110. Where do I find mailing lists?
You rent (and sometimes buy) from other people.
You exchange them with other people, especially competitors’ old names.
You compile them externally – directories etc.
Your own list – your database.
Compiled In House (i.e. other products, other divisions).
How many lists are there?
Lists available in UK
Lists available in USA
What does it cost to rent a list?
Rates per ‘000
£75 to £100
Selection and production charges extra
You can negotiate for
– Old names
– Repeat usage
111. What does list rental cover?
Rental is usually for a single use.
But you can negotiate alternative terms: multiple use, alternative use (e.g. telephone).
You can even buy (a quick but not ideal way to set up an instant database).
A non-mailable extract from the list for analysis.
Usually broker or owner inserts ‘seed’ names to prevent cheating.
Normally you pay for names mailed – not for duplicates – but the industry ‘norm’ is that renters pay for at least 85% of the original quantity
Larger users can usually negotiate a better ‘net name’ agreement.
112. What do list brokers charge? And who pays?
Normally list brokers manage the transaction between list owner/manager and you.
They make 20% commission from owner.
Compiled In House (i.e. other products, other divisions)
113. What if some of the names are out of date or inaccurate? Do I pay?
Normally you pay for names mailed – not for duplicates but the industry ‘norm’ is that renters pay for at least 85% of the original quantity
If you’re a larger user, then you can usually negotiate a better ‘net name’ agreement.
114. What different kinds of lists are there?
Buyers / Enquirers
Seminar / conference / exhibition attendees
Questionnaire respondents / referrals
Promotion respondents (ther divisions)
115. What questions should I ask a list broker?
You must remember that the list broker is there to sell you something – and salespeople are obviously biased.
So you should always ask: where does the mailing list come from? What is the profile? Are they buyers or enquirers? How recently have they bought or responded; how frequently; how much did they spend; how did they pay? – anything that seems relevant.
Who else uses the list? Clearly if one of your competitors uses it regularly it’s likely to work for you.
How often is it mailed? The more a list is mailed, clearly the more responsive it is, otherwise people wouldn’t bother.
What selections are available? It may be that some names on the list will work for you, whereas others won’t.
What are the conditions for future use? It’s not much use testing a mailing list to find out you can’t use it again.
And don’t forget about cost, the format the list will be delivered in and whether it is mailsorted.
116. How do I find out which lists are most likely to work for me?
If you compare your list with ones you are thinking of testing, you can predict which are likely to work.
If the duplication factor between the two lists is 25 percent, your mailing has about an 86% chance of succeeding.
If it is under 10 percent the likelihood of success is about 3 percent.
These figures apply to large customer mailers, though the principle applies to any business.
117. When should I buy a list?
You normally rent a list, you don’t buy it; it is most unlikely that somebody else’s list is going to be perfect for you, even if they were prepared to sell it. Exceptions are when you stumble on something clearly ideal, such as a customer list of a competitor who has gone broke.
Thus, on one occasion this writer was able to rent a list of names of habitual gamblers who are obviously the perfect prospect for anyone in that market. Despite the fact that £1 a name was the price paid – and this was over 20 years ago – the investment was well worth it which leads to one final point:
Just make sure that you are obtaining a list that you buy legally and it is not something which has been stealthily nicked by somebody from a previous employer.
118. How do I check the quality of a list?
You do an Nth name check. That is, you telephone a random selection of names to see what percentage are accurate. This can also be done through the mail, by incentivising people to reply.
It is vital that you do check the quality of a list, particularly when mailing to businesses, since people change their jobs with great frequency.
119. What use is the Electoral register?
Compiled annually this covers over 90% of UK households. You can use it as a mailing list, though without additional data it isn’t selective enough to be much use. There are very few products which appeal to everybody.
You can also use it to check out of date names and to add data e.g. how long at address, household composition, new entrants (movers and 18 year olds) etc. It is quite possible to mail households that have moved recently very effectively with certain products and services.
120. Why do names and addresses get screwed up so often?
I went to a leading expert on this subject, Ian Goodman.
Mrs. J Smith
The computer compares these two names at the same address and decides they are duplicates. In fact they refer to mother and daughter. What if Mrs. Smith is a bad credit risk, but her daughter isn’t?
Invalid marketing information:
The computer discards Janet Smith after crediting her history of purchases to Mrs. Smith, who now appears to be a splendid customer.
