product combination.

In the case of customer trigger events   the marketing activities are aimed at customer trigger events for current customers. These trigger events are generated as a result of an inward-bound communication from the customer (or third party, eg a solicitor) or a change in customer–bank behaviour. Typical customer trigger events would include:

• change of core customer data e.g. address changes
• customer complaint
• product or service information request, eg deed request, tax status change
• account activity, eg abnormal transaction.

This paper focuses on the last three types of communication. Collectively called trigger or event-based marketing their importance is growing as organisations recognise the value of making communications timely and relevant.

DEFINITION

The terms ‘event’ and ‘trigger’ have been used interchangeably. For the purpose of this paper the following definitions are used:

• a trigger is a change in data that could provide input into an event definition, eg change of address
• an event is defined by one or more triggers, eg change of address combined with change of marital status implying a customer life cycle event: marriage.

A trigger or an event may be used to drive a marketing communication where they provide evidence of a financial need that can be served by an organisation.

IMPORTANCE OF CUSTOMERCENTRICITY

Product life cycle events have played an important role in the past in driving marketing communications as they have been easy to execute, even when the organisation’s system(s) are productcentric. Customer life cycle and customer trigger events on the other hand, have required the organisation to   be able to execute customercentric communications. Generally speaking this is now becoming the norm, certainly in most financial service organisations and is facilitating customer life cycle and trigger event marketing activities.

BENEFITS OF EVENT-BASED MARKETING COMMUNICATIONS

If effectively executed these event-based marketing communication activities significantly out-perform traditional targeted communications covering a similar subject matter. The primary reason is that these communications are more likely to be timely and relevant. The trigger or event provides both time input and context input for the communication.

EVENT-BASED MARKETING COMPONENTS

The following section explores the key components required to support event-based marketing. These are:

• event analysis
• event detection
• event-based campaign management
• event opportunity

The following section looks as these components in more detail.

Event analysis

The identification of triggers and events was based on business rules with little or no statistical analysis going into identifying potential triggers or events. A number of organisations have, however, started using advanced statistical techniques to identify potential triggers on the marketing database. A range of statistical techniques have been used either to validate that a business rule-based trigger was predictive or to identify previously unknown triggers or groups of triggers (events).

The following statistical techniques should be supported:

• regression analysis
• tree analysis
• cluster analysis
• associative analysis
• neural networks
• pattern searching
• text

Text mining has been used to search the content of e-mails being sent to a service centre to detect key words or combinations of words, which act as triggers for follow-up marketing communications.

A solution that is going to be used to support event-based marketing should be able to support the statistical validation of a business rule-based trigger and/or provide a statistically robust process for the identification of triggers.

The solution should include a set of business processes that structure the event analysis process for both business rule-based and statistically-derived triggers.

Event detection

Having identified that a particular trigger or set of triggers defines an event, the organisation has to put in place the necessary procedures to detect customers who have exhibited a particular trigger or group of triggers. This requires the application of rules against the customer base on a regular basis or in real time.

The following types of data may be implicated:

• customer profile
• account
• account history
• customer account involvement
• account transactions
• operational contact

The detection process can be problematic. The following are some examples of the issues that may have to be addressed as part of any solution.

Change history

One of the difficulties with some trigger types is that they require the detection process to look for changes in a particular data item. This is difficult, as many marketing databases do not hold history of changes at a data item level, eg change in marital status. In such cases processes will need to be created that detect these changes or maintain a history on key data items.

Query performance

Looking for a particular pattern of transactions in a large marketing database, with 100m transaction records will have implications for performance if not correctly architected. The solution must be designed with these performance requirements in mind. This may be achieved using ‘raw horse power’ and/or good database design.

Volume of trigger rules

Over time many businesses develop large libraries of trigger rules that need to be applied to the database. This will have implications for system performance and communication   prioritisation.   The solution should be able to apply a large number of rules on a daily basis and facilitate the maintenance of the trigger rules. The issue of prioritisation is covered later in the paper.

