Traffic so far:

May: 1500 unique users
20 from organic search

June: 3000 unique users
55 from organic search

Inbound links to this website: Zero that I've asked for. TBH I haven't checked search console in two weeks. ;-)

Does "3000 unique users" mean 3000 unique human users?

Does "zero that I've asked for" mean "but plenty that arose spontaneously"?

You don't need a Search Console to tell you where your visitors came from. Lots of social media referrals? Or lots of type-ins from the same IP ranges?

So, I'm building websites from scratch again like it's 10 years ago and I'm having a lot of fun doing it. There are unbelievable tools that let you build projects in a day and for $5

I gotta say, that is not how I would define "from scratch".

Lucy24, how would you define "from scratch?"

I would define it as etiher

1. refining sand into silicon, ore into iron, aluminum, platinum, and other metals, building chips, setting up an electric and communications grid, getting millions of people to sign on, building a server in my garage and connecting it to the grid I built and then figuring out how to send information down the pipeline

2. What Jake did. Start with an idea and end with a website.

As Carl Sagan said, "To make an apple pie from scratch, you must first create the universe."

>> build projects in a day and for $5

See - if you gave Jake a few years, he could probably create the universe, but why do that when we already have one?

>>unbelievable tools

What are some of your current favorites? Are you building from some MVC framework or from an "out of the box" solution that happens to be MVC

how would you define "from scratch?"

Writing your own HTML, CSS and (if used) PHP-or-equivalent. Otherwise it's like making a cake from a mix and then saying it's "from scratch" because it came out of your own oven instead of from a bakery.

Does "3000 unique users" mean 3000 unique human users?

Yes, lucy24. I've been doing this long enough that I know the difference between human and bot traffic.

Does "zero that I've asked for" mean "but plenty that arose spontaneously"?

Wow. Such appalling cynicism. No, it doesn't. I'm not trying to trick the community.

Writing your own HTML, CSS and (if used) PHP-or-equivalent. Otherwise it's like making a cake from a mix and then saying it's "from scratch" because it came out of your own oven instead of from a bakery.

And that's exactly what I did. I'm not going to roll my own MVC framework, just like I wouldn't write my own regex parser.

What are some of your current favorites? Are you building from some MVC framework or from an "out of the box" solution that happens to be MVC

For simple stuff, bottle.py. For more complex things, flask. gae-init is like a dream come true if you're developing for App Engine.

--
All of my user metrics look good:

Time on site is > 2 min and av. pages/session is 5.8
Bounce rate of course varies by channel, but site wide av. 30% total which is really good
Repeat rate is about 55% in the second month (!), which is amazing
Conversion rate for information sheet download is 37%, and email acquisition rate is around 6%

In May, traffic was primarily from social media (almost 100%), but this month it's a healthy mix of social (40%), direct (20%), email (20%), and other referrals (20%). I've been able to tie effort in terms of hours back to performance in terms of traffic, something not so easily achievable with "search engine optimization".

I think that 10K/day built strictly on non-paid media is still very achievable in 2016, despite the perceived difficulty of "traditional SEO". And not only that, but this stupid hobby website can scale to millions of users automatically and is cached at 90+ edge nodes around the world because Google automatically provides a built in CDN to you for all static assets coming off of their storage nodes. Cost for May hosting was $9.90. And I like that Google is responsible for the maintenance and security of my servers.

I'm pretty excited about all of this.

[edited by: bakedjake at 11:07 pm (utc) on Jun 24, 2016]

Because speed is really important for user experience, especially considering how many mobile devices operate on already saturated networks, monitoring the performance of your application is really important. But this used to be hard and the domain of esoteric software.

With the cloud platforms, this instrumentation is built in. For example, I can look at my homepage, and get a list of all requests in the last 10 minutes and what the server latency (processing time) was for each request. I can set alerts for outliers (for example, I get notified if 3 or more requests within a minute period are exceeding a benchmark I set).

