The 12 Most Important Things Business Owners
Need To Know Before Trying Direct Mail

The easiest way to explain this is with an example:

I got asked once at a business meeting whether I could help someone with a mailing project. So I asked him to tell me about it!

He told me he had a friend who had written a book about his experiences as a drug smuggler, how he got caught, went to jail and did outstanding work helping prisoners get their high school diplomas. I've read the book and it is an interesting story that's going to be made into a movie.

He wanted help selling this book by direct mail to social workers and people in the justice system who could benefit from his story.

I asked him what it sold for and he said $19.95. It would cost $3.50 to print the book and the packaging to ship it. Then I asked if he had anything else to sell the people who bought the book. He said no.

At this point I could tell this project couldn't work so I told him why:

With a sales price of $19.95 and a cost to print the book and package it for shipping of $3.50, that leaves a gross profit of $16.45. We assumed he could charge shipping and handling to cover postage and fulfillment. Take away another dollar for the cost to take the orders on the phone or process them and your profit is now $15.45.

Making an assumption of a 60 cent cost each for the mailing, the break-even point is 3.88% (.6/15.45 = .0388). To make any money at all, say just doubling his mailing cost, he'd need a 7.76% return. Highly unlikely.

What really kills this is that he has no back-end, or other products to sell his customers and the book is a one-time sale. No one needs two. On top of that, his target market of social workers don't have much money either! (If he had other products and services to sell his book customers, it could work well even if he lost money getting the first sale, but not without).

He might also be able to make it work in a very niched magazine or newsletter as a direct response space ad, but that's a whole other story!

The lesson here is to figure out what response you need for a success before doing anything else to see if it's feasible.

Assuming you are wanting to sell a product or service by mail, the next step is to make sure your market is big enough and you can identify who you need to mail to.

There is a reference book of lists that you need to become familiar with. It's called the Direct Mail List Source by Standard Rate and Data. It's available at big city libraries or directly from the publisher at www.srds.com. This book contains information on over 32,000 direct response lists.

This book will not only help you find a list, but it can also tell you how big your market is. Most people outside of the direct mail industry have never heard of this reference or don't know how to use it this way.

Here's what to do: Look in the Standard Rate and Data book to find lists for rent from your competitors. If you can't find anyone selling whatever you want to sell, that's not a good sign. DM may not work for your product.

Next, after you've found these lists, look for the "90-day hotline" buyers.

"Hotline" means the most recent buyers. So, a 90-day hotline, means their customers from the last 90 days. Most lists have this feature because they can charge more for these recent buyers so they segment it out. This tells you how many customers they've sold in 90 days, and the average order of those customers.

Now, multiply the average order by the number of customers, and you've got their sales for 90 days. Multiply that by four, and you now have that company's sales for the year.

Go to the next list and do this again. Once you have gone through all of the lists in your market, add them up and that's how big the market is for direct response sales of your product.

If you find your market is small, and you have a couple lists of 10,000 names, and their 90-day buyers are 1,500 people, you'll maybe find out that that market is too small to even want to mess with.

So, once you've figured out this total market size, again consider if your project makes sense. You could start by using a number like five percent market penetration, which you'll not usually get with one mailing; it'll take multiple mailings, usually, to get that much. So, how much business is available?

If you find that the total market is $100,000, then five percent of that is only $5,000. So, you might want to take another look at your project — it may not be worth it.

Do you want to go through all of the expense of developing and testing a mailing just to make $5,000 in sales? Take a look at your market that way and you might find out that it just doesn't make sense.

Here's a better way to develop a direct mail product in the first place — do this analysis first, find a market where there are lots of buyers and then develop your product!

The lifetime value of a customer is what an average customer is worth to you over their lifetime as a customer. This is important to know so you can determine what you can afford to pay for a new customer.

Understanding this can keep you from making bad marketing decisions. For an example, let's say you did a marketing campaign to get new customers and you made $10,000 in sales from the program but the campaign cost $15,000. Did it work?

It would be easy to conclude the program was a failure and that's what 99% of all people would say.

But, what if that $10,000 in sales came from 30 new customers? If your lifetime value of a customer is $3,000, these 30 customers are ultimately worth $90,000 to you for a cost of only $5,000. That's $166 to get each customer who will be worth $3,000 ($5,000/30 = $166.67.)

Looks like a success to me, not a failure!

By knowing this, the same campaign with the same result has changed from a disaster to a profitable program that should be continued. I've actually had this happen with clients and my first contribution to their business has been to continue doing something that worked that they thought failed!

Profiling customers means to learn demographic, geographic, and psychographic things about them so you can more effectively find more people or businesses who are like those who have already bought from you.