False name creation:
The computer creates a non-existent person by amalgamating the names into Mrs. Janet Smith.
A Mr John Smith, Motoring Ltd
B Motoring Ltd
C Mr John Brown, Motoring Ltd
The computer compares A with B and decides that A=B. It marks B for suppression. The computer then compares B with C and decides to mark C for suppression. Thus only A will be mailed.
Same name, different firms:
Mr Smith, Motoring Ltd
Mr J Smith, Cars Ltd
Suppose both records refer to the same address. Either they refer to the same person who works for a group of associated firms, or they refer to two different people whose firms share the same business premises. How does your computer cope?
Mr John Smith, Managing Director, Health Care Ltd
Mr David Jones, MD, Health Care Ltd
Perhaps the firm has joint managing directors. Perhaps one of these records is quite old, referring to a person who has retired. What does your computer make of this?
False title creation:
Mr David Jones, MD, Health Care Ltd
Does your computer ‘improve’ MD to managing director? Or does it realise that Mr Jones may be a doctor?
Computers rarely demonstrate any flexibility about the way they sort names and addresses into geographic sequence prior to deduplication. The tiniest spelling variation in two otherwise identical names and addresses can play havoc with a computer sort, forcing the records so far apart that the computer cannot spot the duplicates.
Suppose the computer is sorting by street name. Consider Alder Road and Older Road. Imagine the number of other street names which may be sequenced alphabetically in between!
Tie-breakers and heart-breakers:
Suppose the computer resolves its sorting problems and notices the two addresses for Alder Road and Older Road. How far will the computer go to test if one of these names has been incorrectly spelt? Consider the possibilities.
Alder Road exists, but Older Road does not;
Older Road exists, but Alder Road does not;
Both Alder Road and Older Road exist;
Neither Alder Road nor Older Road exist;
It is possible to match either street name with another street in the locality, after allowing for a different kind of spelling error. For example, Alder Street.
The computer might resolve these conundrums by referring to a gazetteer of all known streets in the given locality, using any given postcode as a tie-breaker. But it does not take too much imagination to see that the rules required to break the tie are extremely complex, being variably dependent upon both the given names and addresses and the particular mix of real street names recorded in the gazetteer in the given locality. How far does your computer go in resolving these problems? What protections exist against incorrect address correction?
The examples given above assume that the people entering the names and addresses do it properly. On a surprisingly large number of occasions this is not the case, often because they are not really paid enough to do a decent job. A lot of data entry is actually done by people overseas who need the money and don’t know very much about Britain.
121. How do I avoid duplicating my mailings?
You clearly don’t want to mail the same person several times – and reference to question 122 will tell you how this sort of thing can happen.
But another reason is quite simply that several lists may contain the same person. For that reason we de-duplicate – or merge and purge as the expression goes. This is a process where you run the lists against each other to eliminate these duplicates.
A great many otherwise sophisticated organisations appear to be very bad at this, which is why you will frequently find yourself receiving several mailings offering the same thing on more or less the same day.
122. Which factors help most in determining list quality?
Have they bought recently?*
Have they bought frequently?*
How much money do they spend?*
How did they pay?
Charge card? Credit card? Cash? Easy terms?
How up to date is the list?
Is the name and the title of the person given? You’ll often do better mailing by title most critical discriminators within a buyer’s list
* These three questions cover the three main parameters used in data profiling and segmentation: Frequency, Recency and Monetary value, often referred to with the acronym RFM.
123. What is profiling? How does it help?
A profile of a customer or a list of customers tells you what kind of people they are. It includes a number of factors such as where they live, what their incomes are likely to be, the children they’re likely to have – indeed all the things that you’d want to know if you were trying to sell to somebody as an individual.
Those lists which have profiles most similar to yours, are the ones that you are going to do best with.
Analyse your computer file of present customers by value or category.
Then draw up a profile to help locate similar prospects.
The profile will take into account many characteristics …
The following criteria are not exhaustive but show types of data you may use.
- Name (Maybe use ‘Monica or Anna’, ethnic type – even number of initials).
- Address (Acorn or Mosaic type, perhaps household composition from Electoral
- Register, flat or house; named house or street number)
- Past purchases
- Payment method and performance
- Recency and frequency of purchase
- Response to specific promotional messages
- Any ‘Lifestyle’ indicators – NDL etc.