There is a direct correlation between site speed and bounce rate, so this is a nice tool to have available.

@bakedJaked
Interesting topic, I'm currently working on bottle.py web-app with a MongoDB backend. This is my first serious go at it, but it seems like your working really fast to be able to nail something together in a day. I have been at it for about three month. Most of the work has been in the data collection and analysis (in Python). Right now I am dealing with form "pre" validation in JS.

What cloud service are you using or do you have experience with? I was considering digital ocean.
Are you using a db backend?

From your checklist:
- AMP (not applicable to this project at this time)
- OpenGraph Tags (definitely using this, I think proper integration into social networks will crucial for success)
- Sitemap.xml (Yes, but seems minor)
- (maybe later) Shopping Feed (not-applicable)
- schema (Yes for sure as applicable, at a minimum for breadcrumbs)
- oembed (What is this?)

I can truly separate my traditional "promotion" efforts (PR, influencer outreach) from my "technical" efforts (making sure I'm structuring my data properly for ingestion by multiple sources). Thinking in this way has really helped me to focus my efforts lately on the most effective avenues for promotion.

I couldn't agree more.

I love the approach of putting lots of different building blocks together to automate menial tasks. We have a saying around here that if you have to do something twice, it should be automated if possible.

Some of you may know about the incredible free tool ImageOptim for Mac to optimize your website images (you're running regular PageSpeed tests and looking at image sizes, right?), but did you know that the amazing developer (Kornel Lesiński) behind the software also has a web service at: [imageoptim.com...]

It's free in their beta period. I've been using it for a while, and it is incredible. Whenever my users upload images, they're automatically resized and optimized to my specifications. I like things like this that save my users and administrators time - and offload really difficult tasks to real experts.

The web service is simple to use - build a URL with some parameters, and get your processed image back.

[edited by: bakedjake at 10:44 pm (utc) on Jun 24, 2016]

Thanks for jumping in NickMNS!

What cloud service are you using or do you have experience with? I was considering digital ocean.

A few actually. Digital Ocean is a VPS provider, not really cloud. When I say cloud I'm talking about the cloud platform providers, mainly Azure, AWS, and Google Cloud.

I do use Digital Ocean for a production website on another project and find them to be a very good low cost solution.

Are you using a db backend?

I actually started with CloudSQL (Google managed MySQL) at first, but ended up rewriting the application to use datastore. Datastore is NoSQL that lets you do SQL, but the way you structure data in datastore is completely different than SQL and takes some getting used to!

- OpenGraph Tags (definitely using this, I think proper integration into social networks will crucial for success)

Oh man, in so many different ways. My latest idea was to build a custom graph application for Facebook and have my users post events to their timeline. OAuth is really easy to use and since users can one-click authenticate to social networks, I just use that as my user login mechanism instead of building a whole authentication system. Again, leave the hard stuff to experts.

And I get their consent to post my custom cards to their timeline.

- Sitemap.xml (Yes, but seems minor)

Our job is to make sure that our data is clean and ingestible as easily as possible. Why not take the hour or so to build the single template necessary to do this well, considering every major search engine has built in support for manual and automatic submission of this file. ;-)

- oembed (What is this?)

Custom cards for non-Facebook platforms. Think for WordPress. Right now most applications whitelist sites that can do this but it will be important soon.

Most of the work has been in the data collection and analysis (in Python)

I went to a deep dive talk yesterday on Amazon EMR and Kenesis and their Machine Learning stuff, and they did a full soup to nuts installation/import/transform/export on stage in an hour. It was incredible. Also Amazon is about to launch a web based visual data explorer - it's very primitive compared to RapidMiner/Tableau/etc but the fact that you can, in about 20 command lines, ETL and start clicking around and exploring gigs of data in minutes is just so cool to me. It's really useful for exploring data and getting your transforms right.