If you break your customer list down into 10 segments, you'll probably find that one or two of those have a much higher percent purchase than the other eight or nine, so if you are going to rent a list, it makes sense to only rent the segments of the list like your best customers.

Normally, customer profiling is only done by big companies with mailing lists in the millions because it can be quite expensive to do, but you can do simple versions of profiling affordably.

There are two types of profiling. One is business-to-consumer (BTC) and the other is business-to-business (BTB). With BTC some of the characteristics can be the amount they buy, the geography where they live, and this can be done all the way down to the neighborhood.

BTC Profiling

Demographics can include age, sex, how many kids, and things like that. You can even do a combination of geographic characteristics and demographics which is called geo-demographics. To make it even more complicated, you can add the psychographic lifestyle factors like hobbies and interest factors.

If you set up customer profiling that combines all of these things, it can get expensive, but the more data you can get on your customers, or the more data you have on them, the better your model will be.

To find companies that do this, put terms like "customer profiling," "list profiling" or "data mining" into Google and plenty of them will come up. Cost for this runs $3,000 and up.

I have a software program that allows me to do disposable income profiling for free. If I were selling to consumers and could only pick one criteria, that's the one I'd pick. Many direct mail experts say to start by testing your project to the top 10% of the incomes on your list because if it doesn't work with them, it probably will not work with people having less money.

BTB Profiling

For business-to-business profiling, one of the first ways to sort is by SIC codes up to eight digits, which gets into very specific industry groups.

You can add to that the size of the company, the number of employees, the credit rating, metro areas vs. rural areas or many other criteria. One indicator that can be very revealing is dollar sales per employee. To calculate this, divide the company's revenue by their number of employees. In any given industry, the more revenue they have per employee, the more disposable business income they have.

The businesses at the bottom end of that calculation are likely in financial difficulty, and may not have the money to buy whatever you are selling.

I found in one of my businesses that more of my prospects are located in the rural areas than they are in the metro areas. I don't know why, but in certain parts of the country they just don't seem to buy. That's good to know because then you can stop mailing into those areas.

For this business, I had Info USA do an SIC profile for $500. This did not include any of the other criteria, but that's OK.

Here's how I used that information. First, I verified which industries were best prospects for my product. My top five SIC's, out of about 30, accounted for more than half of my sales.

Then, I took the total number of companies in each SIC by state (from the Info USA catalog) and compared that to how many customers I had in each SIC in the state. Dividing one into the other gave me my market penetration by SIC and by state.

The differences were dramatic. This allowed me to focus my future mailings into the areas that have been proven to be the most receptive.

My Best Spy Tip

Rent your competitor's list and profile it! Take advantage of all the money they have spent developing their list and get that info for free! On top of that, have them send you a copy of what they mailed, too!

Mailing lists can be grouped into two categories, compiled lists and amp; response lists.

Compiled Lists

A compiled list means a list that is compiled from a database, such as the Yellow Pages. Companies like Info USA compile the Yellow Pages from all over the country into this one, huge database that has 13 or 14 million companies and many millions of consumers. Then, they sell that list, sorted however you need it: by type of business, by size of business, by SIC code.

For consumer lists, you can get sorts based on income level, based on residence address, based on male/female, black/white, kids/no kids. You can get people by their birth dates. You can get the owners of '67 red Corvettes who live in Fort Lauderdale. You can get an incredible amount of variety and variability.

Probably anything you can possibly think of, you can pretty much find. But those are compiled lists, they're compiled from some sort of database somewhere, and they're not mail-order buyers.

Response Lists

The second kind of list is called a response list. And it is just what it says. These are people who have responded to some sort of direct-response medium. It could be a TV infomercial, mailing or a catalog that someone responded to and subsequently made a purchase from.

You can get data about these people sorted out in many, many different ways. If you want to sell something to people who like to fish, then find a catalog of fishing stuff, and you could probably rent that list, which has people who have bought what you're trying to sell.

So, how do you know when to use each kind of list?

When you want to sell something by direct mail, you'll do a lot better with a response list. Compiled lists are unlikely to do anywhere near as well.

On the other hand, if you're doing a lead generation mailing, you could use compiled lists or response lists, but if you need a narrow target market, you might find a compiled list to be better because it'll be more complete.

For example, if you're trying to sell to attorneys in Denver, you might be able to find a list of attorneys who bought something by mail, but you might get less that 10 percent of Denver's attorneys that way.

With a compiled list, you can get all of them. If what you're trying to do is generate leads so that a salesperson can call on them, then you might find that you do better with the compiled list.