Model scores

Some organisations have found that changes in model scores provide valuable triggers for events. This means that large numbers of behavioural models (statistical procedures) may need to be run against the marketing database on a regular basis. This may have implications for system performance and the maintenance of model score history. The solution should support the use of models to drive trigger-based campaigns.

The solution should facilitate the estimation of trigger volumes so that a commercial assessment (business case) can be made before event-based campaigns can be established.

Event-based campaign management Once a trigger or set of triggers has been identified the next step is to use these triggers to drive the execution of event-based campaigns. A single trigger could be used to drive a number of campaigns. The solution should support the following campaign management processes:

• maintenance of campaign reference data
• identification of target audience for campaigns
• prioritisation of campaigns
• application of global contact rules
• creation of communication cells within campaign
• definition of data requirements for communication
• execution of communication by relevant channel
• scheduling of campaign repetition
• campaign performance

The development of a large number of event-based campaigns often results in the need to prioritise the communications that should be sent to a customer or prospect. This prioritisation process needs to take into account other communications that a customer is due to receive in the defined time window. If this is to be done the business needs to establish a consistent metric (eg target ROI or propensity to purchase) or set of business rules that the system can apply either at the campaign level or customer level.

It should be possible to integrate event-based campaigns with other campaign types within the same campaign management application. This will facilitate:

• planning and management of all campaigns
• prioritisation across different campaign types
• consistency in measurement processes
• effective use of marketing and channel resources.

Event-based campaigns do drive some incremental requirements that the solution should support. These include the ability to:

• do test counts
• project future volumes for event-based campaigns (on a monthly, weekly or daily basis)
• prioritise within campaign type and then across campaign types
• apply global contact rules (these are rules that are used to frame the prioritisation process)
• execute small-volume campaigns (often requiring a higher degree of automation and tighter integration with channels)
• support more complex reporting requirements as the campaign may run for a longer period of time
• support more complex personalisation of the communication (the trigger provides an excellent context for the communication).

Event opportunity measurement The use of global contact rules and campaign prioritisation results in a number of events not being progressed as opportunities. In order that the impact of these two processes can be measured and refined over time, there is normally a requirement to monitor the number of event-based opportunities that have not been progressed. The solution should:

• record all event-based opportunities detected
• flag those opportunities that have not been progressed and why
• allow the potential business value of the non-progressed opportunities to be determined
• simulate the impact of changes in: global contact rules; campaign priorities; customer communication priorities.

The solution should facilitate the recording of the lost opportunities and allow them to be progressed at a later stage where resource and business priorities allow.

IMPACT OF EVENT-BASED MARKETING ON THE CAMPAIGN MANAGEMENT PROCESSES

The adoption of event-based marketing affects both the technology and marketing business processes. The following section explores the impact that this type of approach has on the marketing processes.

The following is a typical process flow for marketing campaigns and the impact that event-based marketing has on the processes.

Agree marketing plan

The lack of clear understanding of the potential numbers of event-based communications that will be initiated in the coming year makes planning very difficult. This means that assumptions need to be made on likely volumes and response rates for a type of marketing activity that is primarily driven by the customer and not the business.

Produce campaign plan

The longitudinal nature of event-based marketing campaigns means care has to be taken to ensure that they do not

conflict with other campaigns that are going to run in the year. Campaign priorities also need to be established at this stage to ensure that any likely conflicts can be taken into account during the planning stage. Some organisations also agree global contact rules at this stage in the process.

Produce campaign business case

An organisation’s lack of ability to predict accurately the volumes and performance of event-based marketing campaigns means that they often make the business case using a range of values. These values are then used to set thresholds for campaign performance. If a campaign performance falls below these thresholds then the campaign will be terminated.

Produce campaign brief

The timely nature of event-based communications means that service level agreements (SLA) with both internal and external suppliers are often much tighter than for traditional campaigns. In order that these SLAs can be met, business processes will need to be refined.

Manage Response from Supplier

 Separate contract frameworks are often used for the support of event-based marketing activities allowing the procurement process to be streamlined.

