Hello people , I am planning to build a new tool that will allow to create campaigns quickly for the big 3 search engines and maybe for others too (smth similiar to speedppc and other tools out there that I don't remember the name ). However i am still unsure if it would be better to create a windows app ( .net ) or a web based one . What would you like better?
Campaign builder tool - Web based or pc app (.net)
- Thread starter bbrock32
- Start date
Good luck, it's a ton of work if you want it to turn out decent ... it's also insanely gratifying.
If you want it to completely dominate. Make sure to include a keyword research tool, that pulls keywords from a bunch of sources with multiple seed keywords.. and here is the important part. Make it so you can take the results and group them into adgroups with a specified number (user defined) of keywords per adgroup, anywhere from 1 to 100 based on the keyword similarities.
No one that I know of has done this yet.. and I would pay definitely pay $100+ for a tool like this.
No one that I know of has done this yet.. and I would pay definitely pay $100+ for a tool like this.
$100+ !!! YOU MUST BE BAWLIN' HARD!
I don't think any coder would even start something like that for less than $1K broooooo.
That's interesting, but it sounds like you'd get a lot of false negatives.
Hypothetically speaking, if you were in charge of the automatic ad grouping of keywords, how would you go about this?
yea, it really sucks that he would only sell it to a single person and probably only a single time fee and thus only get paid $100 for it. its too bad its illegal to offer it to multiple customers.
:ugone2far:
a benefit of desktop software is its more likely to be trusted by your customers who are generally fearful of keywords being stolen and so they wouldnt want this data stored on your servers.
a benefit of web software is (for your customers) access anywhere and (for you) monthly subscription billing model and less piracy.
If i was building this, it would be 100% about which type of application development I was more comfortable with because this is not a trival task.
For a web based one, a lot of the ground work has already been done for you in Derek Beau's open source tool: How About an Open Source PPC Campaign Builder? : PROFITAPOLIS
Thanks for your ideas. I think I will build a web based one with a nice ajax interface.
What features would like , that aren't in other programs?
What features would like , that aren't in other programs?
Ive got competitive scrapers and some slick ppc tools I use, and on AVERAGE theyve cost me over 2k each...and usually take a bout 3-4 weeks to complete. Yeah, they do help, but no, they dont do all the work FOR you.
If anyone is considering taking this on - feel free to PM me. I've made a couple different campaign builders/refiners. I was never 100% happy with the results. Don't get me wrong, they were very nice groupings... but I just know that I can do it better manually.
For anyone interested... I used soundex and a custom script that basically assigned a numeric value to each keyword/phrase based on the characters that made it up. I used a blended average of the two variables to put the keywords into groups.
You can put in a modifier via text box or slider to specify the tight/looseness of the groups.
For anyone interested... I used soundex and a custom script that basically assigned a numeric value to each keyword/phrase based on the characters that made it up. I used a blended average of the two variables to put the keywords into groups.
You can put in a modifier via text box or slider to specify the tight/looseness of the groups.
Ok, so I'm totally fascinated now
One of the main reasons that I love PHP is because they already have functions in place to handle this kind of stuff. 4 of them to solve the same problem as a matter of fact. Your link led me to explore my options (thanks).
1. soundex - I didn't like this option because it outputs a phrase to A###, where A is the 1st letter of the word and # is a numeric representation of the rest of the word.
This might be a great solution for stuff like spell checking, it sucks for matching keywords because google & goatsie both would have similar structure. I'd be interested to hear why you chose soundex over the other options to match phrases.
2. metaphone - A very strong solution for matching syllables up as single letters (as I understand it to work). I'd rather work with numbers so.....
3. levenshtein - This function calculates the distance between 2 phrases. The smaller the number is, the better. This might be the right solution to the problem of matching key phrases ... I'm going to dig a little deeper the next opportunity I have.
4. similar_text - All things considered, this is the best choice IMO. similar_text compares 2 strings and returns the number of matching letters.
My reasoning for using similar_test is that if you take the number of matching letters as a % of the original for each word, you should be able to put a threshold in place that can match phrases up relatively well.