Another example where a compiled list is going to work best is for retail stores. If a retail store is trying to do a promotion and they want to reach the people within a few miles of the store, then it's very common to do a mailing to all the households within, say, three miles of the store.

Or, if you have a specialty store of some sort, then target those people who live within the distance shoppers come from to get to your store. That would be, again, a compiled list.

An Example Of Selling By Mail

Let's say you're selling an expensive putter, for golfers, and you want to sell them by mail.

You could get a compiled list of sales managers, figuring that sales managers make more than sales people, that maybe, sales managers or vice presidents of sales would be a good target, because they presumably earn more money.

You also figure that most of these sales managers are out playing golf all the time, so they might be great prospects.

Well, that might sound good initially, and if you're talking to someone who's trying to sell you a compiled list, that might be exactly what they recommend.

However, what if you got a list of the subscribers to, say, Golf magazine? Now you're not guessing whether they play golf or not. You know that, not only do they play golf, but they also like to play golf enough so that they bought a magazine about playing golf.

And they bought that magazine by mail. That means they're not in the 30 percent of the people in the United States who never buy by mail. So the Golf Magazine subscriber list would probably work way, way better than that compiled list of sales managers.

But let's take it a step further.

How about if we could find a list of people who bought a golf product by mail from some other company? Now, that's even better than the magazine. If you're selling a golf product by mail and can find a list of people who've already bought a golf product by mail, then that's even better.

But, we can get a step better than that. How about if they bought that golf product last month? Now, they're recent buyers of a golf product, and they're even more likely to buy from you.

And lastly: how about if the product they bought last month cost more than $200? Since your product is $200, if you can find somebody who bought a similar product that cost that much or more money, recently, then imagine how much better a response you'll get from that list compared to your list of sales managers.

It's a huge difference. You could easily have 10 times the response going from one extreme to the other, which can easily be the difference between your mailing working or not working!

You can find these response lists in the Direct Mail List Source book mentioned earlier by Standard Rate & Data (www.srds.com).

Now that you are far enough along to consider testing, how do you evaluate which lists to test from the 32,000+ lists available? Keep in mind that even if you have the best headline, the best copy and an irresistible offer,

The first thing to keep in mind is not to believe the rate card that gives you information about the list. A rate card is a sales document, and many people who don't know much about lists or about renting lists, look at that and that's all they read. They make a decision about whether to rent the list, or not, based on the rate card.

You're not finding out what you need to know. Furthermore, some of the things on the rate card may just be flat out lies. They might say they have 50,000 mail-order buyers, when in reality, they only have 10,000 mail-order buyers and 32,000 who inquired, but never bought.

If you don't know the right questions to ask, then you'll get taken, so here are the 15 questions to ask before renting a response list:

What Is The Size Of The List And The Cost-Per-Thousand? The reason this matters is you want to know how many names are there to roll out to if you do a test and it works. Think about it. If you have a list of 10,000 names, and you're going to test 5,000, the typical minimum order, your only rollout is the other 5,000.

Is It Related To Your Offer? Are the customers on your rented list buying something similar to what you're selling? Don't try to sell steaks to vegetarians. Make sure it makes sense.

How Recent Are The Names? This is called "recency," meaning how recently have the people bought? This factor is so important that, normally, the lists are categorized by how old the names are.

Names can get tremendously out-of-date, and the response can drop in half after they become a year old. So that's why they categorize the list by how old segments of it are.

If you ever hear the term "RFM," that means recency, frequency and monetary value. Many direct marketers consider those three the most important criteria.

What Did They Spend (Monetary Value?) Make sure if you are trying to sell something that you don't buy a list of contest entrants who didn't buy anything but are considered a response list because they responded to a direct response offer.

If you are trying to sell something, a list of buyers of that thing is better than a list of people who entered a contest to get it for free! Don't rent a list of buyers who spent $20 if your product sells for $300.

Are The Names Inquirers Or Buyers? If all the names are lumped together to make it appear that they're all buyers, then you need to ask about this. Make sure they really are buyers, not inquirers. Odds are, there are some of each and that's okay. You're definitely going to want to test with the buyers though, not with the inquirers.

Where Did The Names Come From? Are the names direct-mail generated, or did they come from another source? What you want to buy here is a response list, meaning it was direct-response generated. If the names are not all direct-response generated, then the list is not as good.

In addition to wanting to know where the names came from, you want to see a sample of the mailing piece, or marketing that was used to collect the names.

This is important because even if the names were direct response generated, you want to know if it was a free sweepstakes they responded to or a catalog where they bought something.

Are They Multi-Buyers? Multiple buyers are people who have bought more than once. The list owner may have multi-buyers, but can you select by multi-buyers? They may not have that many of them, so they may not be segmented out.