- Title or function
- S.I.C. code
- Firm size (employees)
- Firm size (turnover)
- Purchasing behaviour
- Promotional response
These lists are not exhaustive but show types of data you could use.
124. How can I benefit from a list profile?
You use the profile:
To identify in your customer/prospect file which are most likely to be good customers.
Concentrate on them
To identify those least likely to become good customers.
Spend less on them
Compare with other lists (e.g. rented lists or in-house records for other products/divisions) before deciding whether to invest in them.
125. What is the difference between a list and a database?
Lists were relatively inexpensive because everyone understood what they were, and thus what they were paying for when they compiled or rented one.
‘Database’ is a word many do not entirely understand. A very profitable industry has evolved to cater for this ignorance, with predictable results: many people spend prodigious sums, even millions, without gaining a great deal of satisfaction.
But if you start by thinking about lists, you will end by understanding databases and what they can do rather well.
Once you add any information beyond the name and address to a list, it starts to become a database. Because you learn something about the nature of the people on that list. Thus, a telephone number tells you they have enough money to have a phone – a small but significant piece of information.
Other much more detailed information such as the type of property they live in, the number of transactions you have had with them, their ages, and so forth all build up a picture of them.
The accuracy of this picture will determine your ability to communicate effectively. The closer you can get to picturing the real individual, the better a job you will do – which is why the database is so important.
126. What should go onto a database?
Everything you’d want to know about somebody if you were selling to them personally.
What kind of people are they? What kind of area do they live in? What sort of accommodation? What do they own? How many people in the household? How old are they? Married or single?
What’s your relationship with them? How did you first make contact with them? How often have you had contact with them? How recently? How did they react? What have they bought? When? How much did they pay? How did they pay? How recently have they paid?
What size and type of firm? How many employees? How many offices? Where? What’s their credit rating? What do you know about their assets, capital equipment, financing?
When and how did you come to deal with them? What’s the promotional history? i.e. what mailings, telephone calls, sales visits have been made — with what results?
What have they bought? When? How did they pay? How quickly did they pay?
How do they make decisions? Who influences, specifies, decides, and pays? You may wish to talk to each individually, addressing his or her motivations.
127. How can I plan my database simply?
If your organisation is of any size, then you will employ experts for this but if it isn’t, here is a checklist prepared by my colleague Andrew Boddington to give a pretty good idea of what to look for.
It is also useful even if you are employing experts who, as you well know, have a tendency to get things wrong quite spectacularly from time to time.
12 basics when building and using a database
Install a flexible, useable PC system for humans, not computer whizzes to use. You don’t want to be blaming poor results on the computer.
Map out the sequence of typical contacts with different types of customer
- customer or business
- decision maker or influencer
- enquirer or buyer
- frequent or infrequent buyer
Think of all the logical pieces of information these contacts could produce and allow for them in your system.
Aim to record every contact between you and your customer or prospect. This is how our brain remembers the building of relationships.
Get business and personal details 100% accurate. Nothing else is acceptable. You might as well shut the door in their face if you don’t.
Link your database to a practical WP and printer, to produce letters which are quick, easy , variable and presentable
Train all staff on the importance of accurate data wherever it can be recorded. It is as important as putting money in your cash register.
Give one person clear responsibility for the database and its administration. Make another person responsible for planning and executing all that is communicated to it.
Use questionnaires. People will reply if you use charm – and explain they are used to help improve service. Many more will reply if you give a small incentive. (Anything up to 70% can be expected for your own customers).
Send the results back to respondents: they will appreciate it.
Plan well in advance. Have a calendar of mailing and seasonal events for customers – especially those who are worth most money.
Pay great attention to detail.
128. What’s a lifestyle database?
Lifestyle databases were originally developed in the United States by NDL.
The data is gathered by questionnaires distributed in several ways – direct mail, inserts in publications, door to door and on guarantee cards. Customers are encouraged to respond by various incentives – the opportunity to win a sweepstake, free samples of various products and so on. Up to 150 questions may be asked on people’s behaviour and other characteristics including what they like to buy.
Advertisers sponsor questions related to the market they’re interested in. They use the names and addresses and data produced to target better.
An example of this was Chivas Regal who wished to target people who drank a competitive brand of whisky.
