SUBSCRIBE

Winning the Campaign

Asking these 10 important questions before deployment can help your marketing campaign management project succeed

By Ruby Kennedy

Campaign management (CM) is a unique blend that combines technology with the art of marketing to help present highly tailored offers to prospects and customers. By offering the right message, to the right individual, at the right time, through the right channel, your company can meet its profitability goals as well as your customers’ product and service needs.

For some companies and divisions, CM projects are mission critical; indeed, marketing automation and CM are their only missions. But all companies that can better target and serve their customers will be more successful. A recent Deloitte & Touche study concluded that customer-centric companies are 60 percent more profitable than non-customer-centric companies, and that because of increased customer loyalty, the former are twice as likely to exceed their business goals.

However, CM projects are subject to the same pitfalls — such as project creep and changing specifications — that can plague other complex projects. For a successful implementation, simplicity is the hallmark of success: The simpler the project and the more quickly you complete it, the more likely the success.

The Basics

As with other projects, successful CM projects need high-level (CEO, CIO, or VP) sponsorship that can command the necessary corporate resources and keep the technical, sales, and fulfillment organizations focused. Furthermore, a successful CM implementation requires a project team that includes all the stakeholders: the systems designers and database managers supporting the technology infrastructure, the marketing managers and analysts who will design the campaigns, the end users who will implement and execute the campaigns, and the marketing or finance analysts who will evaluate or measure campaign success. It also requires a close working relationship between the vendors and the marketing organization, as well as a disciplined project management approach.

Limiting the scope of projects and focusing sharply on specific objectives also increase the chances of success. For example, data repositories and data marts for customer-centric marketing are smaller (typically with fewer data feeds and end users), more homogeneous, and faster and easier to implement than enterprise- wide data warehouses.

These management issues aside, you must also address several fundamental technical and marketing issues that fall into three categories: people, processes, and technology. You must understand all three of these areas and the interrelationships among them, which make up the underpinnings of any successful solution. This foundation will determine whether you will complete your project on time and on budget, as well as if the CM installation will become a sales and marketing tool that can take your organization to the next level.

Because a CM tool can be the critical focus in a marketing organization and because automating CM can greatly affect the marketing business processes, you must consider “human” issues such as software acceptance, training and knowledge transfer, and process re-engineering as well as the more obvious software functionality and integration ones.

My goal in this article is to help you address the technology side of the equation. I’ve boiled down the issues involved here into 10 technology questions you should ask your CM vendor. These 10 questions fall into three categories: installation, database integration, and CM.

Installation

Exactly how long will the project take?

As we all know, longer projects are more likely to fail. Furthermore, in the fast-paced world of online marketing, long projects are far more likely to solve yesterday’s problems, not today’s. Marketing organizations are also increasingly dynamic, adopting new best practices and leveraging new technologies to change the way they do business. Consequently, smaller, faster projects will help these organizations react more quickly, learn from previous experience, and use the “divide and conquer” approach to tackle key business problems. Trying to solve everyone’s problems in a single project greatly expands the project’s scope, and ultimately, the risk of failure.

Smaller projects also require less staff time and fewer resources. In addition, because effective CM can substantially improve campaign results, faster projects can produce an earlier return on investment (ROI).

To meet these goals, your CM software provider must present a definitive methodology and time table. It should also present a typical deployment plan, with a Gantt chart, that outlines project steps, responsibilities, dependencies, and timeframes.

Six months or more is the typical length of a data warehousing project, but six months is certainly too long for a CM project. Generally, such projects should take two to three months, assuming that customer or prospect data is available in a data warehouse, database, or flat files. However, if the CM software supports multiple data sources simultaneously (including flat files) without requiring a proprietary data model or data replication or transformation, the implementation can move along much faster. Ideally, you should be able to complete installation in a few weeks and stage the first campaign within one month.

How much will IT staff be involved?

Having a good understanding of IT staff commitment is critical. Software that minimizes the demands on IT, both at installation and as campaigns are implemented in dynamically changing environments, can be a tremendous help to already burdened staffs. To meet that goal, you should consider the resources required for initial installation as well as ongoing maintenance. Because most companies have dynamic IT environments, CM software that can easily adapt to changing data sources, data schemas, and formats provide the greatest flexibility and lower cost of ownership over time.

The issue involves more than time, however; it also involves expertise. For example, what skills are required? Does your staff have these skills? Do they have the time to take on another project? Basically, the more complex the integration of the CM and data mining systems, the greater the time, expertise, and risk involved. Vendors should clearly outline the skills and tasks involved to a level of detail that will help project planners make preliminary staff assignments.

What project management tools and resources will you need?

As with any large IT project, when installing a CM system you are buying not only software, but also buying into the vendor. Thus, you must examine the vendor’s methodology closely. Does it have a “best practices” deployment plan that begins with clearly stated goals and provides project management tools and checklists that help management and key staff members supervise the project? If so, the methodology should outline each step in the process and specify the sign- offs and approvals required. (See Figure 1.)

