A Rare Glimpse Inside the List Industry
[Editor’s note: Kirk Nagle, president of marketing list-brokerage firm Direct Response Associates in Allentown, PA, responded with a long, enlightening comment to the article I posted last week where my alter ego, Shuvitt Inyurasss, tried in vain to negotiate a deal for a list of proctologists.
Though the article was a stab at humor, it was also meant to educate readers about the potential pitfalls of buying data and how to spot signs that a data vendor is shady. Two of the issues I brought up were pricing and whether or not the email names were being sold for perpetual use.
The firm I was dealing with last week was selling names for $500 per thousand. I thought that figure was unusually high. The company was also selling the names for perpetual use, a practice that squares just fine with postal direct mail but is considered antithetical to permission-based email marketing.
What follows is long. A lot of it refers to the postal-list industry. But I believe it will be worth your time.]
Subject: $500 per thousand
I noticed a slight miss-use of English, [in the exchange published last week] which often indicates an overseas operation, but these people might be as legit as any list source and most lists sold here in the USA are terrible to begin with.
We do rent/sell business emails for endless-use (nobody cares the actual rental is for 1-year on all "unlimited-use" postal and email lists because they deteriorate so quickly). The largest business list compiler in the nation charges more than $500 per 1,000. Their quote of $500 per 1,000 suggests they [the company Shuvitt Inyurass negotiated with last week] might be resellers from that source. We would verify that by requesting counts from this source as well as the known compiler to see if they match. That’s one way we prove to our clients we sell real lists directly from the leading compilers.
Another topic was the lack of details on list sources. No email or postal list source EVER gives details on sources unless it's a subscription, response, or similar list (10% of the list industry) and I mean none! It is always nebulous (300+ proprietary sources, public records, telco, registrations, and telephone verified, etc.). That's why we say the list industry is the true "Wild West".
There are no sources given, and no adequate comparisons between lists. The cost to scientifically test is prohibitive for 99% of all mailers. There's no true guarantees, no consistency in how lists are selected (by the client or the list producer), no accountability, almost no understanding of choices, no recourse for junk lists, and there are thousands of places to buy lists.
All compiled lists are filled with modeled data so you never, ever get the list you order. If you target homeowners with $50K-income or businesses that sell $1mm you'll get some model someone wrote somewhere picking the records most likely to “fit”. Nobody knows where they got their expertise to do that, let alone where they got their records in the first place, let alone what data they have to allow them to "model" income or business sales. The IRS doesn't release data, you didn't tell anybody else your income or sales and neither did anybody else. Self-reported data exists on only a very small number of records and isn't updated for 5+-years at a time. Dun & Bradstreet has actual financial statements on less than 5% of their database yet every record has financials (95% modeled). InfoUSA has no financial statements (except on 3,000-publicly traded companies) and gathers no credit references so their financials are 100% modeled as are their “credit ratings”.
In short, almost all list information is modeled (fake). There are 60,000+ lists advertised in the lists of lists advertised to the list industry. Some guy working from his bedroom will sell you left-handed lawyers and 18-year old millionaires that want what you're selling. They're all in there but, again, it's almost all modeled data. Nice houses have high income, are owned, and have kids of this age and have this income and a dog. That's almost all modeled. How do you know if Superlists is using real data and GreatestLists isn’t? We never buy lists with unknown “vanity” names. Somebody is reselling something they get somewhere else. There are only a handful of truly legitimate list compilers.
Next, lists are already horrendously out of date (by years even when they’re just refreshed). And lists are ugly (especially business lists) and we've been hit by "Inyurass" type names as well as curse words and racial slurs even from "reputable" list compilers.
Next, with many people in many places making models there are obvious, huge skews as to how many fit a given parameter. That's why one list source has 5,000 and another has 50,000 records meeting your parameters.
