9 Steps OnlineMarketplaces.com Uses to Produce 'Machine Beating Ceiling Content'

As Rand Fishkin pointed out recently, "AI content is the new floor". Well if that's the floor when it comes to content, then reports based on original research are surely the ceiling.

The machines can't do those yet.

The following was originally drafted as an internal tutorial for people making content at OnlineMarketplace.com but I've adapted it here for our readers (and anyone else who might be interested) to get a look into the work that goes into our original reports.

These are the steps we use to produce our machine-beating 'ceiling content' like the Real Estate Portal Ownership Report we put out earlier this month...

Step 1 - Idea to Picture

We record a lot of metrics for the content we produce but we don't keep track of when or where we came up with good ideas. If we did, I'd be willing to bet that most of the best ideas for the content we produce have come to me on my daily walk.

No surprises.

There are plenty of studies out there about how your brain is most creative when you're chilled out, right?

The point here is that the ideas I have tried to come up with have almost always been the ones that performed poorly or were given up on. Remember any of these classics?

Just because an idea originated outside my apartment doesn't automatically qualify it for a write up though...

If the idea can't be quickly distilled into a one-sentence question with a metric (number of leading real estate portals for example) and a dimension (let's say... business model categories) it's usually no good.

Next up is to picture that question on a page.

Good analysis articles are just context around a picture. That picture is the data you're going to get and plot on some graphs.

The clearer you can mentally picture it, the more succinct your analysis will be. The more succinct your analysis, the more people engage with it.

If you can't sketch what those graphs will look like, then get your shoes back on and walk another mile.

Step 2 - Need to know, hard to find (data collection planning)

The tedious phase of content creation tends to begin with two questions:

1) Does this data exist?

2) Where can I find it?

A "no" for the first question doesn't necessarily mean that the idea is going to go the way of Boomin (niche joke there). There's sometimes a question 1.5 which can be bad for your sanity and eyesight...

1.5) If the data doesn't exist yet, can I make or find it myself?

Put in the hours.

Finding out if the data exists (maybe you know a third party that has it or maybe it's readily available to the public), where you might find it, what format you might find it in and when and how you are going to collect it is essential before embarking on step three.

Step 3 - Tedium (data collection)

So you've decided on original data for your original content. You fool.

There is a correlation between these three things.

• How original some data is.
• How tedious it is to collect.
• How well received the content that comes from it is.

The output I am most proud of in my professional career is directly responsible for my failing eyesight.

Once again, it's time to put in the hours. The only advice here is advice that you won't like...

1. If the data is even slightly complicated or subjective, collect it all yourself. If third parties get involved it won't be standardized.
2. Don't ever be tempted to "just come back later" to a complicated, arcane or hard-to-find data set.
3. Make a tutorial (for yourself) on how to collect it and how to format it. You might forget.
4. Copy the URLs of your sources and put them in a note somewhere.
5. Lots of cups of tea and loads of biscuits.

Even once you've spent all winter squinting at the financial data for real estate portals or looked through 800 websites to see if they have a pet-friendly filter on their rentals section, there may well be more tedium in store.

Step 4 - Data formatting

If you managed to get cool data from a third party, then you will definitely have to do this. Even if you collected the data yourself, there's a good chance you'll have to do this.

The first thing here is figuring out which tool you want to use to make your graphs.

You can always make static graphs on things like Canva which are fine as long as your data is absolutely 100% static and never going to change or you're a weirdo perfectionist that enjoys editing something 35 times before publishing it.

The likelihood is though that you're probably going to want a tool that's connected to the spreadsheet your data is on and updates automatically. There are loads out there but the three we use for various tasks are:

1. wpDataTables from TMS (a WordPress plugin we use for small graphs on our articles).
2. Looker (we much prefer this to awful, corporate pdfs for our big reports)
3. Datawrapper (really nice for certain types of graphs and adding annotations)

Whichever visualization tool you go for, you probably need to do some copy-pasting, pruning, column name-changing, changing of the data types, swearing, and crying because there is no way your data is in the right format on the first try.

