Article #26: Sample from SSI-SNAP

How good is MR sample ordered on-line? Tim Macer finds out

Playing the Internet number game

The shoes didn't fit, the shirt was out of stock, but the wine was drinkable: these are some of my recent experiences of Internet shopping. Oh, and the telephone sample was exactly what we ordered!

Using a piece of software called SSI-SNAP, which claims to be the "world's only on-line sample ordering system" you can now order sample over the Internet. Sample on tap from the comfort of your PC sounds like a good idea. But for a research 'consumable' where quality can only truly be measured rather late in the day, when you see the tables, you want to be certain this sample is at least as good as what is obtainable elsewhere.

Click on the image for an enlarged view

SSI-SNAP, produced by Connecticut-based Survey Samples Inc. (no relation to Mercator's all-in-one questionnaire and analysis package SNAP) is presented as a sophisticated client/server solution where you use the Internet or a dial-up connection to draw sample directly from SSI's databases.The main screen deals with the basics, such as the quantity you are ordering plus geographic data. For UK sample this starts at country, and goes down to county, postcode (first two letters of the 'outbound' portion), Neilsen region and even ITV, including all the overlaps. It is a typical Windows point-and-click tool that would take any PC-literate user about half an hour to master.

A separate screen allows you to tailor your sample to include or exclude specific regions, from a whole country right down to an exchange code or postcode (again first two letters only). Another screen lets you refine your sample by screening out business phone numbers, any numbers that you have surveyed or non-active phone numbers. In the latter case, SSI will trial-dial all the numbers it selects and only deliver ones that connect. They claim to be able to do this without making nuisance calls, as intercept messages cut in before the time it takes to ring the phone.

You process the order on-line, and in most cases, sample will be sent back to you within minutes. The end result is the same randomly generated sample you would receive if you placed an order with SSI over the telephone or by fax. SSI's databases contain all the numbers in the phone book for the countries they cover. But these are not the numbers they send you. As SSI Vice President Terry Coen told me, "I make my living from selling people made up numbers" A lot of numbers at that. Last year was their twenty first of existence, during which they supplied sample for over 43,000 surveys worldwide.

Samples are built up from random numbers based on valid codes and existing numbers, but in proportion to the distribution of households throughout the country, not by the incidence of listed numbers. Ex-directory numbers account for 34% of all residential numbers according to a survey by Gallup UK. As Terry Coen put it: "With cable and other providers, like Cable & Wireless, it is higher now. And in London, we are probably closer to 50% unlisted. If you use listed numbers, you will over-measure the areas where listing is more common, such as rural areas. Our database is weighted towards the population".

John Choi is CATI Manager at Millward Brown, and one of SSI-SNAP's major customers in the UK. "It is very good," he told Research. He particularly likes the way that every order you place is kept on the PC, so you can open up an existing order and request another identical batch whenever you need to. As they undertake a lot of continuous research at Millward Brown, this feature is especially convenient, as is the ability to order replicate samples. These are ideal for tracking studies, as very similar phone numbers are substituted from the original set, giving an identical geographic spread.

John Choi continued: "We dial to the US to get the sample. You can also use the Internet. You download a zip [compressed] file so it does not take long - 15 to 20 minutes to get 40,000 numbers." He always screens their sample for non-working numbers. "If you don't screen the numbers our experience is that 26% of the sample in the UK is unusable." By screening, they reduce the need to dial non-working numbers. If you ask them to screen the numbers, they then check the numbers for you. "We find after that only 1% are not connectable, so it is really very efficient. We have been using it for two years and that is the average we get. I don't see that there is any problem with the quality of the sample. Personally I think this is the best on the market."

Of course, there are other providers. And random digit dialling sample is not without its critics. Added to which, Terry Coen reports reluctance on this side of the Atlantic to order on-line rather than fax. But surely the speed and convenience of the Internet means that soon SSI-SNAP claim to be "the world's only" will be as out of date as last year's phone book. Or should I say this year's phone book?



Software very easy to use

A very quick way to obtain sample

High incidence of usable numbers

Single source for international sample

Orders are saved and can be re-used for repeat orders

Need to install software on a PC: can't use a Web browser

Not effective for b-to-b

RDD sample not good for small area or low incidence samples


SSI, phone free from UK 00 800 3589 7421 or +1 203 255 4200


Published in Research, the independent voice of market research, February 1999, Issue 393.

© Copyright Tim Macer 1999. All rights reserved. Reproduced with permission.

Top | Index