I still like to use RapidMiner to help me visualize my transformation process. It's good for documentation too. RapidMiner is open source, runs on any platform, and you can script it with Python and R.

The deep dive talk sounds interesting, was it in TO? I'm in Montreal.

As for rapidMiner, I should probably get to know it, but my first impression is that it is some kind of black box approach to data science. I am now used to working in Python, using Pandas Numpy and SciKit Learn, I can get most of what needs to be done completed relatively quickly. The issue with this is always data quality. Finding data, cleaning data and evaluating data is in my experience the most time consuming. Bad data into a black box solution will probably produce a nasty output. But that said I know very little about RapidMiner, when I have some free time I'll have to look into more seriously.

Such appalling cynicism.

No, I honestly couldn't figure out whether you meant "nobody links to me, it's all type-ins" or "I didn't have to ask anyone to link to me, they just did anyway because I'm that good" (which latter would be happy news and far from cynical :)).

This thread is not about seo is dead it is about "I've found new Tools to Play with, Fred!" and that's okay.

Code it right, even with plug and play components, have a viable topic in a niche that has traffic and voila! instant fame (and perhaps) fortune. Tools work. Humans are, after all, tool using animals. :)

Not only that but we can do big things very quickly!

Super easy machine digestible multi-language/currency websites are easy with frameworks like Babel now. Adding support for any currency Stripe supports is a couple of lines of code. Doing this 10 years ago would have been very difficult! Now you can do it in an afternoon.

Why not have your content in six languages? Translation outsourcing is a well solved problem.

The deep dive talk sounds interesting, was it in TO? I'm in Montreal.

Yep. They have it up on slideshare and on their AWS Summit blog though so you can see it and try all the stuff yourself.

If you're ever in TO, say hello!

@ bakedjake I'm going to check it out.

Sounds more like a sales pitch...so is it about these tools or is it about starting from scratch and no penalties? Why would Google love a site just because of coding, I thought it was about content....or does Google now hate Wordpress and similar CMS platforms? I can slap up a Wordpress site in a couple of hours. I'm missing something...

@samwest, sales pitch? maybe but the tools I use are all free so I wont be receiving any commissions. I think the point is that there are many tools out there that allow you to develop a site quickly and easily, even for the most complex ideas. This allows you to address some of the technical SEO issue from the get go, and frees up time for PR, sales and promotion after completion.

As for Wordpress and other CMS's they are great tools, I personally have never used them.

No samwest, I don't have anything to sell you. :) I'm just having a lot of fun and am quite excited. Apologies if I'm all over the place!

With regards to code, flask is a mainstream python framework for building web applications. Most of what I build is built on it. Google App Engine (which I use as a glorified web host, basically) supports it in a "controlled" way that lets it scale very well, as long as you follow their rules in building your website/applicatoin. There's no magic here, and all of the tools I use are open source and free. flask lets me be a better SEO because it lets me make decisions about how I present my content, but it provides a lot of basic stuff (form validation on the backend, for example) that I'm always going to need to do, so it saves me a lot of time.

WordPress is just a "more developed" version of what I do and is a fine platform. In fact, the way that wordpress works is fundamentally the same as the stuff I'm building. My flask application uses templates and a template language that is similar to the way you build templates in WordPress.

You can run wordpress on Google App Engine too if you want to try it out yourself: [googlecloudplatform.github.io...]

At the end of the day, SEO is all about 1) structuring content for ingestion by crawlers, and 2) promoting that content. These new tools give me better, faster, and more powerful ways to structure my data properly for search engines - and other social media sites, which are more and more acting like search engines and crawling content in much the same way. And it's becoming clear to me that there are very solid ways to bootstrap a sustainable website without relying on traditional organic website SERPs. If I can build it through image/social/video/local/shopping then let's get on with it and make sure we're presenting our data in the most effective way to maximize clickthrough from those sources.

I'm amused by the "on page content doesn't matter" meme. Metadata hasn't mattered this much since we had Inktomi!