If they can be segmented out, then you want to buy the most recent names that bought the amount you are selling, and also bought more than once.

Test the multi-buyers first, because if the best people you can test don't work, then the others aren't going to work either.

What Selections (Or Sorts) Are Available? Depending on the size of the list, the selections, or sorts, may vary. Find out what selections they allow, like the person's job title, income or salary, other demographic or geographic criteria. The rate card may not even show all of the available selections or sorts, so ask.

Do they have psychographic (lifestyle factor) sorts? If your target market can be better targeted using things like religion, politics, hobbies, things people like to do.

Does The List Have A High Duplication Rate With Your House List? That means, if you already have a customer list and you do what they call a merge-purge, which is to combine two lists and see how many duplicates there are, is the amount of duplicates high?

You might think that that might be a bad thing. But it isn't, it's a good thing.

Here's why.

If there's a lot of duplication between your customer list and the list you're renting, that's a good indication that the list you're renting is the same type of person as your customer list, and that's great. In fact, if you have a choice of three or four lists, then I'd pick the one that had the highest duplication with my customer list, and test that one first.

Does The List Have A History Of Repeat Mailers? Who else has rented this list? Ask them. Who are the mailers who have rented this list? Did they test it? And then, did they roll out their tests? Did they then do it again? Did they repeat mail? Find out. Call other mailers and ask them how the list worked for them.

Do Your Competitors Rent The List?There again, unlike what you might think, that's a good thing. If your competitors have rented it, and especially if they have rented it more than once, then that's a sign it works. If it works for them, then it can work for you.

If The List Is A Buyers List, How Did People Pay? Did they pay up front with a credit card? Or was it one of those offers where you pay nothing until next year? Obviously, your better prospects are the people who just put it on a credit card, as opposed to people who aren't going to pay for two years, and then maybe default then.

What Is The Rollout Potential Of The List? I mentioned this before, referring to a test of 5,000 names on a list of 10,000 leaving only a 5,000-name rollout. Meanwhile, a test of 5,000 names on a list of 100,000 has a 95,000-name rollout if the test works. All things being equal, the one with the larger rollout potential is the better list.

When Was The Last Time The List Was Cleaned? "Cleaning" means database updating. The statistics are that one-fifth of the population moves each year. Which means if the list is a year old and hasn't been cleaned, 20 percent of the addresses will be outdated and not deliverable.

In the case of business addresses, especially with individuals at business addresses, 50 percent of that list can be obsolete after only a year. And that's not so much the businesses moving, but the people moving.

What Formats Are Available? There are many different formats for lists. You can get them on magnetic tape, disks, labels, labels with adhesive, or labels without adhesive, called "Cheshire labels."

One of the formats that you need to be careful with is when you're going to do a mail merge. You don't want the list to come in all caps, the post office-approved format for automation mailings.

You're going to want to have upper and lower case lettering. Why? Because if you're doing mail merge and the person's name is printed in all capital letters in the middle of the letter, then it will look really obvious that it's not a personal letter.

You'll lose the whole effect. So, if you intend to do any personalization, make sure the list can be sent in upper and lower case.

In a list company, the sales people who are selling you the list are frequently paid by how much they sell. This can be a big problem. Since their incentive is to sell more names, they may not give you the random test you asked for on your order, but a bunch of hot line names instead. They figure you'll come back for a big reorder then and they'll make more money than if your test didn't work and they got no reorder.

Aside from being unethical, this is bad because if you base your rollout on the data from these names, you'll get killed!

The most common reason a rollout of a successful test doesn't match the test results is because when you asked the list company to give you a random sample of a list to test, they gave you hotline names instead. Even if you asked for a random sample or what they refer to as an "nth name select", meaning, if you want 5,000 names from a 50,000-name list, the nth name means they give you every 10th name.

The problem is that you cannot tell if they did it right or not. How are you going to tell by looking at the list whether it's an "nth name select" list or whether it's hotline names? There's no way in the world you can know.

So now, what are you going to do? To keep from being ripped off, you have to be able to tell whether the list company did what they were supposed to do, or the nth name selection isn't going to do you any good.

So how can you tell?

A direct mail guru I know, Marty Chenard, came up with the idea to look at the first letter distribution of people's last names. When I did this with the Denver White Pages, I went through and counted the number of pages where people's last names start with each letter.

There were 1,685 pages in the White Pages, and I counted them all up by what page each letter started and ended on, and, in Denver anyway, the letter "S" is 10.44 percent of the total names. The letter "X" is .01 percent.