The lists produced can be rented by non-sponsoring firms. There is a further opportunity because the firms building the databases send out communications to respondees – vouchers, merchandise and so forth.
The great advantage of this form of database is that you can select individuals and households based upon genuine knowledge about them as opposed to supposition.
129. What is a mailing list worth?
A mailing list of your own customers, is a pearl beyond price.
To understand why, ask yourself why firms take over other firms and close down all their branches. It’s because the only thing that interests them is the customers, particularly in the financial services industry.
Other ways to look at the value of a mailing list are based upon what it will produce in revenue. First of all, if you exploit it properly it will produce a tremendous amount of revenue for you. Secondly, if you rent it out, it will also produce revenue.
There have been quite a few examples in the mail order business of firms whose entire profit could be attributed to the rental of their mailing list – everything else simply broke even.
Clearly the value of the list will be determined by what people are prepared to pay for it. That can be established by looking at what the rental rates are like for other lists of the same kind. It is not unreasonable to assume that you could rent out your list once every month and maybe more often.
If you only have a small list of 100,000 names and you rent it out just once a month. Under such circumstances only good reasons of corporate policy can drive you to non-rental.
But you must take precautions. Seed your list with trap names – the names of people who work for you, or are associated with you, to monitor what goes out.
And you must let those who rent the list know that your list is so protected. Equally, before letting anyone use your list, make sure they show you what they are sending out.
Other sensible precautions include storing a copy of your list elsewhere, in case of fire or theft – indeed, anything you would normally do with something of value. Simply think about the list not just as some names, but as a precious business asset and act accordingly. Maybe just by calling it your database you will treat it with appropriate respect.
Incidentally, one consequence of renting out your list is that you get to see other people’s packages and learn how they are doing.
130. What should I study on a database?
Just about anything.
You may find a particular group of people is exceptionally responsive – or unresponsive. You may note that a particular group of people was mailed a year ago, which gives the opportunity to send out something relevant now.
You may notice that a particular group of people have been mailed many times and not responded giving you grounds to start thinking how you might motivate them to do so or whether you should bother at all.
It cannot be emphasised too strongly that careful study of the database and the behaviour of the individuals on that database and consequent segmentation of your messages almost invariably pays off.
A good example is British Airways who sent out a successful mailing to people they had noticed had travelled one way with them but not returned.
These individuals were offered an incentive to fly both ways – 500 air miles. Every piece of information which is relevant can help you in building up a picture of your best prospects or customers and enable you to do a better job for those people. And the principle applies whether you are talking about business-to-business or customer marketing. The information you use to enrich your database can come from many sources, internal or external. And you can acquire it in many ways.
One way is to look at who has responded to your previous communications, and discern the characteristics they share with other respondents. The converse is equally true: you can see who does not respond. In this way, you can target your communications to those most likely to respond and eliminate those least likely.
This means that you can take a list of people which as a whole would not be profitable for you to mail and by eliminating the least responsive names, cut down the size of that list and make it profitable to mail the remaining names. In effect, you transform failure into success. Alternatively, you can make a list which is already profitable more profitable.
Generally available information can enrich your knowledge of the people on your database. For instance, in the public domain there is often information about residential areas: what sort of people live in them, what sort of accommodation they inhabit. This is usually available through Royal Mail or census statistics.
You can also gain information from your order forms; often simple but valuable data, such as people’s phone numbered, how old they are, how many children they have and of what sex and age.
By comparing this with the responses from other customers you can predict likely responses from those people. Every shred of information helps build up a picture of the people you are talking to and their probable behaviour.
If you like to blind your colleagues with science, the phrase ‘regression analysis’ can be used to fine effect when discussing this subject.
If you like to impress your colleagues with good profit and loss statements, generally this technique – though not cheap – tends to generate more profit than it costs: indeed, you can double your profitability.
Of course, unless you have your list properly organised, with every possible factor about your buyers incorporated into the database, such analysis is very difficult.
131. Which lists pull best?
Your names are better than others’ names.
People who are known buyers through the post are better than people who are not known buyers.
People who are known responders will do better than people who are not known responders.
Promotional names are better than enquirers.
Enquirers are better than compiled names.
Internally compiled are better than externally.
Frequently mailed are better than infrequently.