Execute campaign

The campaigns tend to be executed on a daily basis. This means that there is often a higher degree of automation and tighter integration with the communication delivery channels. The sign-off processes are also streamlined to allow things like copy and collateral to be revised in a more timely manner. Resource planning also has to be tightened to ensure sufficient resource is available to exploit the opportunities.

Monitor campaign

It is essential with event-based marketing campaigns that tighter monitoring processes are put in place to ensure that marketing can rapidly respond to changes in campaign performance. This is more critical because of the automated nature of the processes.

Analyse campaign performance

Post-campaign analysis often looks at additional marketing metrics surrounding lost opportunities, service levels and campaign performance. This is often complicated by the use of refined offers as the event-based campaigns are being executed.

CONCLUSION

Event-based marketing has an important role to play in driving customer communication in a modern financial services organisation. The timely and relevant nature of these communications means that performance is normally better than for traditionally targeted marketing communications. Delivering event-based marketing has an impact on the technology and processes supporting marketing and other functions. These need to be addressed if the final solution is to deliver business benefits.

Abstract This paper describes the Auto-ID technology and its key components. It explores some of the issues associated with its widespread use. Finally it explores its potential business applications in marketing.

INTRODUCTION

The Auto-ID Center at Massachusetts Institute of Technology (MIT)1 announced the launch of version 1.0 of the EPCglobal Network in September 2003. This was a key milestone in the launch of a global set of standards and technologies that allow individual items to be tagged with microchips or radio frequency identification (RFID) tags.

These tags carry the electronic product code (EPC), which allows these objects to be uniquely identified and, through wireless technology, detailed information to be maintained on the object. Thus, the products on the shelf can not only talk to you, they have a distributed memory.

This paper describes the Auto-ID technology and its key components and explores some of the issues associated with its widespread use. Finally it discusses the potential business applications in marketing.

AUTO-ID TECHNOLOGY

The primary focus of the Auto-ID technology is to embed the EPC into product items; this results in objects that are intelligent and can communicate.

There are several components that make up Auto-ID technology.2 These include:

• eTag, an electronic tag
• EPC, a unique identifier
• object name service (ONS)
• Savant systems
• physical markup language (PML)
• business applications

The EPC

The EPC is a corner stone of Auto-ID technology. It is a string of numbers that provides a unique identification; for instance, instead of referring to a class of products — as universal product codes (UPCs) do — the EPC refers to a specific instance of a product.

eTag

To facilitate Auto-ID, the EPC is embedded in a memory chip contained within a smart tag (eTag) on individual products. The chip in turn is connected to an antenna. This allows the eTag to be scanned by a radio frequency reader, which transmits the embedded identity code of the product to a network, where the detailed information on the product is kept.

This detailed information can then be communicated from the network to provide whatever information is necessary about that product.

RFID is the basis for current Auto-ID technology. It is important to note that the baseline functionality of these eTags provides read-only access to the EPC. No detailed information need be kept on the eTag.

The current technology and standards do not preclude other tags with read–write functionality or even more advanced capabilities. However, as additional functions and capabilities increase, so will the cost of eTags. At the present time, read–write tags tend to be slower and have a shorter range than   their read-only counterparts.

The implementation of EPC does not depend on RFID technology; any way of quickly and easily reading a unique ID from a product will work. RFID is the most common option at the moment, but other technologies are being tested.

The ONS

The next link in the Auto-ID chain is the ONS. The ONS tells computer systems where to find information about any object that carries an EPC.

ONS is based in part on the internet’s existing Domain Name System (DNS), which routes information to appropriate network interfaces. The ONS will likely be many times larger than the DNS, serving as a lightning fast ‘post office’ that locates data for the trillions   of objects that will eventually carry an EPC.

Savant systems

Savant is a software technology that acts as the ‘central nervous system’ of the EPCglobal Network. Savant manages and moves information in a way that does   not overload existing corporate and public networks.

PML

PML is a new standard language for describing physical objects, in the same way that hypertext markup language (HTML) is the common language on which most internet web pages are based. The PML describes the physical characteristics of an object and almost anything can be contained within the description, such as weight or calorific content, repair instructions and audit trails. PML will allow manufacturers and retailers to specify and customise the information tracked on products. (It is also technically possible for the consumer to start to collect information on the objects that they own.)

