Context for Application Development
We’ve been elbow-deep building a data model for some time, and I’ve managed to commit a neophyte mistake in offshore management — to paraphrase Mr. Pink, “I’m acting like a first year f***ing manager.”
All too often, we, as managers, provide specifications to developers and expect them to build tools without any context. Context helps developers make better decisions because they can reference broader business rules to develop logic and write code. Contextual information is also critical against the background of offshore outsourcing where teams consist of people with varied backgrounds absenting the underlying cultural assumptions we often take for granted. Giving developers clear direction and the rationale behind the work — business objectives, underlying business rules, etc. makes software faster and better.
We’re working on data management module for a platform designed to help us manage call center campaigns more efficiently. Part of this process is describing a set of patterns for restricting bad data to sculpt a reasonably effective list — data suppression.
Interestingly, data suppression has two meanings. For government data (the original big data), data suppression refers to withholding information from public records to protect individual privacy. Census.gov has an entertaining definition: “Data suppression refers to the various methods or restrictions that are applied to limit the disclosure of information about individual respondents and to reduce the number of estimates with unacceptable levels of statistical reliability.”
But I digress.
Bad Data Begone
We use various methods and restrictions to limit the number of lead records that are self-evidently unsellable or statistically unlikely to result in a conversion. Data hygiene and discernment are critical to contact center operations because:
- Clean, fresh data results in more connected calls, a necessary condition for right-party contacts
- Obviously, unsellable entities retard floor momentum, thereby hurting agent morale.
Lists Are the Foundation for Good Campaigns
The direct marketer’s adage about campaigns, “there are three components to any campaign: the list, the offer, and the creative,” provides some context to our lead suppression discussion.
Marketers ascribe two-thirds of a campaign’s success to a good list. Good lists result from the portmanteau of smart data procurement and effective data management.
Smart list acquisition identifies characteristics that signify prospects with the means and motive for conversion then assembling enough data about those prospects to contact them. Broadly, there are two techniques for data assembly: compilation and response. Compiled lists are derived from a variety of sources like online data collection and data broker purchases. Response lists, by contrast, include opt-ins and/or people with the predisposition to accept your offer. Response lists exchange higher data costs for higher conversion rates.
Compiled lists’ less targeted data works for volume techniques like telemarketing. For most telemarketing campaigns, it’s not feasible to rely upon response lists exclusively because the volumes required to feed the dialer are too large for all but boutique programs. Often telemarketers associate new connects — newly formed or moved businesses — as a responsive list as those businesses are assumed to require extensive business services and will re-evaluate existing vendor relationships.
The offer, not the product or service, is what the recipient gets for responding. Price can be the offer. For example, inducements like “50% off” or “two for one” drive conversions. Infomercials create a sense of value and urgency by piling on items (“wait…there’s more”) and urgency (“act before midnight tonight”). Online services use free trials to great effect.
This is where the line between product development and sales approach breaks down — offer structures for broadcast direct response, online marketing, and telemarketing all integrate the offer into their product design.
Where the offer is often defined as part of the product/campaign design, it’s expression obtains in the script. A compelling script must present the offer clearly and effectively within 30 seconds of reaching the decision maker. As the call unfolds, the script deploys language to build perceived value and mitigate risk (“look at everything you get”, “you can cancel anytime”).
Working from a well-defined, compelling, properly-scripted offer, agent skills are a major component of the creative. Conversions arise from effective presentations, objection handling, and closing skills.
There are a variety of activities that make lists easier to manage and optimize: using response data insofar as it’s possible, working with reputable data brokers and frequent opt-out culling. The most effective techniques for lead management, in our experience, are detailed suppression tables that remove bad records before introducing them to the campaign database and outside validation to remove bad and non-compliant data. In this section, I’ll present the basic process for suppression and our dialing cycles without delving too deeply into the actual suppression logic.
To keep things straight, we use the nomenclature defined below. It is essential to maintain a strict discipline on how data is described among the development team, the operations team, and the ecosystem centers. This keeps all parties on the same page with data movement and management.
- Lead — individual record.
- Batch — leads categories by source.
- List — leads distributed to each center for 2 week dialing period.
- First Import — one-time event where new leads are added to the database.
- Distribution — share lead list with center or email application.
- Recall — return lead list or conversions for processing.<
- Standard Load — the lead count supplied by center calculated at 500 leads/agent/day
- Alpha List — (by center) the first export we provide the center to provide ample leads for dialing This Site. This is double the standard load.
- Omega List — (by center) list supplied bi-weekly and recalled after two weeks of dialing.
- Zulu List — alternating list supplied bi-weekly and recalled after two weeks of dialing.
- Conversion — the campaign’s key event: a trail, opt-in, request for information, set appointment, or a sale.
We run suppression at three points in our lead process. These work in conjunction with update protocols, which define how we manage records on recall, and distribution protocols, which define how we sort records for export. For reference, I’ve included definitions for the import and recall events.
Lead Suppression Protocols
The following protocols provide the high-level logic for data suppression. We use these when we introduce new or recalled data into our system. Recalled data has been active at a call center or in a mail campaign, and we run suppression before we allow this data to enter our system with updates to the record’s history.
Duplicate suppression is self-explanatory, whereas category and keyword suppression are the results of internal tables gleaned from general and campaign experience. We’ve found that small government agencies can resemble small businesses and specific SIC/NAICS code restriction combined with keyword restrictions handle most of the inconvertible leads. Our tables are also manually updated when we encounter import dispositions for data that cannot be sold. Finally, we use RealValidation to handle our last suppression protocols by removing bad or regulated data.