Stories about Software


Reasoning About and Projecting SEO Content ROI

Hello folks.  And thanks for sticking with me as I continue my relentless assault on any remaining soul I have yet to suck out of content creation.  I’ll try to finish it off today.

And there’s no better way to do that than to start talking about return on investment (ROI).  Return on investment is amount that you wind up reaping (or losing) from a given investment of money or labor.

There are two main ways to think of ROI for SEO-minded content:

  • Comparing the ROI (or cost) of SEO campaigns to that of other lead acquisition tactics, measuring relative ROI.
  • Reasoning directly about the revenue (well, profit, really) generated by your SEO-minded content efforts.

I’m going to dive into both extensively.  But first, let’s take a detour through some business terms and some extremely opinionated takes on digital marketing metrics.

How Should We Measure a Content Program? (Hint: There is a Right Answer)

For my money, one of the most surreal questions that I see when DevRels or content marketers get together and talk shop is a simple one.

What’s a good metric for the success of a content program?

This seems both innocuous and like a good question.  But here’s what I hear when someone asks that:

What’s a good cardinal direction to drive in?  I’m a south kind of guy, but I’ve been hearing a lot of good things lately about northeast.

Both questions are kind of absurd, because both beg the contextual question of “what are you trying to do / where are you going?”  Metrics and tactics both are meaningless absent articulated goals.

If DevRels and content marketing managers are asking this question as a matter of shop, someone in the org chart above them has faceplanted in setting them up for success.  Nobody should need to flail around aimlessly for a success metric or a driving direction.

Luckily, however, unlike “which direction should I drive,” there is a universally right answer about how to measure the success of a commercial content program.  You measure that with the relationship between customer acquisition cost (CAC) and customer profitability.  This is your return on investment in the purest form: how much profit do you realize from spending money or time on content?

Real Quick, Business 101

Creating content for a business is a marketing activity or, more broadly, a growth (customer acquisition) activity.  In a post I wrote about profit some years ago, I articulated three systems for business:

  1. Acquisition: a funnel-shaped system that takes “the world” as input at the top and poops out customers at the bottom. (e.g. sales and marketing).
  2. Service Delivery: a system that takes a customer as input (from acquisition) and delivers a satisfied customer as output. (e.g. you as a freelancer not sucking at writing code during the code-writing hours you sell your client).
  3. Overhead: a support system that provides the necessary ‘fuel’ to acquisition and delivery. (e.g. invoicing, time tracking, a CRM, etc).

Tasked with measuring acquisition via content, I imagine the first thing that you think is “measure the number of customers the content produces.”  And, while this isn’t the worst thing to measure, it’s far from the best.

After all, what happens if it produces terrible customers?  Or, more subtly, what happens if it delivers customers, but you spend more money acquiring the customer than the customers ever pay you?

CAC, LTV, And Ratios

Say you spend $500K building a content library and that library, over the course of its existence, brings in 10 new customers, each of whom gives you $100?


I assume you wouldn’t consider this program successful because it produced 10 customers while wasting (at least) $490K of your money.  But it gets even muddier.  What if those 10 customers each paid you $50K?

Well, then you’ve broken even: paid $500K to get $500K in business.  But that’s not good, since servicing a customer has cost.  Your CAC has to be greater than the realized revenue for a customer (which you might start to approximate with a figure called lifetime value, or LTV).

You can read more about this if you want because I need to wrap up this rabbit hole and get to the point.  And that point is that content (or any growth activity) needs to yield profitable customers.  It does this by maximizing the LTV (really profitability) to CAC ratio.

So that’s the only metric that truly matters, in the end.

Why Do You Never Hear About This Metric?

And yet, you never hear content marketer types talk about this.  Why is that?  If you’re a content-flavored technician, it’s understandable that you wouldn’t know a ton about business profitability, but why is nobody telling them or demanding accountability here?

The answer is simple: because it’s REALLY hard to do that.

Holding marketers accountable for the LTV/CAC ratio wouldn’t be fair, since marketers have no control over how profitable your service delivery is.  LTV/CAC spans organizational silos to such a degree that you’d have to stop at the CEO’s office in most orgs to find the lowest level of the org chart that could control both concerns.

Now, the growth side of the business can entirely control CAC, but marketing can’t.  After all, what if marketing makes it rain leads through the website, but sales is terrible at closing those deals?  Unfair to measure the marketers in this fashion.

So the business silos further and the measurements become both more fair and more useless.  You start measuring marketing efforts based solely on leads.  Then maybe you narrow it further for content marketers to “site visits” or “atta boy comments on social” or even the worst sort of metric: “pieces of content produced.”

(As an aside, I think pieces of content produced can actually be an excellent improvement metric for overly image-conscious early stage founders, but that’s a topic for another time)

And in the end, you wind up with DevRels on marketing forums, asking what to measure and what direction to drive this weekend.  And then you, as a non-marketing freelancer, take your cues from this as you figure out how to measure success with your own marketing.

Content, SEO, CAC, and ROI

So for the sake of the rest of the post, let’s walk things as close as we can get to the P&L while still hanging out in the growth portion of the business.  Let’s consider CAC.

If you create a bunch of content that never produces any business, the cost of your acquisition is infinite (your $ spent, divided by 0).  If you spend half a million to generate a single customer, your CAC is half a million.  Spend less, generate more customers, and things improve from there.