I've got a pretty good idea of how to match synonous phrases that are way off in spelling using the system I already have in place.
I'm excited to see how this turns out, and yes, I'm definitely going to give it a shot and will take any advice you guys can feed me.
Since I can't get this off my mind, I decided to whip something up using my similar_text ideas above and the results are decent I'd say.
There's not a lot of intelligence to it at this point other than matching up shared characters, finding a % difference from the original and then allowing thresholds ranging from 82%-70% depending on the length of the original phrase. I also filter out phrases that are not within 5 characters in length of the original.
Here's a sample
Obviously there is still some manual tweaking needed, look at words like supplements and pictures slipping in there ... However, I'd never trust anything automatic to be 100% correct.
What kinds of maximum/minimum limits for words should be followed when adgrouping?
There's not a lot of intelligence to it at this point other than matching up shared characters, finding a % difference from the original and then allowing thresholds ranging from 82%-70% depending on the length of the original phrase. I also filter out phrases that are not within 5 characters in length of the original.
Here's a sample
HTML:
fast weight loss <- root word
healthy weight loss, fast weight loss, steady weight loss, cause weight loss, easy weight loss, water weight loss, natural weight loss, best weight loss, ultimate weight loss, fat weight loss, fasting weight loss, obesity weight loss, instant weight loss, health weight loss, fastest weight loss, fitness weight loss, cylaris weight loss, fast weight lose, fast weightloss
weight loss surgery <- root word
weight loss surgery, weight loss program, weight loss supplements, weight-loss surgery, weight loss suppliment, weight loss suppliments, weight loss surgeries, weight loss procedure, weight loss secrets, weight loss stories, weight loss pictures, weight loss support, weight loss supplement, weight loss software, new weight loss surgery, weight loss surgery com, my weight loss surgery, www weight loss surgery, weight loss surgery uk, vsg weight loss surgery, weight loss by surgery, or weight loss surgery, weight loss surgery tax, weight loss surgery tv, weight loss surgery in, 46 weight loss surgery, a weight loss surgery, do weight loss surgery, get weight loss surgery, of weight loss surgery, on weight loss surgery, op weight loss surgery, the weight loss surgery, to weight loss surgery, weight loss surgery at, weight loss surgery ft, weight loss surgery las, weight loss surgery los, weight loss surgery san, weight loss surgery st, weight lose surgery, weightloss surgery, weight loss surgeons, weight loss surgeonObviously there is still some manual tweaking needed, look at words like supplements and pictures slipping in there ... However, I'd never trust anything automatic to be 100% correct.
What kinds of maximum/minimum limits for words should be followed when adgrouping?
both really
I'm currently doing the matching of words by characters ... +/- 5
What's the max number of words people would like to see as an adgroup?
Basically should they be tight groups of 10 or 100 words that are loosely related?
Good posts erect. You got me interested so here's what I did:
I took the similar_text() score for all values. Calculated and saved the differences(strlen($word) - similar_text($root, $word)). Took the standard deviation of the differences, then ran through all the values again to get the z-value of a particular "difference score". I used the z-value as a qualifier to decide what should be matched and played around with it for a while.
Here's the results of a qualifying z-value of < 0 (everything that fell below the mean, less is more here) :
Here's the results for < -.5 (the more below 0 the more "tight" the matches) :
I don't know if this is any better than just using a percentage threshold though. Also, I didn't do any filtering of overlapping keywords since I just wanted to test using similar_text() and a z-value to get matches and not get into the details of solving a real problem. I'm just fucking around trying to learn something that could help me automate shit in the future, although I'd never fully trust a non-human to create ad groups for me.
I took the similar_text() score for all values. Calculated and saved the differences(strlen($word) - similar_text($root, $word)). Took the standard deviation of the differences, then ran through all the values again to get the z-value of a particular "difference score". I used the z-value as a qualifier to decide what should be matched and played around with it for a while.