And these cloud providers have massively redundant hosting options that scale incredibly well available for pennies a day. Again, I really like the fact that my site is hosted by Google. It's cheap, it's fast, SE/SM crawlers sit very close to where my website lives, and I like that my content is cached in 100 locations around the world on their dime.

My advice to webmasters who use all these "fancy free / 3rd party tools" would be to start building a transfer script because when all these tools get shut down (and the majority of they will), you are left with nothing ;)

My advice to webmasters who use all these "fancy free / 3rd party tools" would be to start building a transfer script because when all these tools get shut down (and the majority of they will), you are left with nothing ;)

Url Hölzle, Google's 8th employee, thinks their cloud offerings could make more money than ads:
[businessinsider.com...]

Amazon Web Services is approaching a $10 billion-a-year business:
[recode.net...]

I will take my chances.

MY problem with cloud platforms is that they lock you into doing things their way, and they limit what you can do. AWS is an exception provided you just you it as a scalable VPS provider, but then it just becomes an expensive VPS provider.

It is a lot easier (and cheaper) to run a VPS or dedicated server than it used to be.

Writing your own HTML, CSS and (if used) PHP-or-equivalent. Otherwise it's like making a cake from a mix and then saying it's "from scratch" because it came out of your own oven instead of from a bakery.

I disagree. That is like saying that you only really made a radio from scratch if you made your own transistors. Using existing components is just sensible, and lets you focus on doing the site specific stuff yourself. I would say it is not from scratch if the major functionality is not custom written. Should you write your own webserver as well? I tired that once, it worked, but its a waste of time.

There is a continuum. I do as little from scratch as possible. I use Django because it does a lot of things for me, but rarely constrains what I can do. I have used bottle.py for a simple API, and I am learning Seaside (which is really more suitable for web apps rather than web sites).

MY problem with cloud platforms is that they lock you into doing things their way, and they limit what you can do

Yes, true, that is a necessary tradeoff. Auto-scaling is hard, and they want you to develop in ways that makes it easier to support. That is an OK tradeoff for me - I'm fine with the constraints, because in exchange, Google manages my servers.

But you don't have to use App Engine (their platform as a service), you can spin up your own VPS or use containers and run your servers however you feel like. There's a million ways to do it. If I need to do something custom, they give me the tools to do that too.

If you're worried about lock-in, check out [kubernetes.io....] Containers across cloud providers and support for rolling your own on-prem stuff.

It is a lot easier (and cheaper) to run a VPS or dedicated server than it used to be.

It definitely is.

I agree with Lucy. I've built my own web platform due to the shortcomings of, well everyone I've ever encountered in real life as well as every piece of web software I've ever come across. All of my PHP, SQL, XHTML5, CSS, JavaScript and Apache related code I wrote from scratch and the longest string of someone else's code might not be longer than a single line of code posted in reply back when I was primarily learning, that is if any of that code has since survived unaltered if not outright replaced. Yes it takes longer but I can get a website up and running in thirty seconds flat and Bing, Google and Yandex verified in under two minutes. I used no frameworks, in fact I loathe frameworks especially client side ones. Everyone is in a rush to get things done faster but when you actually take the time to do things the right way it ends up letting you go much faster than any one else. Don't get me wrong, I'm glad you got something up and running fairly quickly and I hope it works as smoothly as any third party stuff possibly can. I should also take a moment to dispel the idea that from scratch means reinventing silicon elements and everything else a computer is made out of, nothing is more arrogant than smugly and easily discounting the enormous effort an individual or group of people put in to something that they're passionate about. Either way, if something works it works and the only way to judge something is how any given system works when it comes in to operation regardless of the fact that all these different systems require different amounts of time to come in to operation.

John

@bakedjake There is also Open Stack which does not tie you in to any one provider, but gives you quite a lot of tools.