There will not be a vast repository of PML descriptions. Ultimate implementation of the PML descriptions will result in highly distributed data.

Manufacturers, retailers and consumers will all have unique views to data.

Business applications

Potential application for Auto-ID in business are numerous. They include:

• manufacturing process control
• inventory management
• supply chain optimisation
• regulatory compliance
• recall management and

In all these areas, Auto-ID offers the potential for significant savings, as well as new sources of incremental revenue.

New services will start to emerge as objects start to become smart and interactive. As the technology becomes pervasive, benefits will extend

throughout the entire value chain and for the consumer.

Auto-ID technology has the capability to redefine the global marketplace by embedding intelligence, identity and internet connectivity into everyday objects. The EPC unites elements of the entire supply chain, making it an interactive, dynamic cycle from raw material and distribution to

point-of-purchase and recycling, and back to raw material. Products equipped with smart tags will interact with manufacturers, their trading partners and each other to form an optimally efficient cycle of direct, real-time supply and demand.

APPLICATIONS IN MARKETING

The primary focus of Auto-ID applications has been the supply chain, where   it   is   believed   the   highest benefits will come. But Auto-ID and similar technologies have a number of unique   features   that   could   provide value for marketing. These include the ability to:

• uniquely identify an object
• integrate data from a wide range of sources
• read the EPC wirelessly
• provide communication during the product purchase decision
• access the consumer after a purchase is made.

These features allow the marketer to explore a number of new activities. Many of these can be grouped under the concept of a ‘personal shopping assistant’. Examples of these applications in the store and in the home are discussed below.

In the store

Select an item and view product attributes such as:

• where it was made
• how long it has been in the store
• its expiry date
• what the calorific value is
• what it

Select an item and

• remotely look up the contents of the home larder and confirm if the item is required, alerting the customer where appropriate
• alert customer about products to which they may be allergic, eg contains peanuts
• alert customer that they have a discount voucher for an

Select an item and view usage information, such as:

• a recipe for an item
• product survey information
• product instructions
• comparison of prices at other

Prompt customer with location of item:

• pre-defined shopping lists can warn customer when they are in the proximity of a required item
• a selected recipe can warn the customer when they are in proximity of the required

Select an item and receive a promotional offer:

• provide the customer with a promotional offer at the purchase decision point based on the current contents of their basket
• provide the customer with a promotional offer at the purchase decision point based on previous purchase

In the home

Automated shopping lists can:

• use information about items in household and consumption patterns to automatically create a shopping list for a particular store
• use information about pricing to optimise shopping based on price comparisons.

Select an item and:

• warn if an item has gone past its expiry date
• alert customer to a product to which a member of their household may be allergic, eg contains

This is just a short list of potential marketing applications; the key point is that this technology will allow us to integrate data from a wide range of sources wirelessly.

ISSUES WITH AUTO-ID

As with any revolutionary technology, there will be challenges to overcome in Auto-ID implementation. Some challenges are technological in nature, some economic, and some societal. Issues include:

• privacy
• accuracy
• interference
• performance
• frequency availability
• security
• data

These are discussed in more detail below.

Privacy

Perhaps the most controversial issue is that of privacy. The ability to track an item after it has been purchased raises a number concerns for consumers. Although there are limits to the current technology, in the future it may be possible to trace clothing stolen from a store for example. If this then gets sold on to an individual who   visits a location with a radio receiver, the system could then ‘check’ to see if it had been paid for.

As consumers see value in the technology, and if these genuine privacy concerns are addressed by the industry, acceptance is likely to increase.

There are also some legitimate competitive issues that come under the heading of privacy. For instance, since the EPCs will be global and unique, it may be possible to determine specific product   information   from   the   EPC given enough data. Imagine gaining knowledge of your competitors’ shelf assortment and inventory levels by walking   through   a   store   accompanied by   a   hand-held reader.