^(?:/articles/[^1-2].*) Leanring RegEx is the best advice I can give here.        

Step 5 - Bullet points around graphs (the best, easiest and shortest step)

The reason you do all of that is to see the data in a lovely graph that makes everything clearer.

Once you've untangled your data and found your preferred tool, that's exactly what you get to do next. The eureka moment you've waited for. The best bit.

Don't try to make the graphs perfect now, you'll mess around with them in the edit. Just make a first, intelligible version of each and then write bullet points next to them explaining why they're interesting.

You should now have what we in the business call 'a narrative'. Something to put as a title and on social media posts.

Step 6 - Iteration and fact-checking

Maybe you were able to construct a good narrative from your graph on the first try. Maybe the values in the graph made perfect sense at first glance.

Probably not though.

Usually, you can add or remove data to make what you want to say more succinct. Often you see something that doesn't make sense.

Whatever the case, now is a good time to question what the data is showing you and then go back through it and check that step three was done properly. Remember those obscure URLs you (hopefully) jotted down?

Mike DelPrete has written an excellent guide to how he goes about this process which includes a good example of data visualization iteration. I'm not ashamed to say that I copy what he does on many things and that's one of them.

Step 7 - Opinions

Unless you're in the top tier of respected people in the niche (like Mike) you're going to want to solicit some expert opinions to lend clout to your narrative.

Luckily we have one of those in-house. Simon Baker is always willing to jump on a Zoom call for YouTube and gets the tone right every time.

Getting busy people to give up their time requires good email writing skills and patience but it's worth it.

Obviously, they might share the content once it comes out but even better than that, they might actually tell you something you hadn't thought of or say something counter-narrative (content marketer holy grail stuff).

If anyone reading this thinks they're an expert on real estate portals, doesn't think like everyone else and, importantly, responds to emails in a timely manner, get in touch.

Step 8 - Writing and Editing

Once you have graphs, bullet points and some expert opinion, writing is a breeze.

What you'll be writing for a typical OnlineMarketplaces.com piece of original research are...

• The words around the outside of a report.
• A landing page - you subscribe to our mailing list, we give you the report.
• An accompanying article (linking to the landing page).
• An email to go out to the subscribers (linking directly to the report).
• At least two social media posts (one before publication to hype it).

Here you'll want to find the most interesting point from the report, distil it into a sentence with a number in it and repeat it consistently until you're tired of it.

The hook is the hook for a reason.

You know the drill here... Get everything drafted. Wait a day or two. Edit it. Play with the graphs. Play with the style. Make it look cool. Edit again. Publish.

Step 9 - Promotion

Anything I could possibly come up with here would be vastly inferior to what Andy Crestodina wrote on the subject as recently as last week. The man knows his onions!

Yes, writing articles like this one is one of the steps in Andy's playbook.

Step 9+1 - Analysis

Knowing what type (and subtype) of content, what word count, what subjects, what social media distribution, which graphics and what days of the week work best has greatly improved what we do.

Good reporting and analysis require discipline though...

• A report is only as good as its data.
• Journalists are only as good as their sources.
• *Content analysis is only as good as the spreadsheet it comes from*

We like to think we're good to freelancers. We pay on time and we try to get them a nice portfolio of work and an introduction to a lot of contacts.

That said, if they don't fill in all the fields on the big spreadsheet that feeds the internal analytics, I have been known to be grumpy at them.

For reports I have a special property on Google Analytics that I use to track performance without the need for a spreadsheet. All you have to do is paste the ID of the property into the report settings on Looker. Another great reason to not put reports out as PDFs.

After 9 steps hopefully, you'll have not only made machine-beating ceiling content but also you'll have collected some valuable data which nobody else has and you'll have refined the process a bit to make it easier next time.