Later the same product was suggested in still another form. The previous verdict made the change look dubious. The advertiser hardly thought a test worth while. But he submitted the question to a few thousand women in a similar way and 91 per cent voted for it. Now he has a unique product which promises to largely increase his sales.

These tests cost about $1,000 each. The first one saved him a very costly mistake. The second will probably bring him large profits.

Then we use test campaigns to try out new methods on advertising already successful. Thus we constantly seek for better methods, without interrupting plans already proved out.

In five years for one food advertiser we tried out over fifty separate plans. Every little while we found an improvement, so the results of our advertising constantly grew. At the end of five years we found the best plan of all. It reduced our cost of selling by 75 per cent. That is, it was four times more effective than the best plan used before.

That is what mail order advertisers do—try out plan after plan to constantly reduce the cost. Why should any general advertiser be less business-like and careful?

Another service of the test campaign is this:

An advertiser is doing mediocre advertising. A skilled advertising agent feels that he can greatly increase results. The advertiser is doubtful. He is doing fairly well. He has alliances which he hesitates to break. So he is inclined to let well enough alone.

Now the question can be submitted to the verdict of a test. The new agent may take a few towns, without interfering with the general campaign. Then compare his results with the general results and prove his greater skill.

Plausible arguments are easy in this line. One man after another comes to an advertiser to claim superior knowledge or ability. It is hard to decide, and decisions may be wrong.

Now actual figures gained at a small cost can settle the question definitely. The advertiser makes no commitment. It is like saying to a salesman, "Go out for a week and prove." A large percentage of all the advertising done would change hands if this method were applied.

Again we come back to scientific advertising. Suppose a chemist would say in an arbitrary way that this compound was best, or that better. You would little respect his opinion. He makes tests—sometimes hundreds of tests—to actually know which is best. He will never state a supposition before he has proved it. How long before advertisers in general will apply that exactness to advertising?