Accuracy

Readers cannot be guaranteed to be able to communicate with all tags in a volume all of the time. Environmental issues, the make-up of the products being tagged and the volume of tags to be read all impact on read accuracies. The degree of concern is proportional to how much an enterprise relies on absolute data. RFID offers many advantages over manual or semi-automated data collection processes. Any shortcomings in accuracy can be mitigated through the use of redundant readers, information auditing and process redesign.

Interference

As readers proliferate, more occurrences of interference will be seen. Depending on the frequencies and powers used, devices such as mobile phones, wireless handsets and industrial equipment may be affected. As such a widespread penetration of radio frequency (RF) technology has not been undertaken before, it is difficult to state categorically what will be impacted.

The perceived health risks of this many RFs may also be a concern. While there is no evidence that there are any negative effects at the power and frequency levels associated with RFID, it has not yet been rolled out on such a large-scale. More research and monitoring will need to be conducted to address the public’s concerns in this matter.

Performance

Smart objects could generate tremendous amounts of data. This much data will not be accessible if stored in a massive central repository, so some distribution of data will be necessary. This raises a number of performance issues.

Auto-ID is based on RFID technology. Anyone who has used a mobile phone will be aware of the issues associated with access to a network. In order to work therefore, the data associated with EPCs will need to be available on demand, anywhere.

Frequency availability

Since RFID currently uses sections of the unlicensed RF spectrum, the available parts of the spectrum that are usable for RFID are an issue. Although there are some frequencies that are common, there is no universal standard.

13.56 MHz and 2.45 GHz are both worldwide standard industrial, scientific and medical (ISM) frequencies. These are available in most parts of the world, albeit at slightly different regulations.

More useful in terms of read range and speeds are tags operating at roughly 915 MHz, or ultra-high frequency (UHF). The UHF spectrum around 900 MHz is not universally available at the same frequency and power levels worldwide, however.

This issue will be addressed through two potential methods. The first alternative is multi-frequency readers — overall RF system design (integration of antenna, readers and tags) is the most difficult part of the problem here. The second is to select a common frequency. Obviously, since this involves millions of stakeholders, the lead-time on this will be considerable. This does not, however, deal with the fact that not all frequencies work well for every application (although some work well across virtually all applications).

Security

Security will be paramount and may be viewed at a number of levels, including:

• read security (or being able to read the tag)
• security of the data
• other security

For users of the technology to feel comfortable, there will need to be assurances that no one will be able to ‘hack’ into a smart object. As long as tags are read-only and are difficult to counterfeit, then security will be high. Users of Auto-ID technology will also need to rely on the security of Auto-ID data on the network.

Data ownership

Related to security, data ownership is an issue. Who owns the massive amounts of event information associated with an object?

It is clear that the manufacturer owns the design specifications and other PML-type data for a given product. It is clear who owns captured data — the owner of the reader that reads the tag. It   is less clear, however, how information will be shared.

Many parties will be privy to, and will update, the data for an object as it passes though a supply chain. Will those collecting the data ever want to share data? Does an end-user (consumer) ultimately own a product and its data   and, if so, how does use of that data for process improvement or data mining impact privacy?

Lastly, although killing a tag when purchased has been discussed as an option, this method eliminates future recycling benefits. It also introduces the potential for tags to be killed maliciously or by accident, before they should be.

CONCLUSIONS

The ability to uniquely identify an item through the use of an EPC is a natural extension of the UPC. Allowing this   EPC to be wirelessly read and integrated with detailed data across a global network is a major leap in functionality, which will provide manufacturers, retailers and consumers with significant benefits.

The development of Auto-ID technology is evolving, but widespread use is unlikely for many years. The early adopters are likely to be industries where the value of a unit is high and tracking individual items is important. Prime targets include pharmaceuticals and the automotive sectors.

To date, the focus has been on improving supply chain management. Little attempt has been made to explore the potential marketing applications of this technology. The author believes that it has a number of unique features that will prove valuable to marketers when developing point of sale communications. It also offers the opportunity to extend   the marketing communication process into the home environment and access the full product lifecycle.