An Essay on Modern SEO. A Call to Pipulate.

by Mike Levin

SEO stands for “Search Engine Optimization” in the field of marketing and advertising. If you write it, they will come. You are assured a certain amount of reliable, recurring visitors to your website merely by virtue of existing. Or are you?

Does SEO still work?

Am I still getting traffic by virtue of merely existing? What is quality, anyway? Yup, we’re relying on AI more than PageRank now I think. Or we soon will be. So is all that an old type of “organic” traffic dying off? Do we all now just have to pay for traffic, or be one of those few websites somehow deemed best? What does the data in our website tell us?

Looking For Data

How do we know? The default dashboards of Search Console and Google Analytics? Custom reports? Particular drill-down paths through the menus? How do you know, and even then, how do you know what to do as a result of charts and graphs and exportable lists? What questions are we even asking, and to whom do we ask them and how?

We Find Google GSC & GA APIs

We ask our questions of Google, of course. Usually through their GSC or GA APIs. With simple expression, and usually through Python and in Jupyter. Also in a way where we could make it re-send us that answer again with the data of a new day. Real-time’s great, but bat ain’t bad, ‘cause we all have to sleep. Think wake-up emails of your bidding.

Automatic Clarity-inducing Reminders

That’s what we seek here, right? Clarity. A wake-up call you rigged to ensure a good day. Actionable and repeatable. Got a concept? Here’s some code. Flesh it out, refine it and customize. Send it as an email. Throw it into a scheduler. The email is no more than you need and no less than a satisfactory answer. 80/20-rule. Strike the right balance of tech liability and perfection.

Expressing Inquiries With Ease

Eventually the language of expression fades into the background as a transparent tool. You forget about a hammer while hammering or else you’ll miss the nail. SEO, you say? What are the basics today? Express it simply.

Looking For Trends

Wow’s my site doing over time? So the data coming back is almost certainly going to be time-series. We’re going to want to look at trend-lines for stuff. But what stuff? Which keywords are leading people to my site through search, of course! What are the landing pages? How many are influx of new people and how many are folks I see all the time at my site?

First the Google Search Console API

Answering this is usually a Google game these days. You use Google Search Console for SEO data and Google Analytics for Users and Sessions. PageViews are on the way out, but we’ll defer the GA discussion for later. Let’s first focus on GSC. GSC only returns the 3 metrics, impressions, clicks and position. It also returns ctr (click-through rate), but that’s just a ratio derived from impressions and clicks. Well, let me just show you a bit of response…

{'rows': [{'keys': ['pipulate'],
   'clicks': 17,
   'impressions': 72,
   'ctr': 0.2361111111111111,
   'position': 3.0694444444444446},
  {'keys': ['levinux'],
   'clicks': 4,
   'impressions': 42,
   'ctr': 0.09523809523809523,
   'position': 5.023809523809524},
...
  {'keys': ['youtube xvid'],
   'clicks': 0,
   'impressions': 11,
   'ctr': 0,
   'position': 11.818181818181818},
  {'keys': ['تولبار اليكسا للفايرفوكس'],
   'clicks': 0,
   'impressions': 13,
   'ctr': 0,
   'position': 19.846153846153847}],
 'responseAggregationType': 'byProperty'}

The dictionary in the response containing ‘keys’ as a key is followed by a list of keywords, a list that’s usually 1 long. Sometimes, there’s a URL or Platform or Geo data in there, depending on the dimensions included in your request. So reaching into these objects can be bit tricky, parsed by fancy combinations of list comprehensions, lambdas and mapping. This site will provide plenty of examples.

Looking For Trends

Okay, so what keywords are leading people to my site? What are the trends? Where am I on the rise? Which keywords and landing pages are those? Where am I on the decline? Just give me the highlights — with only 20% of the effort I imagine could go into a project like this, how do I get 80% of the benefit?

Trends, The Basics

Don’t fail to do the basics before the advanced stuff. There is much you can visualize with just your GSC data and matplotlib in Jupyter. It’s not very sexy, but it proves you’re competent. Take in a bunch of X & Y values, and plot them in a grid. Make the dates on your X-axis readable. Make the metric used on the Y-axis clear.

Automatically Emailing Actionable Findings

Sure you can do that in Excel, but can you automate the sending of that graph in email with no webserver in the picture and no proprietary software? If you can provide net connection and Linux machine, you’ve got site trend monitoring as an ongoing SEO service you could offer, without depending on paid-for 3rd party software or services. A few good Python scripts and you can deliver what most small to medium size businesses need to know to optimize their sites. Let’s break it down.

The Basics Is Linear Regression

Graph one line. Then try quantifying a trend with a linear regression. Sort a dataframe by the steepest negative and positive line-slopes. Find the most in-danger and up-and-coming tends. Look at those keywords and landing pages. You can use your sites own Search Console data to do these exercises. You don’t rely on all that much to do such informative and actionable investigations like this. This is the basic minimum skill set you should look for in somebody claiming to be an SEO. The good news is it’s simpler than you think.