Now don’t be too surprised at seeing men in the workplace; it’s the twenty first century after all, and as many of you know Mack has finally decided to add some testosterone to her team.
For those of you who don’t know me (so, all of you, really), I’m Reid, the intern. In between the coffee runs and manual labor that I assume are the standard fare for any Intern, I’ve been helping Tyler build a tool to streamline keyword research.
In the, hopefully not-too-distant future, this tool will scan the bulky, data strewn spreadsheets that I’ve come to know and love (and by love, of course, I mean fear), and spit out a sleek, new spreadsheet complete with a sorted list of keywords to target in our campaigns.
Actually, it kind of does that already. (Emphasis on the “kind of”). As of last week, we’ve got a working alpha up and running and now we’re looking to test it, turn that “kind of” into an “absolutely and with grace, ease, and aplomb.”
This is where you, our beloved lab-rats-to-be, come in.
“Help me, Obi Wan Kenobi; you’re my only hope…”
To improve our code we need to know how much it sucks, where it sucks, and why it sucks. We figure that the best way to determine this is by shamelessly handing it off to other people so that we can get data from outside the office.
What exactly do we want from you?
We are looking for criticism, but we need it to be constructive. (And not just because we’re 21st century men, complete with emotions and stuff). That’s the only way we’re going to be able to make the tool better.
Ideally, we’d like you to run old keywords through our code, keywords that you’ve tried out and determined the success (or failure) of as a source of traffic.
Then check the output of our tool against which keywords you finally ended up using (or not using), see how they match up, and, if you’re feeling generous, report back to us.
Are our top keywords the same as yours? What about the low ranked keywords? Do your actual experiences with these keywords reinforce the value estimates that our tool placed on them? This is the kind of stuff we want to know.
Once we have a good picture of how the tool works in its current form we’ll tweak our algorithms, add new ones if we need to, and try to get our results more on track with what actually works.
“Beneath this mask there is an idea, Mr. Creedy.”
Sadly, tweaks and minor fixes will only get us so far. If we want to move our tool out of alpha, into beta, and eventually into a finished product to share with the SEO community (which we intend to do, in case you hadn’t realized) we need to continue developing it.
Right now, our tool analyzes two major metrics as well as a user input weight for each keyword (I’ll go more into this later).
We can get a rough idea of which keywords are going to be valuable to us from just these three factors. In fact, so far, our outputs have matched up fairly well with the keyword research we have on hand. But, honestly, these three factors are just a few trees in the middle of a forest.
Anybody who has performed their own keyword research has their own techniques and set of metrics that they use to judge a keyword. To really hone this tool, we want to learn from your techniques and your metrics. By understanding the relationships between pieces of keyword data we can enhance our algorithms and rank keywords more accurately.
So please, please, pretty please, with sugar on top and a cherry or two: tell us what you do. What math do you use? What weight do you give particular variables? How do they interact with each other? If you’re willing to share, we promise it’ll be worth your while.
“Wait a minute, Doc. Ah… Are you telling me that you built a time machine… out of a DeLorean?”
There’s just one thing left. If you’re going to be giving us feedback on our tool, well, you should probably know how it works. The tool is designed to take several factors and calculate an output number. The output then ranks keywords from most valuable to least valuable. We’re currently evaluating two major metrics and a user input value in our algorithm.
The first metric is the Google Adwords monthly searches. Now each of you (probably) has your own special way of doing keyword research. Whether you pull broad match, phrase match, exact match, or some combination of these three is entirely up to you. Regardless of what actual values you use, just make sure you are consistent throughout the spreadsheet. If you use broad match for one keyword, make sure you use broad match for the other. Apples to apples – ya know?
As a fun experiment, you could even try two different sheets, one with broad match and one with exact match to see what happens.
“Go ahead, make my day.”
Our second metric comes straight from the big boys and girls over at Moz. Much like Google Adwords, the Moz Analytics tools will give you a difficulty ranking for keywords based on the who is trying to rank for them and just how hard they’re trying.
Unlike the Google Adwords competition ranking, though, we find the Moz Keyword Difficulty Score to be pretty accurate. So, when we want to know what kind of fight we’ll be getting into when we target a keyword, Moz are our go-to guys (and gals).
Finally, we’ve factored in the need for human input in this whole process. (Because until they’re smart enough to pull a Battlestar Galactica or a HAL, computers can’t entirely replace us).
The third variable our alpha code processes is a “transactional value”. The idea behind this is that, at a glance, most every keyword is going to indicate intent on the part of the searcher and that intent will have a certain value to your specific client. Our job, and the job of the fancy meat organs in our head buckets (which, again, barring Cylon evolution is an entirely human trait), is to determine what this value is.
For example, your client sells discount dress shoes. So a searcher using a keyword like “cheap brown leather shoes” is fairly likely to convert. This keyword probably ranks about an 8 or 9 (on a scale of 0-10).
On the other hand, a user searching “Does Tom Cruise wear brown leather shoes?” is a lot less likely to be of use to you. (Unless your blog runs a monthly feature on the footwear of eccentric celebrities). So you’d rank it around 0 or 1.
Be warned, though, that there’s a lot of room for subjective interpretation in the transactional value. The first reaction of two people to the same keyword could be wildly different, and when humans are asked to rate something on a scale we have a tendency to artificially inflate or deflate the values we assign.
To avoid submitting garbage inputs (and therefore getting garbage outputs), you need to do extensive research, consult with your clients, and be as sure as you can be of the values you attach to your keywords.
“Fasten your seatbelts. It’s going to be a bumpy ride.” (TL;DR)
So, how does this work? It’s really quite simple.
1. Create a CSV without any headers. The first column should be the keyword, the second column should be the Moz difficulty score, the third column should be the total search volume (local or global), and the fourth column should be likelihood of conversion (on a rounded scale between 0-10). (Here is an example of what a final CSV should look like)
2. Upload the CSV here.
3. Download the results.
4. Review your results. At this point the numbers are not on a scale, but the higher the number, the better the score.
5. Give us your feedback! Email us at firstname.lastname@example.org or comment below with your thoughts.
We’d be eternally grateful for any feedback you can share. Although we can’t pay you, we do offer hugs and Skittles.