Which One of These Methods are You Using Right Now to Increase Conversions:
- You think it sounds good, you hope buyers will act – but you're not sure.
- You pull some text from an overflowing “swipe file” and hope to get lucky.
- You throw money at a campaigns & tweak until you go broke or get it right.
- What? … You mean I should actually be testing this stuff?
Seriously, if you're using any of these strategies to test and tweak your offers – you are wasting too much money, time and effort and not getting the results you need!
Conversion is emotion… Testing is about Statistics and requires a scientific approach!
- Benefits Of AdvantageBot
AdvantageBot contains a built-in database of over two hundred and seventy thousand profitable and unprofitable ads. It allows you to score your own text against this database for profitability.
It's really simple. You just copy/paste a headline, sentence, order link, Adwords ad, resource box or any other text that is meant to convert into the edit box in AdvantageBot. Once you do that, you click the “Score” button. In milliseconds, thousands of words and punctuation are compared to the internal database and scored. As far as you can tell… INSTANTLY, you see the score right under where you typed.
If the text you just typed scored higher than any prior text you typed and scored in this session, your text is also copied to the top of that screen so you can see the winning text at all times.
Change a single word. If it scores higher, the new winner will go up top. Try putting an exclamation point at the end of the sentence instead of a period. If that scores higher, your text gets copied to the winner's box. Keep working on any text as long as you would like. When you are happy that you've tried everything you can think of… just copy the winning text from the top and paste it wherever it goes.
Just think about the money and time you can save testing new headlines and ads. Just think of the money you are about to earn from increased sales.
AdvantageBot software allows you to score any ad copy after comparing it with a massive database of profitable and unprofitable ads. You simply choose the highest score and shave thousands off dollars and months of testing.
- When you use AdvantageBot to score your text you are effectively eliminating guesswork and prolonged, expensive testing. This affords you faster results and minimizes ‘bad ad' spend.
You can optimize any copy, across multiple niches, market sectors and regions quickly for yourself and/or your clients. Perfect for marketers, business owners, copywriters and ad/marketing agencies.
Built-in database of hundreds of thousands of ads – We'd argue the most powerful advertising and marketing tool ever created.
Instant results from Scientific Proven Methods – AdvantageBot has taken over 7 years to become fully realized and presented as it is today with tens of thousands of dollars spent in testing and data analysis.
Will it work for you? Why not try it and find out?
- How We Made AdvantageBot
Here is the scientific, step=by-step process we used to create the AdvantageBot scoring dataset:
1. Gather advertisements from a wide variety of periodical sources. We used trade magazines, magic ads classified sheets, penny saver classified sheets, Google Adwords, Overture ads, classified ads in multiple wide distribution newspapers, Yellow Page ads from New York City, Los Angeles, Miami, Chicago and Sacremento, sales sheets from popular Sunday edition broadsheets in both the US and the UK, social media run advertising including Facebook, LinkedIn, Twitter and Youtube – Even billboards across various Interstate highways for our initial study. We save very little variability in the token scores among the variety of ads although there was some variability with punctuation and numerical scores.
2. Find the same publication for weeks or months later depending on the type of ad. Trade magazines had to be at least 6 months apart. Classified ads had to be at least 3 weeks apart. Yellow page ads had to be at least one year apart in the cases where multiple yellow pages directories were printed in a single year. Google Adwords ads needed to be only one week apart. The amount of time between the two publications was determined by the amount of time that was typically purchased by an advertiser. If a magazine advertisement was typically purchased for three months at a time, then issues of that magazine to be captured has to be at least four months apart. The same rules were applied to all periodicals.
3. Capture all ads in each periodical. This was sometimes performed by scanning and then using OCR. Yellow page ads were typed in manually due to the variety of graphics and fonts utilized. The same was true for billboards. Each ad was captured by three different contractors. The same three contractors captured the same publication's later issue, but with other capturing work in between. This was an attempt to make the capturing work more consistent and less biased. The captures where then compared and anything less than a 95% match resulted in more captures of those ads by different contractors until at least three of the captures had a 95% commonality. Commonality was measured by tokenizing the ads, sorting and then counting matchies vs. mis-matches.
4. The official tokenizing now took place among the three captures for each ad. All carriage returns and line feeds were converted to spaces. All double spaces were then converted into single spaces. Each item separated by spaces was deemed a token. In addition, the punctuation marks and digits were also added manually as tokens.