Here's the results of a qualifying z-value of < 0 (everything that fell below the mean, less is more here) :
Code:
root : fast weight loss
Array
(
[0] => healthy weight loss
[1] => fast weight loss
[2] => steady weight loss
[3] => cause weight loss
[4] => easy weight loss
[5] => water weight loss
[6] => natural weight loss
[7] => best weight loss
[8] => ultimate weight loss
[9] => fat weight loss
[10] => fasting weight loss
[11] => obesity weight loss
[12] => instant weight loss
[13] => health weight loss
[14] => fastest weight loss
[15] => fitness weight loss
[16] => cylaris weight loss
[17] => fast weight lose
[18] => fast weightloss
)
root : weight loss surgery
Array
(
[0] => fat weight loss
[1] => weight loss surgery
[2] => weight loss program
[3] => weight-loss surgery
[4] => weight loss surgeries
[5] => weight loss secrets
[6] => weight loss stories
[7] => weight loss support
[8] => new weight loss surgery
[9] => weight loss surgery com
[10] => my weight loss surgery
[11] => www weight loss surgery
[12] => weight loss surgery uk
[13] => vsg weight loss surgery
[14] => weight loss by surgery
[15] => or weight loss surgery
[16] => weight loss surgery tax
[17] => weight loss surgery tv
[18] => weight loss surgery in
[19] => 46 weight loss surgery
[20] => a weight loss surgery
[21] => do weight loss surgery
[22] => get weight loss surgery
[23] => of weight loss surgery
[24] => on weight loss surgery
[25] => op weight loss surgery
[26] => the weight loss surgery
[27] => to weight loss surgery
[28] => weight loss surgery at
[29] => weight loss surgery ft
[30] => weight loss surgery las
[31] => weight loss surgery los
[32] => weight loss surgery san
[33] => weight loss surgery st
[34] => weight lose surgery
[35] => weightloss surgery
[36] => weight loss surgeons
[37] => weight loss surgeon
)
Code:
root : fast weight loss
Array
(
[0] => healthy weight loss
[1] => fast weight loss
[2] => steady weight loss
[3] => cause weight loss
[4] => easy weight loss
[5] => water weight loss
[6] => natural weight loss
[7] => best weight loss
[8] => ultimate weight loss
[9] => fat weight loss
[10] => fasting weight loss
[11] => obesity weight loss
[12] => instant weight loss
[13] => health weight loss
[14] => fastest weight loss
[15] => fitness weight loss
[16] => cylaris weight loss
[17] => fast weight lose
[18] => fast weightloss
)
root : weight loss surgery
Array
(
[0] => weight loss surgery
[1] => weight-loss surgery
[2] => weight loss surgeries
[3] => my weight loss surgery
[4] => weight loss surgery uk
[5] => weight loss by surgery
[6] => or weight loss surgery
[7] => weight loss surgery tv
[8] => weight loss surgery in
[9] => 46 weight loss surgery
[10] => a weight loss surgery
[11] => do weight loss surgery
[12] => of weight loss surgery
[13] => on weight loss surgery
[14] => op weight loss surgery
[15] => to weight loss surgery
[16] => weight loss surgery at
[17] => weight loss surgery ft
[18] => weight loss surgery st
[19] => weight lose surgery
[20] => weightloss surgery
[21] => weight loss surgeons
[22] => weight loss surgeon
)here's a php-based keyword grouper that utilizes metaphones, soundex, levenshtein distance and a few other methods:
http://best3acai.com/kwgrouper/grouper.php
it's still pretty crude in the sense that it cannot take large datasets of phrases (1,000+) and needs to have the ability to delegate "orphaned" phrases into the closest, relevant adgroup.
anyone that wants to build on top of this and make it better, holla at me.
http://best3acai.com/kwgrouper/grouper.php
it's still pretty crude in the sense that it cannot take large datasets of phrases (1,000+) and needs to have the ability to delegate "orphaned" phrases into the closest, relevant adgroup.
anyone that wants to build on top of this and make it better, holla at me.
source? The only keyword grouper thingie I found searching was a windows based program. I'd love to peak into your php version
hollain'