I do not have any massive scalability problems to deal with that need a cloud platform, but I do find a lot of people using AWS, App Engine, Heroku and the like because they require less setup and it is easier to get started.

I loathe frameworks especially client side ones.

To put it in context, what frameworks have you used and what did you dislike about them?

Leaving the front end aside, what is the disadvantage of back end frameworks? You develop quicker and you have better tested code. The speed of development is not just a one off: it helps every time you change anything. Unless you have the resources to develop a good framework internally, using a framework will give you either better organised code, or just plain better code.

What is the point of writing your own database abstraction layer, or form validation, or routing, or sessions or CSRF protection?

PHP is something of an exception, because of its roots. It started as a framework and features kept getting added until it turned into a language, so it is a framework and a language. Not a very good framework by itself, which is why so many people either use a framework on top or it or write their own components to do the things a framework does.

I should also take a moment to dispel the idea that from scratch means reinventing silicon elements and everything else a computer is made out of

If you are replying to me, I need to clarify: my point is that nothing is truly from scratch. Some people go as far as writing their own web servers just for their site. That really is doing things from scratch! Almost everyone uses at least some existing libraries or components.

Some basic tools are used by everyone (servers, net connectivity, storage/retrieval, merchant processing, cdn/cloud) and think nothing of it because they didn't HAVE to think of it.

Old days it was HTML and go (add a little Perl for some server side stuff) then

BANG!

New tools based on OLD TOOLS, either defined/refined and, as the years pass, bloated with extraneous, though not necessarily useless functionality.

Current framework options are the next step in evolution of the same old stuff.

At this point it becomes more important to weigh the value/need to change/migrate from/to old style to frameworks, etc. and, in all cases, whether scope and scale will be required.

For a few sites --- admittedly a small number --- I've opted for Old School for the very simple reason the content begs it and the sites load fast! For more interactive/visual sites the new tools in the same old tool box have been nice to have.

What is the point of writing your own database abstraction layer, or form validation, or routing, or sessions or CSRF protection?

Yes! Exactly!

At this point it becomes more important to weigh the value/need to change/migrate from/to old style to frameworks, etc. and, in all cases, whether scope and scale will be required.

Agreed. For me, there's always an endless list of things I want to do. We must do more work in 2016 to launch a successful website than we did in 2006. These opinionated frameworks and tools help me execute on my ideas more quickly.

If you like to handcraft your HTML and CSS, you can take that static content and deploy to Google's cloud using gae-init in about 10 minutes with a few short terminal commands. It'll run faster than a PHP VPS solution, the servers are hosted at Google, it's got a built in content delivery network, and because it's a static site it can run in an F1 instance all day long which means it'll be hosted for free.

As mentioned, there are different levels of backend frameworks. You can can choose one that is as opinionated (WordPress) or not (flask, which provides ZERO opinion on how content is rendered to the browser and makes you do it from scratch) in terms of output. But to assume that unless you're uploading HTML or PHP files via FTP you don't have control over your site's load time or output is a mistake. And to assume that these tools are hard to use is also a mistake.

in fact I loathe frameworks especially client side ones

I generally dislike client side frameworks too but let's face it: Bootstrap is eating the world. And a lot of people are excited about Angular.

For a few sites --- admittedly a small number --- I've opted for Old School for the very simple reason the content begs it and the sites load fast!

Me too. Static HTML still has its place.

As mentioned, there are different levels of backend frameworks. You can can choose one that is as opinionated (WordPress) or not (flask, which provides ZERO opinion on how content is rendered to the browser and makes you do it from scratch) in terms of output.

I feel Wordpress is better suited to used as a drop in CMS, rather than as a framework for development. I would far rather use Flask for most sites that require actual development. In fact, I mostly use Django which is rather more opinionated than Flask, but still very flaxible. If I did PHP I would probably use Symfony or Drupal. The only thing I do not altogether like about full stack frameworks is using an ORM to generate schemas and queries instead of writing SQL, but its well worth it.