132. Can I predict if a list will work – before I mail?
Reader’s Digest in the US, is one of the most intelligent commentators on the subject of lists. In December 1982, an article of his, ‘Foreshadowing response: the muscle in your merge’, demonstrated in the most startling way how important it is to remember the fact that new customers’ responses can be predicted from old.
Grossman noted that if you compare the lists that are working for you currently with ones you are thinking of testing, you can predict with a high degree of certainty which lists are likely to work.
When people rent lists, you have to complete a computer run to eliminate duplication – known in the trade as ‘merge and purge’. Having removed this wasteful duplication the average mailer carries on. Not Grossman, because he takes a look to see how similar the lists are. He points out that in his experience:
1. If the duplication factor between the two lists is 25 %, your mailing has about an 86 % chance of succeeding;
2. If it is under 10 % the likelihood of success is about 3
Much more recently a direct marketer attending a seminar I was conducting told me her firm had discovered the same figures applied to within 2 %.
133. What if I have no historical data?
You cannot possibly afford to guess what is likely to happen. So you must start building data as fast as you can, because as it accumulates it will become increasingly valuable to you.
Lacking such data, you should be extremely conservative in your approach. Obviously you will be testing everything before you try to spend any large amount of money. The results of those early tests provide the foundation for your future historical data. As you acquire more knowledge, you can be more daring about what you do, because you will have discovered, as time has passed, how much customers are prepared to buy and how profitable they are.
But do remember a customer may stay with you for an average of 5, 6 or 7 years. It is not until that time has elapsed that you really know what you can afford to spend to acquire customers.
Very sophisticated organisations may wait months or even years before they actually profit from any newly recruited customer.
134. How do I build a database?
A list can belong to anybody; but your database only belongs to you, designed to meet your particular needs. It is a list of names and addresses to which are appended all the things you think might be relevant; the information you can discover that will help you deal better with these people: by talking to them about the right things at the right time.
One point to remember: don’t waste time and money acquiring information about people and placing it on your database – which costs money – when that information is, however interesting, of no real value. This is a mistake often made.
There are only two ways in which you get people on to your database. You may know enough about them to identify them as prospects for what you’re offering. You therefore place them on your database and start communicating with them until you find it is no longer worthwhile, because they are clearly not going to respond. Or they identify themselves by replying to communications of one kind or another – advertisements, commercials, take-ones, or syndicated questionnaires (more about this in a moment).
Any business must have a list of customers – even if it’s only to send out bills. This is your database. All you have to do now is enrich it by adding information about these people on to it.
Most people actually have important lists under their noses which will form the basis for their database and either fail to see their value or don’t realise the names are there just for the compiling.
Some years ago, somebody in the very competitive business of photographic developing came to see us. He spent most of the meeting telling us what a hot shot firm he had. At the end, we asked him what he did with his database. ‘What database?’ he asked. ‘The list of all the hundreds and thousands of people who send in for free films,’ we replied.
‘Oh, those. We take their coupons and use them to send the film back – it saves money,’ he replied. ‘Making a list would be far too expensive.’
Anyone who thinks that a list of satisfied customers is not worth the expense of compiling is being very shortsighted indeed. And when you are in a highly competitive – even cut-throat – business like the one this gentleman was in, you should be particularly anxious to communicate regularly with the customers to retain them.
So the first place to look for lists which can build you a database is in your own backyard. Some of the sources are so obvious it’s easy to overlook them.
135. Collecting customer names
Gordon Grossman once wrote an article called: ‘If your customers won’t make you rich, then who will?’
It’s vitally important that you trap the name of every customer you have, or have had in the past. They will come in from your direct advertising and mailing obviously. But what if you’re just starting out on direct communications? Then you must look elsewhere.
Get everyone who deals with customers to record their names and addresses. Your telephonists. Your receptionists in the showroom. Your complaints department. Your service people. Your sales people. Your marketing people.
If you have retail outlets or agents or a chain of dealers, get them all to do the same.
Do you offer a guarantee? Get those names and addresses on file.
What about your past records? Invoices you have sent out, old customers (a lapsed customer is one of your very best prospects), past enquiries.
Do you attend exhibitions or shows? Collect names there.
How about competitions? Build in a reply device to capture the names of entrants. They should be useful, too.
136. How do I keep a list up-to-date? Do we have details right?
One of the big mistakes mailers make is not to give people a regular opportunity – every time you mail if possible – to confirm that the details you have about them are correct. If you have an old list it often pays to do a mailing with no other objective. It worked for us with a large client who had names which may have been very much out of date.
