Compiled data is critically important to B-to-B marketers for two reasons:
It tends to target relatively narrow audiences with limited volumes. Compared with response files, compiled data generally provides better coverage.
Each account has a high revenue opportunity, so it’s essential to gain access to every potential prospect.
For the broadest possible coverage, business marketers use compiled files generated from telephone records, credit data and other sources. These lists are sold direct or through brokers.
SCOPE AND INTENT OF THE STUDY
But thanks to the Internet, two important developments have revolutionized compiled B-to-B data. The first is that entirely new data sources have emerged based on information gathered using the Internet itself. Jigsaw and ZoomInfo are good examples. The second is easy online ordering, now offered by data compilers of all stripes. Marketers can search fields, generate counts and place orders via browser-based interfaces — and download the data instantly.
In light of these changes we decided to research the online sources of B-to-B data to assess their accuracy and completeness. Late last year 10 vendors agreed to answer a series of questions about their data and business practices. We express our appreciation to the participants.
Business marketers are interested in volume (“How many good contacts can I retrieve from this system?”), completeness (“Can I get every field I want?”) and accuracy (“Is the contact information correct?”).
Getting counts is a fairly straightforward process. For this study we identified 10 industries commonly of interest to business marketers and asked the vendors to tell us how many companies they had in each of the 10, as indicated by SIC.
Then we selected a well-known firm in each of the 10 industries, and asked the vendors to tell us how many contacts they had in those companies. We also asked whether they code firms using NAICS (the North American Industry Classification System). Finally, we requested that they report the number of complete contacts — “complete” meaning that the record included full name, postal address, title, telephone/fax numbers and e-mail address.
COUNTS AND RECORDS REPORTED
Counts are one thing, but assessing data accuracy can be difficult because the vendor’s record must be compared to some standard of correctness to determine the “truth” about the contacts on the vendor’s database.
OBSERVATIONS ABOUT THE DATA
To solve this problem we persuaded 10 businesspeople in a variety of industries to provide us with their accurate current contact information and allow us to publish their records as reported by the participating vendors. We’d like to express our gratitude to these individuals as well.
Fig. 1 shows the company counts in each of the 10 industries reported by the vendors in response to the question, State the number of U.S. firms you have on your file within each of these 10 SICs.
The counts for contacts at each of nine well-known companies and one university are given in Fig. 2. The question: Provide the total number of contacts you have at each firm (U.S. only), including headquarters and all branch locations.
Complete counts for each industry are presented in Fig. 3. We asked for: The number of complete contact records you have at each firm. “Complete” means full name, postal address, title, telephone and fax numbers, and e-mail address are included.
Fig. 4 is the contact record for one of our 10 intrepid businesspeople. Theresa Kushner was enthusiastic about participating in the study because she’s a database marketer herself. The other nine folks work in IT consulting, data communications, accounting, higher education, publishing, healthcare, manufacturing, law, and the optical components industry.
Going into this study we assumed that B-to-B data is a mature category, and that most vendors have fairly similar access to information about U.S. businesses. We also assumed that data accuracy would be a serious problem since we don’t know any business marketer who doesn’t complain endlessly about it.
We were surprised on both fronts. For one thing, we didn’t expect the wide variance among company counts and contact counts reported by various vendors. In SIC 32 (stone, clay and glass products), for example, the company counts ranged from 385 to 36,352, with all kinds of quantities in between. The least likely data element to be available was e-mail address, followed by fax number. This probably reflects the relative recency of these media as business communications tools.
For another, we were pleased by the level of data accuracy this research revealed. When vendors reported having an individual’s record, it was correct in the vast majority of cases.
The major problem was in coverage. The number of vendors reporting no record on each individual ranged from a minimum of three to a high of seven. Among the 10 individuals, the higher-ranking businesspeople (i.e., CEO, president) tended to enjoy better coverage by the data providers.
Continue on Page 2
Analysis of these results suggests that business marketers should exercise caution when ordering compiled data online.
Here are some guidelines:
Given the wide variances in data quantity and quality, it’s essential that you investigate thoroughly the data sources and maintenance practices of the vendors you’re considering.
Don’t take for granted that all subsidiaries of large compilers will have the same data. Salesgenie, OneSource and idEXEC are all units of InfoGroup, and Selectory and Zapdata are divisions of D&B. All reported dramatic differences.