So what you do is, instead of asking for nth name or random selection, you pick the percentage that matches the letter, and ask for all the names starting with that letter. Now it's easy to tell if they do it. If you asked for everyone with a last name starting with "P," then just look at the list and you can tell.

Let's say you have a 150,000 name list and you want to test 5,000 of those. When you do the math that comes out to be 3.33 %. When you look at your chart below, you'll see that the letter "F" represents 3.38 % of the total.

So now, instead of asking for an nth name or a random selection, you ask them to give you everybody whose last name starts with "F". This should still be random, but guess what you can do now? You can just glance at your list, and if everybody's name doesn't start with "F", then they did it wrong.

So this is a way to make 100% sure they don't mislead you with hotline names. If you need more than 10.4% for your sample, add up whatever letters total the percent you need and order your names that way.

Distribution Of Population In Denver By First Letter Of Last Name

You may have heard that the post office has a history of throwing out bulk mail. Needless to say, they're not allowed to do that. It's the mail carriers who do it, and some of them, when interviewed, say that they only throw out mail on days when there's a lot of it, which means of course that they throw out a lot of mail.

There have been a number of studies done by the Direct Marketing Association, the Post Office, and mailers that show that between 10 and 30 percent of all properly addressed bulk mail never gets delivered.

One study that was done some years back by the Postal Service showed that properly addressed bulk mail was destroyed at all 111 Post Offices and 12 bulk mail centers that they checked.

That was 10 to 15 years ago, so maybe now it's better? But I can give you an example from just a few months ago.

I have a client I'm doing some mailings for, and I was doing a test of a big 6" x 11" color postcard. If sent bulk they get a good price, but if they go by first class, you pay the full postage. So I mailed them by bulk.

You know what happened? I had four or five different segments in my list and this is a promotion for an oil change store that has a North store and a South store in the Denver area.

Everything we'd done before this, the response rate at the North store was about twice the rate of that at the South store. The South store had a lot more competition, and basically, anything done at the North store worked better than anything done at the South.

In this mailing, I sent out equal quantities for both stores of each variable being tested. I believe it was 1,000 pieces around the North store and also around the South store. We got 15 new customers for the South store, but the North store got zero.

This leads me to believe that since the North store has always gotten a better response on everything else we've done, and since we got 15 new customers for the South store and none from the North store, that this mailing never got delivered. I'm 99 percent sure it got thrown out, because it makes no sense for the response rates to be that way. (If the North store had even one response, I'd have known the mailing bombed but at least it got delivered!).

So what do you have to do to keep your mail from being thrown out? You shouldn't have to do anything, but the reality of it is that mail carriers do throw out mail.

The best thing you can do, aside from mailing first class, is to make it look as much like first class as possible. In other words, trick the mail carrier into thinking that it's first class mail, not bulk mail.

Some of the ways to do this are to use "live" stamp, not a meter imprint or a postage meter. Even if you're doing bulk mailing, they do have bulk mail stamps. You will have to pay extra to have the stamp put on, but mailing companies have machines to do it so it shouldn't be much, but it's worth it.

Pick a bulk stamp that has the words "bulk rate" as small as possible so it's not so obvious.

Another thing mailers do is use headlines or teaser copy on the outside of the envelope to help get people to open it. This makes it obvious it's a commercial mailing so I don't recommend it unless you are an expert. We're not talking about getting the envelope opened right now; we're talking about trying to get the envelope delivered.

They Can't Open It If They Don't Get It

So making it look like a first class letter is a good idea. Don't let the lettershop use bar codes, nine-digit zip codes or use the CASS certification criteria or anything else that keeps it from looking like a personal letter. Make sure the address is printed right on the envelope, not on a label.

The best choice, if you can afford, it is to hand-write the address on the envelopes. I suspect, though I haven't tested it, that if you sent it bulk mail using a bulk mail stamp, and hand-addressed the envelope, it might get delivered as first class.

If you test this, do let me know what happens!

I just did an envelope test with typed addresses vs. handwritten addresses and it's running 2 to 1 so far — well worth the extra cost. Like everything in direct marketing, the best thing to do is test for your own product and market.

Gary Halbert — one of the best living copywriters — points out that people sort their mail over a wastebasket. He calls it the "A" pile and the "B" pile with the "B" pile being the wastebasket.

So, your first job is to get in the "A" pile and the best way to do that is to look like personal mail. But people are wise to the gimmicks I for one, look at the stamp and if it says "standard or bulk" rate, in the trash it goes.

So make your mail look like a personal letter. Don't let a lettershop put indicia on it, use a postage meter or use all caps letters to get a postage discount. You can test all those things later if your mailing works, but your first job is to get it to work.

In any testing, give it your best shot right away and ignore the costs. If you can't get it to work doing everything possible to increase response, it's sure not going to work when you start mailing bulk.

I met a real estate attorney who wanted my help but was totally convinced direct mail to people in foreclosure would not work because he sent out 2,500 letters and didn't get one call.

I convinced him to let me try and my first mailing got him 11 people to call from only 280 letters. Needless to say he's very excited and instead of getting a client we formed a real estate limited partnership as 50/50 partners. We've had to get 2 more people to do deals because of all the calls coming in.

The mailing works both because of what it says and how it says it, but also because it gets opened. Here's what I do. Imagine seeing this in your mail and tell me if you would open it!

The envelope is a 6" x 9" manila envelope. It has a very large commemorative stamp on it. It's addressed by hand with a felt tip pen. The return address gives street, city, state with no name. Does that look like a personal letter or what?

So, when it's opened, you don't want to trick the recipient — if what fell out was a stack of color brochures, I'd throw it in the trash because I'd be mad I was tricked into opening it.

So when they open my personalized envelope what do they find inside? A personal letter! With their name in the headline!

Before you start testing first make sure you have a way to track the results. It's important to track all the way through to the sale and not just leads because they don't all necessarily convert at the same rate.

I had a project 5 years ago where I was advertising in 10 national magazines. My highest cost per lead was Entrepreneur magazine but since I tracked all the way through to the sale, I found out that Entrepreneur was the cheapest cost per sale because those leads converted so much better.

Any time you're doing testing, especially tests of multiple lists and multiple variables, you'll need a code for each variable.

Every combination of variables needs a code. If you have three lists and two offers to test all at once, that means you have six cells, and you need to have a code to track each of the six.

If you're doing mailings where the recipients are mailing/faxing in their orders, then you can put, as they call it, a "key code" on the label that gets returned on the order form so you know which code counts for which sale.

On the other hand, if people are calling in orders, or are responding by phone, then whoever's answering the phone has to be trained to ask for those codes. If you ever order anything by mail, you've probably seen these on the mailing labels of almost any mail order catalog and been asked for the code on the mailing label when you called in your order.

In my mail-order company, we had both. We had orders come in the mail, by fax and by phone. When you get phone orders, that's where you can lose your codes.

The way I handled this was to have an objective for the people answering the phone to get 95 percent or better of every sale attributed to a key code. And every month, at the end of the month, I'd look at a report called "sales by key code for the month." I didn't want the sales in the "unknown" category to be any more than five percent of the total sales for the month.

There are ways besides using codes to do tracking. You can track by using different prices. I'm not talking about testing the price, but I'm talking about taking a product that's $24.95, and creating a different price for each variable you're testing. Make the prices $24.95, $24.96, $24.97, etc.

Vary it by a penny, and then the tracking is automatic and doesn't involve people remembering to ask for codes. Wherever possible, I like to be able to track things in a way that doesn't involve people remembering to do something!

You can also track if you're giving away a free report or you're doing a two-step program of some sort. You can change the report numbers. You can say, call for Report #104, #105 or #106. Those can all be the same report, but the report numbers can be your tracking numbers. Tell people that to get your free report they have to ask for it by number, so that we send them the right one.

If You Have A Web Site.

Another concern with tracking is if you have a web site. If you put your URL in your ads and mailings to give them the option of going to the web site, then you'll lose your tracking.

What if 60 percent of the people go to the website as a result of an ad or mailing and order or inquire there? You'll have no clue where they came from. This can cause the stats on your ad or mailing to be so understated that you might conclude that your promotion was a flop, when it really worked just fine!

Here's how to fix that.

Don't use your main URL in your ad. Don't use your URL with extra letters either, because everyone knows to go to the root domain. In other words, for my site
www.joemcvoy.com, if I had a URL in a newspaper ad using a web address of www.joemcvoy.com/newspaper, then people would type www.joemcvoy.com and not the rest. I know. I do that all the time.

There are two ways to fix this:

Set up another domain just for tracking. I could use josephmcvoy.com instead of joemcvoy.com and then add the code like this: www.josephmcvoy.com/10.

I set it up so that if they go to josephmcvoy.com without the following numbers, they get an error message telling them to enter the full URL. When they enter it correctly it goes through my ad tracker and then to joemcvoy.com.

The other way to do this is to have a free redirect service like www.tinyurl.com. You go to that site, enter your ad tracking code and it will shorten it and make it unique like www.tinyurl.com/dx6448.

The easiest way to do this web tracking is to set up an ad tracker code for each ad. The service I recommend is www.1shoppingcart.com.

Since ad tracking codes can be 60 – 70 digits long, or longer, I also use the tinyurl.com site to shorten them for any off-Internet advertising. When you are tracking a pay-per-click term, a banner ad or an online link, it doesn't matter if they are long because no one sees them anyway.

The service I use allows an unlimited number of ad codes and lets you send out broadcast emails and also unlimited autoresponder messages. As of now, it only costs $29 per month.

Now it's time to cover testing. The number one thing is to remember to test the big things first. The big things being: the lists, your offer, your product, your headlines on the sales letter and maybe the time of year. Little things, like the color of the order form, or the things that are minor, you can test later.

When you're doing testing, it's the simplest if you just test one variable at a time. That variable could be the list or any other thing; it could be your offer, your product, anything.

It's also possible to do what they call grid tests, where you test multiple variables. If you're going to do that, then you need to make sure that you understand the reliability of the results and how many responses you need for each cell to conclude anything.

The next thing you need to know is how to make sure the test quantity you're using gives you reliable results. This depends not on the number you mail, but on the number of responses. One good rule of thumb is that, if you get 100 responses, that's a good, reliable test.

On a list of 5,000 names, at a two-percent response rate, that would give you the 100 responses. Which is why you'll frequently hear 5,000 used as the test quantity.

The statistics can be pretty complicated. But, there's a great free calculator under the "Resources" section of the Edith Roman website at www.edithroman.com. Since rollouts typically do worse than the initial test, this calculator will tell you the confidence level and the amount of decline on a rollout.

You choose the confidence level you want, 75%, 85%, 95% and then the % drop off on your rollout that would be OK, and the calculator tells you how many responses you need from your test to give you the confidence level you picked.

Here's a printed version of the chart, showing a few standard choices, so you can see how it works.

Chart To Test Quality

Here are some examples to show you how this works:

The number of responses you need for statistically accurate results depends on the confidence level you want and the percentage of decline on your rollout you can accept. The number of pieces you mail is irrelevant. It's the number of responses that count.

If you want to set up your test so that when you do a rollout, you want a 90% confidence level that the response to your rollout mailing will not drop off (decline) more than 25%, then you need 26.2 responses when you test.

This means that 9 times out of 10, your rollout will not be more than 25% worse response than your test. If your test got you a 2% response, then that means that you can be 90% certain that the response rate on your rollout to the rest of the list will be no worse than 1.5% (25% less than 2.0%).

If you want to be more certain, or don't want that much drop off, then you need more responses. Let's say you want to be 99% certain that you don't have more than a 25% drop off. Now you need 86.9 responses.

The chart tells you how many responses you need. To calculate how many pieces to mail, either use the Edith Roman calculator at www.edithroman.com, or do it yourself. All you need to do is divide the number of responses by your expected response rate of your mailing.

If you expect a 2% response and want to be 99% certain you don't have more than a 25% drop off, then divide the 86.9 by .02 (percent response expressed as a decimal) and you get 4,345 pieces. If you only expect a 1% response, then it would double to 8,690.

Another example, if you are OK with a 85% confidence level that your rollout will not drop off more than 25%, then from the table, you can see you only need 14 responses.

At a 2% expected response rate, you'll only need a mailing of 700 pieces. As you can see, the more confident you want to be, the bigger test you need.

You can be pretty much assured that the response rate on your rollouts will drop from your tests. In all the mailings I ever did in my mail-order company, I never had a rollout do better; they every single time did worse, sometimes quite a lot worse.

Another thing to consider when you're doing testing is the season. You want to really make sure that your mail doesn't get delivered right before a three-day weekend or a holiday.

Mailing Seasonality - By Month

A number of mail order books have published a mailing seasonality by month for consumer offers to help guide you in adjusting your response rates to seasonality.

The first point to make is that you need to develop your own chart for your own products. This one is a compilation of many mailers results for non-seasonal consumer products. If you have a product with seasonality like a Christmas product, your chart will look very different.

Mailing Seasonality Response Chart

Some explanation of these numbers will be helpful. They are based on calling the typical results in October as the baseline—or 100%. The other months are better if higher and worse if lower.

For example, if you did a test in October and got a 2.0% response and then you did a rollout or another test in November, you could expect a 10% drop to 1.8% based on seasonality alone.

Likewise, if you tested in November and got a 2.0% response and then rolled out in January, you could easily get a 23.4% better result—2.0% to 2.47% because 111.2 is 23.4% greater than 90.1 (111.2/90.1 = 1.234).

The Best Day To Have Your Mail Delivered

The ideal day to have your mail delivered is a Tuesday because everybody gets a lot of mail on Monday that accumulated over the weekend. Since Tuesday is typically the slowest mail day, shoot for that.

How To Calculate The Half-Life Of A Mailing—And Why It Matters

The half-life of a mailing is the point at which you have half of your responses. If you put your mailing on a chart, you'll have a big peak at the beginning, and then it'll taper off over time. You might still receive a response, or two, four months from now.

When you know the point where half your responses have come in (the half-life), you can make a decision, roll out, and test again doing four or five times as many mailings in the same period of time as someone who doesn't know this and has to wait for all their responses to know how the mailing did.

This can make a huge difference to your profits and lessen seasonal differences if you can make decisions in weeks instead of months.

In my mail order business, I knew that when a mailing had a week sales drop off from the prior week, I could take the prior week's sales and everything before that, add them up, and that was roughly half that that I was going to get — or the half-life. That meant I knew the results of my mailing in two weeks even though a response or two might still come in up to four months later.

Grid Testing

Grid testing is used to test multiple variables at once. All of the variables are put into a chart format, with each cell getting its own mailing code. This can get quite complicated, and if you want it to be statistically significant, then you also need to have enough responses in each of those cells.

I've seen charts with 50 or more variables. This is not really practical for small businesses because large mailers can do this when they mail in the millions and get good results.

You can, however, just do it anyway, and do maybe two variables, three or four lists, two different offers, or something like that, to give you six or eight cells. It may not be statistically significant, but it can give you a direction for further testing quicker than you would otherwise get it.

I just did that for a client, and I got a pretty good hint. I tested three different outside envelopes and two lists all at once.

I've done this a couple times, and even though the results were not statistically significant, when one is 4 or 5 times better than the others, it sure gives you a hint as to which variables to test in a larger test.

If you don't already have a "control package," or a standard mailing that you're trying to beat, or if you're just starting from nowhere, then try a grid test.

Pick maybe 5 lists, include 2 or 3 offers, and maybe a couple different sets of copy. This will involve a lot of cells that you will need to assign codes for and track, but you will probably find a few cells that do dramatically better than all of the others.

You can then further test those cells for statistically significant results before you do your rollout. This lets you test 30 or 32 cells. Just understand that you don't have statistically significant results, but a test like this can give you an idea of which things to test next to get reliable results.

You can get to the next step a lot faster this way than if you just did one variable at a time.

Here's one other point on testing. I've probably done over 300 mailings in my own business and for clients. I've never had to pull out a statistical chart to analyze a mailing.

I've had mailings get as high as a 30 percent response, when I needed one percent to break-even, and then, I've also done a mailing of 20,000 pieces, and got four orders—essentially zero percent response.

Every mailing I've done has always been really obvious whether it succeeded or not. You may not ever need to get into statistics, but, if you do, use the chart at the Edith Roman website and don't bother getting involved with complicated statistics.

Special Advanced Testing Note—Taguchi Method

Recently a new technology for marketing testing called the Taguchi method has been adapted to testing marketing. This type of testing has been used in engineering for complicated modeling and has now been adapted for testing both direct mail and "pay-per-click" internet marketing.

So far only 30 people have been exposed to the details at a new seminar and I have a friend who attended and am going to be in the beta testing group for the Internet pay-per-click models.

One of the first companies to use this technology was Dell computers and they saw dramatic improvements in response as a result.

This method uses a number of different matrices that test dozens of factors and tens of thousand of permutations and combinations of all the variables all at once, resulting in quickly finding the optimized combination of al the factors.

I have the grids and will be running my first Taguchi test shortly on line and with a direct mail promotion—stay tuned for the results—this is going to revolutionize marketing testing.

I like using the Internet for testing because you can get results so quickly. I set up tests on www.surveymonkey.com for free and then send out email surveys asking people to take the survey.

If you have an email list of customers or prospects that are in your target market, you can get test results in a matter of hours. If you don't, you can always drive traffic to a survey through Google or Overture PPC (pay-per-click) campaigns.

There are also on line companies who will rent you databases for survey purposes—not free but at least quick. All these options are a lot faster and a lot cheaper than sending out surveys by mail.

I hope you have found this information to be helpful and if you want help with a direct mail campaign on-line or off, give me a call.

For More Information Contact:

Joe McVoy
Profitable Marketing Systems, LLC
1100 Nautilus Court
Lafayette, CO 80026
Web: www.ProfitableMarketingSystems.com
Email: Joe@ProfitableMarketingSystems.com
Phone: (720) 890-8760
Fax: (303) 604-6839

You are free to copy this article to use on your web site or blog as long as you make no changes and keep contact info including web site and email address intact.

© 2006 Profitable Marketing Systems, LLC