FIGURE 1 Deployment methodology.

You also need to understand the training and implementation support required before the project begins. What training will you offer, how many marketing managers and administrators will you involve, and what ongoing technical support services will be included? What are the key milestones and how will the project formally end? These answers will determine how easy and intuitive the CM tools will be for campaign designers and implementers.

Database Integration

What changes, if any, will your core database require?

Database integration and mapping are important not only in terms of time and skill sets, but also from a risk and efficiency standpoint. For example, you need to understand how the CM software will interface to all your in-house databases, and, if the CM tool requires database changes, what they are and how they will affect other operations. At the very least, the data tool should map to whatever data format you currently have in place.

But mapping is the most rudimentary level of integration. You also need to know whether the CM tool will require IT staff or marketers to extract, transform, and load data into its own data model or database. If so, you must consider both the initial conversion efforts and the ongoing batch feed process required to keep the CM data current. (See Figure 2.)

FIGURE 2 Are you ready to deploy?

Batch updates, however, introduce complexity, points of failure, and delays into accessing current data. In addition, the opportunity to achieve the ultimate goal of today’s marketers — realtime campaigns — is limited in a batch environment. Ideally, the CM tool will be able to import and export data to any data type, including flat files, in real time. This capability will give users the greatest flexibility across the broadest range of data types and sources.

For example, the ability to monitor responses directly from a Web database — instead of waiting for a batch process to load that information into the centralized marketing database — can make information available immediately to the CM system. This ability would let you send automatic responses, such as order confirmations or a cross-sell offer, to the customer in real time. In addition, the ability to map to flat files lets you use prospect data and appended demographics in campaigns without having to burden IT resources to load it into the database.

Will your database take a performance hit?

Some CM tools rely heavily on database resources, continuously writing to the database as they run campaigns (for example, to store information about which customers fall into which segments or cells as the campaign logic divides customers into different segments, test, and control groups). This approach obviously adds performance overhead; if the CM software relies on database server resources, you should consider purchasing performance enhancement tools such as disk caching as well as the related disk, memory, and networking requirements.

Other CM tools use their own database and data schema. This approach requires you to replicate your data into the CM application’s data mart; thus, creating the new data mart is usually the first step in implementation. This approach can provide the advantage of a dedicated CM database and some performance benefits, but you must schedule periodic batch updates to keep the marketing data fresh. If updates are frequent, the update process alone can significantly degrade database performance. In addition, a proprietary data mart may reduce flexibility in changing your underlying data structure, using external data, and in tying your CM system into other operational CRM systems. Your vendors should help you understand these issues so you can plan accordingly from technological, cost, and timeframe standpoints.

Can the CM tool work with multiple data sources simultaneously?

Some CM tools operate against only one data source at a time, such as its own proprietary data mart or a single database to which it has been mapped. Thus, any campaign that uses data from more than one source would require a merge or join among multiple databases outside the CM tool. In addition, each data source may be a different database or file, have different formats, run on a different platform, and differ in any number of ways. Either way, merging and purging the files will involve a tremendous — and often prohibitive — amount of staff time and computer resources.

THE BIG 10

The most important questions to ask your cm tool vendor 1.Exactly how long will the project take? 2.How much will IT staff be involved? 3.What project management tools and resources will you need? 4.What changes, if any, will your core databases require? 5.Will your databases take a performance hit? 6.Can the CM tool work with multiple data sources simultaneously? 7.Can the CM tool generate truly random samples? 8.What range of campaigns will you be able to run? 9.How scalable is the CM tool in all dimensions? 10.What are the experiences of other companies using the software?

CM tools that can access multiple databases (including flat files) simultaneously and without administrator intervention provide the greatest power and flexibility. The ability to use information from your invoicing system, telemarketing database, Web database, or inventory database can provide additional data to create more effective campaigns. For example, you could automatically suppress promotions to individuals whose account payments are more than 90 days past due or confirm that inventory is in stock before emailing an order confirmation. Particularly when predictive models, data mining, and intelligent segmentation are involved, CM systems that can draw data from multiple sources let marketers experiment with different data without committing it to a centralized marketing database. When the data has proven its value, they can add it to the data feed into the marketing database. For example, you may want to utilize data from a separate billing, inventory, or channel database (such as one on the Web). As long as a persistent unique identifier exists across the data sources, you can join the data across tables in disparate databases using the CM tool’s application server. Merged, denormalized data in a single database would not be required.

Acquisition campaigns are a good example of the power of accessing multiple data sources. Access to third-party databases, or more typically lists in flat files, is critical for executing such campaigns. Usually in these cases, a merge-purge process will eliminate duplicates and existing customers from the lists. If the CM management tool can only access one database at a time, you must not only merge these lists but also load them into a database, consuming precious IT resources and introducing additional delays.