When you do place an order, be very specific about industry selections. Find out if the vendor uses SIC or some kind of conversion algorithm. Not that any method is wrong, but you should know exactly what you’re getting.
Keep an eye out for vendor specialization by industry. NetProspex, for example, appears to have particular depth in the business services category, but doesn’t cover education at all.
Conduct a comparative test before you buy. One way is to run an appending trial, sending each potential vendor a list of 5,000 of your house names and asking them to add data fields. Be sure you include 25 or so records on which you know the “truth,” to assess the accuracy of what comes back. Another method: Order a sample of names and verify their accuracy by telephone.
RUTH P. STEVENS (firstname.lastname@example.org) consults on customer acquisition and retention, and teaches marketing to graduate students at Columbia Business School.
BERNICE GROSSMAN (email@example.com) is president of DMRS Group, a New York-based database marketing consultancy.
|32||Stone, clay and glass products||27,721||2,153||4,620||N/A||1,657||15,297||26,591||28,274||36,352||385|
|56||Apparel and accessory stores||36,325||1,552||6,141||194||5,419||185,704||220,209||193,149||228,619||3,200|
|28||Chemical and allied products||113,484||3,306||8,550||371||4,545||16,358||23,789||30,296||39,618||4,300|
|64||Insurance agents, brokers and services||43,679||3,221||11,794||980||38,303||231,220||245,486 /||217,389
|81||Legal services||71,908||1,769||25,639||5,290||38,669||196,000||343,409 /579,313||318,280||318,863||24,000|
|80||Health services||367,925||4,092||30,689||933||79,931||577,818||941,650 /1,672,599||872,000||875,370||38,500|
|82||Educational services||458,788||1,389||27,155||836||N/A — Does not cover||268,067||284,194||195,018||279,804||40,000 (Including educational institutions)|
|35||Machinery, except electrical||138,615||6,274||25,013||565||5,707||71,792||105,764||90,343||109,286||10,900 (For-profit educational firms only)|
|Do you code firms with NAICS?||No||Yes||Yes||No||Yes||Yes||Yes||No|
| 1NetProspex does not use SIC codes. Counts are from their equivalent categories.
2For certain industries, multiple businesses may exist at a single site. In those cases, Salesgenie reported two figures — the number of unique firmsper address and the total count of all firms at all addresses.
3Counts are approximate. ZoomInfo uses a keyword algorithm that maps closely to SIC codes, but offers much more flexibility than SIC codes alone. For example, you can find insurance companies (SIC 64), but also marine insurance, credit insurance, or property/casualty insurance.
|Morrison & Foerster||9||11||1,524||11||248||290||30||124||N/A||863|
|Hospital Corporation of America||6||336||1,817||1||0||2,758||759||4,206||33||11,255|
|Ohio State University||30||Not found in idEXEC database||6,966||0||N/A — Does not cover education||190||203||713||1||9,025|
|Level 3 Communications||175||73||713||13||110||136||83||393||21||422|
|Morrison & Foerster||5||7||1,524||11||243||0||21||124||N/A||848|
|Hospital Corporation of America||0||98||1,817||0||0||0||261||4,206||33||8,404|
|Ohio State University||20||N/A||6,966||0||N/A||0||168||713||1||1,309|
|Level 3 Communications||156||25||713||11||107||0||31||393||21||380|
| 1NetProspex does not provide fax numbers, so these figures are technically incomplete by our definition.
2OneSource did not provide counts on complete contacts.
|First name||Last name||Title||Company||Address1||City||State||ZIP||Office phone||Fax|
|Correct data||Theresa||Kushner||Director, Customer Intelligence||Cisco Systems||170 West Tasman Drive||San Jose||CAfirstname.lastname@example.org|
|Demandbase||Theresa||Kushner||Cisco Systems||170 West Tasman Drive||San Jose||CA||95134-1706|
|idEXEC||Theresa||Kushner||Cisco Systems||170 West Tasman Drive||San Jose||CA||95134-1706|
|Jigsaw||Theresa||Kushner||Director, Integrated Customer Intelligence||Cisco Systems||170 West Tasman Drive||San Jose||CAemail@example.com|
|NetProspex||Theresa||Kushner||Director of Customer and Marketing Intelligence||Cisco Systems||170 West Tasman Drive||San Jose||CAfirstname.lastname@example.org|
|ZoomInfo||Theresa||Kushner||Director of Customer Intelligence||Cisco Systems|