And perhaps most important is that cheap, old data, wholesale data, minimally-sourced data is available everywhere in the list industry. If you make up a great-sounding list name "Ameri-great lists" and set up a website you can sell the cheapest old crap someone you know gets somewhere. Most list sources sell what they get cheapest, not what’s most accurate. Like the end-user, they have no way to verify what’s good. Sure, everybody says their data is good-but nobody can prove it. Mailings are almost all “bulk” so the mailer never sees the returns, let alone how so many addresses they mailed obviously missed their financial parameters, not to mention THOSE PEOPLE MOVED OUT X YEARS AGO or THAT BUSINESS CLOSED X YEARS AGO. The USPS reported 25% of direct mail is undeliverable as addressed and in many cases 25% more of those records miss their parameters.
After 30-years as list brokers we know all lists are bad, with many being 50%+ inappropriate or somehow incorrect. We always recommend the small, not larger list with more data likely known. We try to limit the selects [address holder attributes such as age and income] we offer our clients as picking modeled data that is built on models continues to exacerbate the inaccuracies and we stick with the expensive, current-month files of the giant, well-known list compilers. And we very, very carefully select our lists. A check-box most would never think of can correctly kill off the least accurate 38% of a list.
All of the same disasters in demographics in postal lists are appended to most email lists, again, unless they are specific subscriber or other response lists.
Email lists are often worse than postal lists because there are so many "live" but unused emails they get delivered but nobody sees them. I have nearly a dozen emails but use only 3 regularly. I use the others for purchases and inquiries so my preferred accounts don't get loaded with spam.
The saving grace is that consumer emails are so cheap many can tolerate a great deal of inaccuracy and if you pay for 50,000 the source will keep blasting until 50,000 are delivered.
Business emails in your hand are expensive so the source and selects used must be considered carefully.
The USPS NCOA [National Change of Address], CASS [Coding Accuracy Support System], DSF [Delivery Sequence File] and other processes catch a few small errors but have no way to verify anything but address accuracy. List sources show monthly “updates” but many are really only running postal processes and obtain “refreshes” from the compiler only quarterly or semi-annually. Business data is so difficult to gather and keep current. Experian, which was the 3rd largest compiler of business lists, stopped selling their own lists 3+-years ago and now resells InfoUSA data. InfoUSA is largely based on phone books, so their sources are often out of date when they first gather the data and it often sits there for 2+-years and so on. Contact names often go 5+-years without true verification. People upload their Outlook contacts to Jigsaw and other trading sites so compilers think this is brand new data yet that person may have left that position 5+-years earlier. Last, only a small percent of households fill out NCOA “forwarding” forms and never tell list compilers they’re moving and no business tells list compilers when they go out of business.
[In an email exchange I asked Nagle if I should clarify that he was talking solely about compiled lists. He responded:]
Yes, I primarily complain about compiled lists. However, nearly ALL lists including subscription and response lists (snail and email), have demographic/lifestyle-type data elements appended from a compiled list source. Magazine publishers don’t ask your age or income or lifestyle interests, yet all of those fields are available on almost every record in almost every list. The problem is that list managers often use the cheapest source to append income, age, etc., so the inaccuracies and skews of various models come through in nearly all lists.
There’s a second issue rarely considered: match rates and quality. There are many matching programs used to match source data to a given snail or email record. Using the default settings of Alvion technology, one of the most common match program we match 25% of business records and the matches aren’t very accurate (you can spot mismatches). When we use the default settings in Dun & Bradstreet Optimizer software we get 60%+ and they’re all quite good, but it is very expensive. D&B has matched hundreds of billions of credit references to credit reports and because the references are often derogatory, any mismatch could generate a lawsuit so they are very careful and only use the best matches for trade references.
In nearly any matching program we can “loosen” the parameters to increase the match rates (which also increases the mismatches). Nobody knows what various matching programs are used by what sources and how “loose” they were with the match results. A list provider can’t sell many records without demographics appended (sales or income as an example), so they are motivated to loosen the parameters as much as possible.
The list industry really is the unregulated and largely misunderstood “Wild West.”