Campaign Management

Can the CM tool generate truly random samples?

Surprisingly, some CM tools do not generate truly random samples for creating control or test groups. Rather, they sample by taking every nth record from the database. This approach can introduce bias based on the sort order of the records — for example, using a customer key might imply a time bias if the keys are sequential. Control groups created in this manner may not pass statistical tests for randomness, and therefore may not provide valid results for comparison purposes when evaluating a campaign. Most CM tools use this method because it’s easy, not because it’s the right thing to do.

Ideally, the CM tool should make it easy to create control sets and calculate sample size requirements for statistically valid measurements. I cannot overstate the importance of true random sampling — it is the only way to get the valid test results that become the basis for all subsequent campaign-design decisions. In addition, it is impossible to accurately measure campaign results if the random sample preceding the campaign is invalid. Therefore, your campaign designers must be comfortable with the level of random sampling offered by the CM tool. You should make sure that the tool’s random sampling capabilities include a random number generator to select records in a manner such that each record has an equal probability of being selected.

What range of campaigns will you be able to run?

Most marketers do not run the same campaign repeatedly through a single channel; they instead conduct integrated, multichannel campaigns. Some CM tools are oriented to one channel to the detriment of others — for example, they only manage email or Web campaigns — but even a pure-play Internet marketer may not want an Internet-only tool because it will limit future growth into other channels and collaborative marketing campaigns.

Your CM tool should support all the customer touch points you use, including the Web, email, direct mail, catalogs, fax servers, the direct sales force, and branch offices in every combination for traditional as well as permission-based campaigns. It also should be able to stage realtime, response-driven campaigns.

Furthermore, don’t let a CM tool’s ease-of-use limitations dictate how you design your marketing campaigns. A tool that is easy to use, permits easy repurposing of campaign components (such as queries, derived variables definitions, and portions of other campaigns), and supports a documented, logical campaign flow will make it easy to create and debug new, complex campaigns.

The software should accommodate simultaneous access to multiple data sources for acquisition campaigns, not only from the technical standpoint I discussed earlier, but also from a CM standpoint. Developing acquisition campaigns can be just as complex as cross-sell, upsell, retention, loyalty, and other marketing campaigns targeted at existing customers. In many ways, acquisition campaigns can be more complex because you probably have less information about your prospects than your customers.

For example, say a credit card issuer has three offers for which the annual fee, interest rate, and pre-approval ratings vary. In this case, the CM tools must be able to determine, based on available data, who is eligible for pre-approval and which offer to make to each targeted individual. In addition, the tools must be able to choose the appropriate channel by balancing factors such as cost, potential value, and prospect preferences. The software should also support a flexible approach across any conceivable marketing campaign. For instance, if you receive no response, a second wave might automatically target the same individuals, perhaps with a more attractive offer or by using a different channel.

Thoroughly examine the software’s optimization capabilities. Every CM vendor claims such capabilities, but few are delivering a robust set of tools. This process requires the close integration of CM software with database modeling programs. During the testing stage, campaign designers should be able to create a marketing model based on responses from the random sample. They can then use this model with the full “universe” to score the rest of the population. Ideally, the scoring should happen dynamically (on demand) so that the most recent database information is available — as opposed to batch scoring periodically and using the old, stored model scores.

Optimization technology can determine the best course of action for each individual based on his or her previous history and profile. The result of this segmentation is personalized, tiered offers at the one-on-one level. The system must be able to evaluate the costs for each offer to present the customer with the greatest perceived value while maximizing profitability. In addition, in the event that multiple offers apply to an individual, the software should have the “intelligence” to select the most profitable offer. In the end, effective optimization can make the difference between a standard campaign and a highly successful one.

How scalable is the CM tool in all dimensions?

Because marketers seldom have the luxury of working on only one thing at a time, scalability is a big issue. For example, with what sized database can the CM tool work efficiently? How many campaigns can you conduct at the same time? How complex can the campaigns be? As I’ve already touched on, the CM tool must support personalized optimization through multiwave campaigns and make different offers based on responses as part of longitudinal campaigns that play out over time. The software also should be flexible enough to mount inbound and outbound campaigns originating from any combination of channels.

These flexibility attributes are especially important today because of a trend toward quickly implemented, smaller, highly individualized campaigns. In this environment, marketers must consider how a CM tool helps them manage multiple marketing activities across campaigns. The CM tool should include or integrate with a Web portal across the enterprise so that all authorized interested personnel can efficiently monitor campaign progress and results.

What are the experiences of other companies using the software?

Talking to companies that have successfully installed and implemented the software is critical. You should be able to get a number of customer references, but don't rely on “canned” testimonials; talk to the customers yourself. Specifically, ask about the time to deploy, the system's flexibility, and their overall satisfaction with the vendor.

You may also want to ask the users themselves several questions, such as:

Rate This Article Rating: 1 (best) 2 3 4 5 (worst) Comments: Optional e-mail address: