Here are some of the highlights of topics you’ll learn about in this blog post on performance marketing tests:
- Why you should have a well-defined objective before doing any A/B split-test with your landing pages, Google Ads, Facebook ads, banners, emails, direct mail pieces, inserts (or any other split-testable medium).
- How you should implement changes indicated by test results.
- What are the strict test guidelines to follow.
- How to establish a “control”.
- How to determine test size.
- How to determine what to test and what not to test, ever.
- How to add new life to an old control promo.
The best way to achieve control and success is by establishing carefully defined testing and record keeping procedures. Also, one of the most important virtues you can bring to performance marketing is to have patience.
Tests of your landing pages, Google Ads, Facebook ads, banners, emails, direct mail pieces, free standing or package inserts, IVRs, SMS texts, telemarketing scripts or even magazine ads are critical to the success of any ongoing performance marketing program.
Yet surprisingly, few companies have a regular testing program, and even among those that do, a substantial number fail to establish firm controls and keep adequate records.
The lack of adequate controls and records makes objective evaluation of test results impossible and organizational knowledge skewed or lost forever.
Test but in a structured way or don’t test at all
The reason for testing is simple. It is the only sure way to determine whether you’re doing things in the best way or whether you can improve results.
In short, testing is the best way to learn in the marketplace. It might even be called a marketing study. One thing is certain: testing is more accurate than conducting “focus group” interviews or other marketing surveys that are largely uncertain due to their declarative, not behavioral, nature.
Tests are usually more reliable and, in many instances, less expensive. Performance marketing permits testing the effectiveness of nearly every element of your promo piece.
Nothing else offers equally reliable market segmentation or results.
When a test is suggested prior to a rollout campaign, someone may protest, “We don’t have time!”
If time is never taken to conduct systematic tests of a program, the company will run out of time and money because it will fail to achieve the desired sales objectives over time. A company cannot afford not to test!
There are many aspects to designing and executing an A/B test.
What Is the Test Objective?
If there is no clear-cut objective, don’t conduct the test. A test objective should summarize precisely what the test is supposed to accomplish.
- Why Test? For example, your objective may be to improve results by rewriting the lead on your landing page, a sales letter or telemarketing script (it’s called opening statement in telemarketing). Or perhaps you think the headline is not strong enough in your package insert, or the descriptive copy used in the direct mail piece don’t adequately describe the product and the benefits it offers to the buyer. For any of these reasons and many more it is desirable to A/B test before making arbitrary decisions to change without testing.
- Measure Results. Of course the first thing you must have is a control piece or ad that will provide a benchmark from which to measure the results of any A/B test conducted. If none has been established you must have more than one ad or landing page so that you can measure one against the other.
- What Is the Objective of the Test? Once the objective has been decided, the test unit can be modified to accomplish the test objective. Without such an objective, when the results are being analyzed later, questions will be asked such as:
- “What were we testing?”
- “Why did we test that?”
- “Who authorized that?”
- “What did we learn from the test?”
- “Have we made corrections in our control landing page to reflect the information learned from the test?”
(Hopefully, many of these questions will be asked anyway. But it isn’t uncommon to see people order A/B tests performed without a clear-cut objective. Then when the results are summarized, you discover no one was sure about the purpose of the test.)
- Define Test Response Needed. Usually the A/B test objective is to increase response and profitability. (Of course, tests may be to improve the visual appearance of your mailing package, to seek a broader reach into the market by changing the mix of audiences used or toward other objectives like lower refunds.) With this objective in mind, projections should be made to determine how many orders will have to be received to break even and make money utilizing the proposed test package.
Write an Objective Test Summary
When all test results have been calculated you should write a very objective test summary. The summary should answer the question posed by the test objective in a short, unequivocal statement:
- Did the test fulfill the test objective?
- If so, how?
- If not, why not?
- What, if any, information achieved by the test should be utilized to improve or correct the current control landing page, ad, insert or operating procedures?
- Have the appropriate persons been assigned the responsibility of implementing the changes needed to comply with the information developed by the A/B test?
A/B Test Summaries
This summary should then become a permanent part of company A/B test records for future reference.
- Take Action. The test summary should be a catalyst for taking action to correct any deficiencies in the current control. Too often, tests are conducted and no action is taken to take advantage of the information gained.
- Assign Responsibility. Management needs to assign a specific person to promptly review the A/B test summary. The marketing team should assign specific persons to take whatever action is indicated by the test results.
- Establish Deadlines. Firm deadlines for implementing changes or corrections should be established with appropriate assignments, in writing, made to those responsible for fulfilling each assigned task.
- Follow-up Action. If any definitive results are obtained someone should be assigned the task of following up to implement the newly developed A/B test information. That individual should make sure that subsequent campaigns incorporate the newly developed information.
Do Not Include Start-Up Costs
When these projections are made, do not include the creative and other start-up costs associated with the test unit. To do so will place an unfair burden on the test unit.
Test quantities may cost more. For some performance marketing channels, especially offline ones (i.e. inserts, direct mail pieces) or in some companies where cost allocation systems are oversimplified (i.e. the cost of designing the landing page is divided by the orders or visits generated during the test), costs for the test unit will be more than for the control because the quantities of test pieces will be smaller. Therefore, use prices quoted on comparable quantities to the control unit to obtain an accurate projection.
What Is a Control?
- Control. A control direct mail package, preprinted insert for space advertising, telemarketing script or a landing page is the one that is working best and against which any new mailings, inserts, landing pages, telemarketing scripts will be tested. The control in any of these media will remain control only as long as it outperforms any item tested against it.
- Establishing Control. If you have not designated any current landing page or ad as the control, do so immediately. Select the one in each medium that is working best. If you do not have a control and are starting out new, A/B test at least three totally new concepts.
Let’s assume that your company has never conducted any tests but has an ongoing traffic generation program directed to specific landing pages, has some newspaper inserts and telemarketing campaigns with word-for-word scripting. Some rudimentary records have been kept on responses and sales that have been achieved by each ad, phone script, insert and landing page.
But, just in case your company hasn’t, choose the best performer (if known) from each medium currently used. If the company has not kept careful records on the performance of each by each ad, phone script, insert and landing page, START NOW!
Call those selected pieces “control. ” Henceforth, any changes in the landing page, ad or insert must be tested against the control in each medium. The control should remain control only until a better performing piece is developed.
Take Four Steps Before Changing Control
After a new material outperforms the control on a larger test, use the following procedure before replacing the control.
Despite large, statistically valid samples some A/B tests may be skewed due to a number of reasons (seasonality, different sources of traffic, weather etc.).
This is why some big performance marketing players will do a three- or four-step confirming campaigns before replacing the control with a new landing page, mailing etc.
So if for example you’re into direct mail or email campaigns and have a huge database of prospects/customers take these steps:
Step 1. Mail a test mailing of 10 to 20 percent of total mailing against control.
Step 2. After summarizing results and if test unit outperforms the control, increase next test unit to 50 percent winning test unit, 50 percent old control.
Step 3. Again, after response has been measured and it still shows the test unit winning, increase test unit to 90 per-cent of the mailing with the old control at 10 percent. If the results remain positive for the test unit, it now becomes the control until a better package can be created. The total number of pieces in each drop is increased as confidence is gained in the new test unit.
Step 4. Even after step 3, some companies will mail another 10 percent of the old control with the rollout to reconfirm that this was the correct decision to make.
Selecting a Control: Common Mistakes
If you have just established a performance marketing program your first choice of the test promotional material (i.e. email or landing page) used will probably be to choose the one that produces the most “up-front” response.
This is what pretty much everyone does and most of the time that may have been the right choice, but not always. That choice could be the undoing of a company.
- The new landing page having the most up-front respondents may have attracted the most deadbeats.
- The new landing page may have attracted a large number of non-converters to the regular offer (people who, for any number of reasons, never purchase the advertised item). They also return goods after looking.
If time permits, it is best to evaluate test material before choosing one as the control and subsequently doing a large rollout.
Sometimes you will find that the email, direct mail piece, insert, landing page etc. that had a lower front-end response rate will actually produce more net profit because the customer converts better, upsells better and offers fewer credit risks.
You may find this to be an overly conservative approach. It is conservative, but it also avoids making very expensive mistakes by switching from one creative to another before positively confirming the improvements made with the new test unit.
Review Orders Before Doing a Rollout
All orders / leads should be reviewed before a rollout especially if your business model involves immediate upselling, check payment, cash on delivery payment, high refund rate, is a multistep model or is anything else than a straight sell with a credit card.
For example if you’re into bill me later schemes you should review credit collection records especially on “bargain” offers. After having reviewed the collection records, your company may find it expedient to accept a larger percentage of bad credit and no-pay customers if the total volume of good business justifies the additional risks.
How Large Should the Test Be?
Let’s imagine for a while that your company sends its prospects and customers a ton of SMS texts every month.
If your company already has a control SMS, your test unit can be a relatively small quantity in comparison with the quantities regularly texted with the control one.
- Select an Appropriate Test Quantity. If the company normally makes small campaigns of 25,000, a suitable test unit could be 5,000. Companies sending texts in hundreds of thousands or more normally should use test units of 25,000 to 50,000 pieces. That’s a rule of thumb and a tip that you DON’T have to have your tests take a half of your normal campaign volume. Make sure though you have a right sample size selected (that’s the next point).
- Quantity Is Important. It is important to use a test quanti-ty sufficient to provide a statistically valid reading. If the normal response rate for the product is very low, a larger test unit should be chosen. The higher the normal response rate per thousand, the smaller the quantity of the test unit needed to achieve statistically valid readings. The higher minimum detectable change between conversion rate you’d like to see, the small the test quantity.
The good thing is you don’t have to be rocket scientist to conduct proper, statistically valid tests. Just use on of the free online sample size calculators (for example the one by Optimizely).
Example: Suppose you have a on-going SMS text campaigns selling baby food subscriptions that are targeted to young mothers. Let’s assume the usual response rate is 2.0 percent.
Suppose you’d like to have a test in which you introduce a higher price (to earn more per one prospective buyer) or change the value proposition (to increase conversions).
These are your options for 2.0 percent conversion rate:
|Minimum detectable change between the control and the test unit
|Your sample size
|57,000 texts for the test unit and 57,000 texts for the control
|13,000 texts for the test unit and 13,000 texts for the control
|5,300 texts for the test unit and 5,300 texts for the control
What this means: if you opt for the 30 percent increase in conversion (i.e. from 2,0 percent to 2,6 percent) and choose 5,300 sample size BUT all you achieve is a 10 percent increase in conversion (i.e. from 2,0 percent to 2,2 percent) your test will be completely invalid.
- Do Clean Testing. In most cases, keep tests “clean,” especially if you already have a reasonably successful control to test against. Test one significant element at a time; everything else in the test units should remain the same as in the control.
- Test Significant Items Only. Limit tests to elements that can significantly increase response. Avoid major changes in direction until the more obvious elements have been tested unless the initial results were so bad that a major change in strategy is needed. For example, opt for copy, offer, price, audience and landing page design tests before testing the color of call-to-action buttons and envelopes, or a new bio of your boss (actually never test colors or bios unless you have extremely huge amounts of traffic like Google or Facebook themselves). Always remember, audience tests usually provide the best chance to increase response rates.
- Schedule Timely Tests. Tests ideally should be conducted in low response periods so that the results can be read in time to make any alterations to the control package prior to heavy campaign season. This permits you to gain maximum advantage from any significant new finding of the tests.
This guideline runs counter to suggestions by some that you test only in the best period. The best period for campaigns is the best time to reconfirm that the new promo is the better before doing a full scale rollout.
- Test Reliable Quantities. If the test is too small and produces a statistically insignificant quantity of orders, you have created a problem instead of solving it. It isn’t prudent to make a major marketing decision based on a difference of two or three orders.
- Test Package. The test package should be identical to the control package, except the single element being tested. That element may be only a single line of copy or the whole letter; a different reply envelope or outside carrier envelope; a new brochure or no brochure; a copy or art change on the landing page, a change in the offer, or price increase or decrease.
Keep Good Records
If you do not keep good records, conducting tests is a waste of time and money. To keep good records you must be able to identify all responses generated by a specific SMS text, email, direct mailing, landing page, magazine ad, insert etc.
Make sure you record such things as:
- The hypothesis behind the test.
- The test objective.
- The test description.
- The test start date and end date.
- Sample sizes.
- Cost per order or lead.
- Revenue per order or lead.
- Selected audience description.
- Selected channel (e.g. web push, Google Ads, email in-house, Gmail ads, Taboola etc.)
- Person responsible.
- Test summary.
- Test conclusion.
If you don’t record these basic things you’re bound to forget implementing your winners, have test conclusions twisted by time and human memory and will certainly rollout tests that should never be rolled out.
Keeping good records is easy and a must have when it comes to running a true performance marketing program at your company.
That’s because refinements of your landing pages and other materials, audience selection and timing of the campaigns are made on the basis of the information gained from the initial and subsequent tests that all get recorded.
These refinements are then tested under stringent controls. Using this procedure the your marketing team will be able to fine-tune mailings and produce business considerably under the maximum allowable sales allowance.
This step-by-step method of send campaign, evaluate, refine, send campaign, evaluate is critical to the success of any performance marketing program.
A company cannot succeed in the long run by jumping from one promo piece to another without testing the copy, audience or creative against the performance of a control unit.
And you cannot properly evaluate any of these factors without keeping good records.
Types of Tests
The number of tests that can be made in a large performance marketing program are limited only by the budget and imagination of the persons involved.
Several types of tests that are very significant and necessary are discussed next.
Of all the tests usually conducted by performance marketers, none is more important than tests conducted to determine which Facebook audience, direct mail list, segment of the in-house database or media channel will work best for the offer being presented.
The first thing performance marketers consider is the total size of the test campaign and then how many different audiences to use.
Usually, a test of 5,000 each of 10 to 20 audiences will give enough data to provide direction to the types of audiences that will work best.
It is always advisable to test market segments within any audience. Frequently a specific segment within an audience will work very well for an offer, yet another segment of the same audience will not.
Some of the principal audience segments to consider for a test are:
- Age, income and education (for Facebook audiences or direct mail lists).
- Amount of purchase, recent date of purchase, frequency of purchase and product purchased (for you in-house database).
- Lifestyle indicators such as special interests and hobbies (for Google Ads targeting, Facebook audiences or direct mail lists).
Audience Facts to Remember
When making your audience selections for any test, remember the following facts:
- If your product falls into a special interest category such as outdoor activities, check to see if there are any specialty database groups of lists that will permit you to make the desired list selections and enhancements all from one source.
- Some segments of your house list may work very well while others will perform poorly. Generally a $5 to $20 buyer will remain in that price range. The $50 to $100 buyer usually will remain in that price range or higher.
- Affinity group lists with an endorsement from the group will outperform any list with the exception of a good house list.
- A third-party endorsement on any selected list may improve response rates 5 percent to 20 percent.
- Donor lists are sometimes poorly maintained and seldom perform very well on direct-mail and mail-order offers.
- Groups of small lists will usually outperform one large list within any market segment.
- Compiled lists or general audiences rarely work as well as mail-order or e-commerce buyer lists.
- When choosing magazine or newspaper lists, choose hot-line subscribers (the most recent subscribers) first. Then try segments of the 12- to 24-month buyers from those lists. The older names cost less money than the hot-line lists. Test to see if they will also perform adequately.
New Product Creative Tests
A new product offering is an ideal time to do some testing. Ideally, these new landing pages or other materials should be done by completely different creative teams.
Each creative team should be given the same product specifications, regulatory restraints and other pertinent data. All should be given the budget and time constraints.
At least three different packages should be prepared. This can be price differences (i.e. the cost for the consumer), offer differences (i.e. the cost for the seller to deliver the promised offer), creative differences.
So when the tests results are in, the cost per order, the revenue per order and total number of orders per package should be the principal measures of success or failure.
Check the Quality of Business Produced
The quality of the business produced is measured by how long the new customer stays active and how much revenue generates.
Customer quality can be influenced by the creative and audience selected. Don’t jump to conclusions about the success or failure of your promotion until the fulfillment and customer renewal cycle is completed.
Sometimes a cheap, sleazy-looking promo will produce a larger front-end response but will result in more returned goods and more no pays (if it’s cash on delivery for European or Asian markets or a bill me later scheme).
By testing a full range of different looks, a company may better determine which is best for them.
Confirm Results Before Making a Final Decision
Creative tests usually provide the only time your marketing team has relatively free rein, especially on a totally new product where you have no previous experience and no known competitors. After the initial round of tests is complete, an evaluation needs to be made quickly to determine which promo package should be chosen as the control.
If you had four test promos and found that two worked almost the same, the cost of product or service delivery may determine which you choose.
But don’t just arbitrarily throw out the more expensive one without careful evaluation. It may be worth testing by slightly modifying some unit in the tested promo to reduce the cost or by changing the copy slightly to make it perform even better.
If none of these seems viable, select the more economical version as control but use the other as well in a test segment to confirm results achieved on the initial test campaign.
Copy testing can be the most productive and probably the least expensive kind of test to do, except for audience testing.
Before changing copy used in your control a careful evaluation and review is needed. The creative and marketing teams should do the evaluation. This evaluation should be as objective as possible. When copy is changed, the changes should be limited to the most significant ideas in the piece and then tested against the control material.
Experience shows that the most successful changes are made in the lead and close of the landing page, sales letter, telemarketer’s script etc. as it concerns the way the offer is presented.
Even more important is usually the banner ad, text link, SMS text, email etc. that drive traffic to your offer presented on the landing page.
15 Questions to Ask When Evaluating Copy Tests:
- Does the lead on the landing page (or the headline on the insert or ad) quickly and accurately states the offer in a positive and benefits-oriented manner?
- Does the close on the landing page (or ad or on the preprint) provide easy-to-follow instructions on what the prospect needs to do to get what is being sold?
- Does the copy clearly establish the need for the product, then back up that need with facts related to the need for the product or service?
- Does the copy on the front panel of the brochure clearly establish what is being sold, summarize its principal benefits and state the offer in a clear and concise manner?
- Does the copy on the online order form clearly state the offer and the costs and obligations of purchasing the product or service? Does it quickly identify what is being sold? Does the application or order form provide ample room to answer all questions asked? Are those questions relevant and necessary? If not, don’t ask them.
- Is teaser copy being used on the outside carrier envelope? Does it contribute materially toward getting people to open the envelope or does it mark your package as just another junk mail?
- Are deadlines being used to get people to respond quickly? If not, consider them for future testing.
- Is a money-back guarantee of satisfaction being offered? If not, why not?
- Are testimonials being used? If not, why not?
- Has a celebrity spokesperson ever been used? If not, consider this to be an important future test idea.
- Are third-party endorsements being used? If not, consider this a future test option. As an example, let’s assume a company does lots of email or direct mail campaigns to an association group or buys banner ads directly from this association’s website. A campaign of the product with an endorsement by the association or its president can substantially improve response rates if the association or its president have a strong relationship with its members or customers.
- Is the copy “time oriented?” Does it create a sense of urgency to act in the reader? Does it urge the recipient to respond quickly? If not, revise the copy and test using more time-oriented copy.
- Does the copy attempt to be cute? Is it presumptuous? Is it negative? Does it include long, convoluted, run-on sentences? Is it loaded with big, seldom-used words? If the answer is “yes” to any of these questions, revise the copy and test against the control.
- Does the copy ask for the order up to five times in the course of a long landing page or sales letter? If not, revise and test. When asking for the order, it isn’t necessary to stop the reader. Indeed it should be done in such a way that it encourages the reader and whets his or her appetite for more information.
- Is adequate technical information given about the product so that the prospect can clearly understand what the product or service will do for him or her?
Product tests are the least-used but sometimes the most productive tests. Normally, most companies will limit product test ideas to focus group interviews or other marketing surveys.
All of them may be useful in creating new products, but none can do the job as well as actually testing in the marketplace. As noted earlier, what people say and what people actually do are two different subjects.
A common mistake is creating a product on the assumption it will sell without having any external input or marketing studies. Companies then will do one online ad campaign or one mailing and expect a miracle.
It would be a miracle if they succeeded. Performance marketing offers the opportunity to test products with carefully selected target audiences. These may be conducted on a local, national or international basis much more quickly, economically and accurately than by retaining a large staff to conduct a survey.
Price testing your products can make a world of difference in your profit and loss reports. Generally, most companies will underprice their products more than overprice them. Of course, competitive pressures are a compelling factor when stablishing the prices of a product or service. B
But consider this:
A large nationally known greeting card company was selling a decorative spoon set that ladies display on walls. The company conducted a carefully controlled price test of the product.
They tested the following prices for each spoon to be included in the set: $29.95, $39.95 and $50.00. To their amazement, they sold more $50.00 spoons than they did at either of the other price points. This result is also contrary to common department store pricing practices, which would hold that the price should have been $49.95 instead of $50.00.
The only explanation is that the perceived value exceeded the real value. Don’t leave money on the table!
Price threats are a sales technique regularly used by some companies. Price threat offers are usually used with the company’s own lists of old customers and inquirers. What they do is use notification of an impend¬ing price increase as motivation to get the prospect to buy. Typically, the opening copy would read something like this: “Buy before a big price increase on November 1 and save $75.00!”
For those skeptical about using such a technique, be assured it works year after year to the same market and often to the same people. Yes, they did actually raise the price, but it usually would be reduced later by changing the offer or using a different group of ancillary equipment. The basic equipment and price didn’t usually change significantly.
There’s a lot of money to be earned in the offline media. And some easy tests to conduct.
Postage tests are a very neglected area of consideration for many mail¬ers. Something as simple as knowing different postage application techniques and what class of postage to use can yield big dividends. These dividends can be in the form of money saved on postage and also on better response to your offer. Any increase in the response without having to increase mailing expense is a real bonus.
Make sure you test live stamps, meter stamps, third class mail in the US. If you’re based in Europe, especially in the European Union, do test sending your direct mail from abroad. Possible savings on postal charges are huge (as in direct mail from abroad may be up to 2 times cheaper than sent locally).
Most companies automatically assume they must have a brochure to sell their product or service, so they have one produced extolling the many virtues of whatever they are selling. Once installed in the mailing program, it is seldom looked at again unless there is a major change in the product or service.
A brochure’s primary function is to show what the letter sells. A brochure is a good place to provide technical detail, show pictures of the product, state warranties, guarantees and list legal and regulatory requirements. Most people read letters and scan brochures.
A good question to ask is, “Have we ever tested copy and design elements on the brochure?” If not, do so. Furthermore, if the brochure isn’t essential to show the product being sold, try a test unit without it. If you are selling insurance, a brochure may not be necessary.
My tests conducted selling life insurance have amply demonstrated this fact. In some cases, by not using a brochure, response was improved and in-the-mail costs per package were reduced.
The front panel on the brochure is a good place to picture the product and highlight its principal benefits. This can be very effectively done by using several short, fact-filled statements about the product and setting them off with “bullets” or check marks.
Order Form Tests
The order form is a key element in any landing page, mailing package and an essen¬tial part of any preprinted insert. As mentioned earlier, it is also one of the most neglected pieces in many performance marketing efforts.
Why is this so? Somehow, everyone just assumes these forms develop without much effort. They are seldom treated with the same interest and enthusiasm as the landing page or sales letter.
Careful attention to the copy and design of the response vehicle can substantially increase response. The order form should be easy to fill out, offer as few choices as possible, be short and be easy to read and to understand. Above all, it should headline what is being sold. The order form also should reassure the person by displaying your company’s money-back guarantee.
How to Add New Life to an Old Campaign
If you have a successful promo that slowly becomes less and less productive, give it a rest. That’s why it is so important to always have test units working against the control. A frequent cause of a quality material dying on the vine is audience fatigue.
The same people keep getting the same appearing, same sounding offer, month after month, year after year. After a while it’s like eating a bowl of corn flakes (without any fresh fruit) for breakfast every morning.
Here are four solutions to this dilemma:
- First, give the promo a rest.
- Try a complete new graphic look with the same copy.
a. Use a different creative leading to the landing page (i.e. change banner ads, email copy).
b. Introduce a prelander/presell page preceding the old landing page.
- If you have had a previous creative package of the same product that worked very well but has been pulled out of service for a year or more, try it again. It’s surprising how often a previously successful promo will work better than the present tired control.
- Test a completely different group of audiences if you can find any that have a psychodemographic profile similar to your present customer.
As you can see, there is no shortage of valid and reasonable items worth testing. Nearly any of the tests already suggested can be effective in increasing sales and reducing costs.
None of those suggested measure up to tests of the offer and audiences, however. Let those two items always remain in the forefront of any new tests. The other kinds of tests are best used for refining, reducing costs and fine-tuning existing campaigns.
- A well-defined test will provide a positive direction for future tests.
- By writing a carefully documented test summary, you set the direction for defining future actions.
- Promptly instituting changes indicated by test data permits taking maximum advantage of the data developed by the test.
- There are many types of tests that can be conducted. Some will help increase response. Some will reduce costs. Others will establish which product will work best.