5. A master list of tokens found in all ads was created and duplicates removed.
6. The ads were then categorized as ‘profitable' and ‘neutral' by looking to see if they were running in each issue of the periodical used in the study. If the January and July issue of a particular trade magazine contained the exact same ad, that ad was put in the ‘profitable' pile. If the ad had changed in any way with the wording (graphics or fonts could change, but not wording), then it was placed in the neutral pile.
7. The total number of ads in each pile were counted for later normalization. The neutral pile was approximately 9 times larger than the profitable pile.
8. Each token in the master list was then counted in all of the ads in the profitable pile vs. all of of the ads in the neutral pile. The normalization value calculated in step 7 was applied to the profitable pile so that the numbers were comparable (we multiplied the count in the profitable pile by how much larger the neutral pile was than the profitable pile). Alternatively the two piles could have been made equal by randomly removing ads from the neutral pile until the total number of ads was equal in both piles. We used the normalization value.
9. If the total count was less than 19 (before normalization) for the token for both piles combined, that token was removed from the master token list and the count was also ignored. If the total count was >=19, then the ratio of the counts after normalization was calculated. The lower number is divided by the larger number resulting in a number between 0 and 1. If the neutral count was more than the profitable count (after normalization) the the number was assigned a negative sign. Otherwise the sign remained positive. For convenience, we multiplied the resulting number by 100 and rounded it to an integer in order to produce the final score (so that we were working with integers between -100 and +100 rather than floating point numbers between -1 and +1.
10. If the score was zero, the token was removed since it would have no effect whatsoever in the final scoring database.
11. The AdvantageBot scoring dataset was almost complete at this point, but there were redundant tokens existing that caused a double count. It turns out that this did not affect predictability, but it did cause higher scores due to the redundancy in the tokens. To remove the redundancy, each token is searched for in the rest of the token list. If the entire token is found to be embedded in another token, the absolute value of the score of the smaller token is subtracted from the score of the larger token. As mentioned, it turned out that this step is not productive even though not taking this step results in double counting of redundant tokens.
12. The list of remaining tokens and their score is the AdvantageBot scoring dataset. It is applied in AdvantageBot by checking for the existence of each token in the text under test and if the token exists, then applying the score to an accumulator. After each token has been tested, the accumulator contains the AdvantageBot score for the text under test.
13. Validation is then performed by scoring advertisements with known profitability along with the same ad in earlier incarnations that had known non-profitability. The percentage of times that AdvantageBot scores predict the already known results is the predictability rate. It initially appeared that we needed 90,000+ ads to create a scoring dataset that had a predictability of approximately 85%. Increasing that to over 200,000 ads does not significantly change that predictability. Reducing down to 12,000 ads reduces the predictability to 78%.
- AdvantageBot In The Future
AdvantageBot – The Most Important Marketing Tool In The Last 100 Years… And For Years To Come…
Predictability remains about 86% regardless of increasing the dataset size to over 280,000. We started with a dataset of 12,000 ads and increased all the way to 286,000 ads with the latest study but with no significant increase in predictability after approximately 90,000 ads.
Currently the text under test is not tested for multiple instances of a token. It is not known if this could improve predictability.
There is a seasonal drift that exceeds 3% which is far greater than the decade drift which is estimated to be approximately 0.5%. A version of AdvantageBot that uses monthly datasets may therefor increase predictability by and additional 3% on a month to month basis.
The variability between UK ads (a smaller part of the dataset) and American ads was significant (over 1%). A separate version of AdvantageBot for Britain vs. the United States would probably increase predictability by an additional 1%. Other English speaking markets have not been tested but our conclusion is that variance would be no greater than 1% predictability.
The month of December appears to be very consistent from year to year, but up to 4% different than it's most different month (which is February). The month-to-month version of AdvantageBot could add yet another 1% of predictability for December and February.
Different ad sources varied by only about 1% with the exception of the use of numbers and punctuation marks. Specific versions of Advantagebot designed for use with particular kinds of ads could increase predictability by 1-3%
Variability between markets (computers vs. flowers used in the study) was surprisingly low. Follow-up studies for other markets could yeild some markets where market specific AdvantageBot scoring datasets could increase predictability by a yet unknown amount although variance is estminated to be less than 1-3%.
Capitalization is currently not tested at all. A future study in the use of Capitalization could yeild further interesting results.
Important: When you purchase AdvantageBot you will also receive a lifetime software update promise from us – All future enhancements to the software and the scoring database will be provided to you free of charge with no ongoing fees. While we do not anticipate this to be frequent as we are currently at the highest level of predictability things can and do change so you can rest assured your copy of AdvantageBot will always give you a true advantage.