137. Who are the best new customers and how do I find them?
Just as your own customers are the best people to make you money, they are the best source of new customers. Always spend time and thought on ways of encouraging them to give you the names of possible new recruits. This activity – known variously as ‘Member-get-Member’ or MGM, and ‘Friend-get-a-Friend’ – tends to be the most cost-effective of all ways of getting new customers.
It will normally pay you to offer incentives in exchange for the names of customers’ friends. (The incentives only being received when the new name actually makes a purchase.) Usually it pays to offer the new recruits a gift, too. That way they do not feel exploited.
I know of no easier way of increasing your profits because the communication that solicits the new customer goes out as part of your regular commerce with existing customers. It rarely requires more than a simple leaflet, sometimes less.
Unlike a mailing or an ad soliciting new members, you have no need to spend too much time persuading people your product is good; your satisfied customer is already convinced. In fact, the most cost-effective example
I have ever seen consisted of a simple one-line request on an envelope flap, offering no incentive at all. Nothing could be less expensive; I have never heard of anything as cost-effective.
Also, try a small message in your communications suggesting to recipients that if they are not interested themselves, they pass the message on to someone they think may be. A good place to do this is behind the window on an envelope in a mailing so the message appears when the contents are removed. Another is in the PS to the letter.
In business-to-business mailings it’s particularly important to do this sort of thing. That’s because although it generally pays you to mail people by name, this is not always true: sometimes the additional cost is too great to justify the additional response.
Often people move positions within organisations, or from one firm to another, so that you may be better off mailing to a title. This is not always the case, but it generally is because of the large number of executives who change their jobs – or lose them – each year. One respected UK list broker stated in 1992 that 30-40 % of all names on the average business list at any time were likely to be inaccurate.
Often, it may not be at all clear who the real decision-maker is in the case of your product or service. That makes it particularly important that if you mail the wrong person they pass the message on to the right person.
Never assume that people will do it automatically. However, if you point out to them that the better a job of informing their colleagues you can do the more it will benefit their firm, they may be more than happy to help you.
138. How responsive are compiled lists?
Not as responsive as people you know have responded to messages.
Generally speaking, compiling only pays if you have a subject which is rather rarefied – for which there is no list available from a broker.
Even then, though, it may pay you to go to one and get him to compile it for you. He may do it cheaply in exchange for the right to hold the list and derive revenue from renting it out to other users. You obviously make sure you retain the right to restrict rental to people you approve of, i.e. non-competitors.
Often the value to you of reaching a named executive can be great enough for it to pay you to telephone firms to find out the right people to talk to. You’ll also pick up valuable information on how buying decisions are made.
139. What is the effect of a teaser or advance mailing?
If you have a mailing that is doing very well, and you want to get even better results, it’s often a good idea to use a teaser or advance mailing, saying “something interesting is coming your way”.
It is even possible to use the telephone in the same way. In a campaign for Xerox in Europe, a telephone call was made to prospects saying, “Keep an eye on your post – you will receive a pleasant surprise in the next two days”.
Teasers, like everything else, should be tested. On one occasion for reasons which we have never been able to understand, we sent out a postcard in advance of a mailing which did absolutely nothing for the response apart from increasing the cost.
140. What happens if I send a follow-up mailing? When should I do so?
If all new customers are profitable you must ask:
‘Are we getting as much new business as we could?’
For example, a follow-up mailing, sent two or three weeks later, usually gets half the response of the first mailing.
141. How long should I keep following up enquiries?
One of the biggest mistakes you can make in direct marketing is to assume that simply because somebody has not bought, they are not interested. Often they simply haven’t got round to it.
Essentially though, you should keep following up enquiries as long as it pays – i.e. as long as you produce more revenue than the mailing costs. By revenue we do not necessarily mean immediate profit, we mean revenue based upon your assessment of what it worth paying to acquire a customer.
142. What impact does cold direct mail have on behaviour?
Martell mailed a cold list to see what impact one mailing would have on behaviour
Amongst other things 72% remembered the mailing after one month. The increase in intention to buy in future was 24% amongst responders, with a remarkable increase of 23% amongst non-responders. Amongst responders, there was an increase of 27% actually claiming to purchase Martell.