Lower is better in CAC-land.

So your goal with SEO programs and content is to throttle your CAC as low as possible, producing customers in cost-effective fashion.  And by the way, ROI is essentially the relationship between CAC and LTV, which means that lower CAC is highly predictive, ultimately, of ROI (assuming you have a viable business with viable fulfillment).

We’ll come back to this in a bit and unpack it a little more, BTW.

Get Yourself Comfortable with the Back of the Napkin

For the rest of this post, I’m going to talk about reasoning through CAC and program cost.  But you need to understand that gathering data in the marketing world is really hard and imprecise.

You’ll need to acclimate yourself to making baseline and ballpark assumptions.  For instance,

If we assume we get 100K visitors per month to the site and we assume a 2% lead qualification rate and a 10% lead conversion rate, our program is generating 200 customers per month.  We built that 100K per month visitor property over the course of 18 months and $200K, so our CAC is $500 and improving, since maintaining the content is cheaper than creating it.

You will have a lack of precision on conversion rates and even raw visitors, so understand that we’re living in a fuzzy world.  Become comfortable with that, measure what you can, and feed what you know back into the model.

SEO Program ROI Model #1: Comparison

Let’s take a deep breath and start with something kind of easy.  You can tout ROI on your content program by comparing it against other, known CAC figures that a company is likely to use.  Take paid search, for instance.

Ahrefs has a feature called “traffic value” that is basically the measure of how much you would have to spend in pay-per-click (PPC) advertising to generate the amount of traffic that you’re getting to your site through organic search (SEO).  So let’s say you spend $100K creating content that results in ahrefs estimating that your monthly traffic value is about $200K.

You can reason through it this way:

To get this traffic over the next two years without the SEO/content program, we would need to spend $200K per month, or just under $5M over 2 years.  Instead, we generated this amount of traffic with $100K of content, making our spend on organic traffic a whopping 50 times more efficient, in terms of CAC.

While you can’t trace this back to ROI, precisely, and you don’t even really know whether any of that traffic converts, the spending efficiency here is going to get any executive’s attention.  Assuming you have even the barest hypothesis about how to convert the traffic, the improvement over a traditional, tried-and-true channel is a no-brainer.

For this reasoning to apply to your own site and business, you would need to have plans to spend money or labor on some other acquisition channel.  You could then project the relative return from SEO and evaluate it for favorability.

SEO Program ROI Model #2: Generated Revenue and Profit

I realize this content is fairly dense so far, packing an awful lot of business-ese into content that has otherwise talked about engineering or content creation topics.  To mitigate that somewhat, I’m going to explain the more difficult model, model 2, with a completely tangible instance.

My hope here is to sacrifice a bit of generality to avoid too much abstraction.  In the process, I might also show you why influencer content, affiliate marketing and other “passive income” streams are such attractive nuisances for business ownership flameouts.  But that’s coincidental collateral good only, if it happens.

Let’s talk affiliate marketing.

A Concrete Example: Hypothetical Affiliate Marketing and DaedTech

One of Hit Subscribe’s clients, Wrangle, makes approval and ticketing workflows in Slack.  I’m using them as a hypothetical affiliate example since a lot of DaedTech’s audience skews indie engineer/business-owner, and I could actually see making successful affiliate sales of Wrangle from DaedTech.

So, let’s say that Wrangle had an affiliate program and, for the sake of simple math, let’s say they paid their affiliate partners (i.e. me) a $10 commission on affiliate conversions.  Let’s establish some reasonable assumptions and then do some ROI modeling.

  • Assume (pretty conservatively) that I could bank on about 1,000 unique visitors to a given post.
  • Assume that the cost of me creating a blog post is about $500.
  • Assume that 1% of readers will click on an affiliate link.
  • Assume that there’s about a 10% affiliate conversion rate on their “add to Slack” CTA, once they hit Wrangle’s site.

This means that I spend $500 earning 1K visitors, 10 of whom will convert to Wrangle’s site, and one of whom will become a customer, earning me $10.  That’s… not great.  My ROI is a pretty depressing negative $490.

Tuning the Variables to Find ROI

As I mentioned earlier, there is a lot of assuming and back-of-the-napkin-ing with this kind of modeling.  So I like to create spreadsheets (or, these days, more sophisticated Airtable models) to turn assumptions into variables and to play with scenarios.  I then run prototyping experiments to tune up the figure and feed this data back into the model.

Let’s do a super crude version of that.

DaedTech has a domain authority of 46 or something these days, which means I can shoot a lot higher than 1,000 visitors per post.  In reality, 1K visitors per post, per month is a better figure. So let’s say that I can actually rustle up something like 25K visitors per post over 2 years.

Now, let’s also assume that I creatively incorporate multiple affiliate deals per post.  So I’m now earning $250 per post from each Wrangle affiliate link and maybe a similar figure for the other 2-3 links that I’m working in.  Now we’re up to $500 to $750 per post in earned revenue and flirting with a viable hustle.

Now I just need to commission 400-500 blog posts per year, and I’ll make a living (or write about 150 and do a lot of SEO planning).

Fixing Your Offerings and Funnel to Find ROI

So if you get really good at SEO, and absolutely grind out content, you might squeak by hawking affiliate links.

Or, you could adopt an actually viable business model.  You know, offer something of actual value to your own readership.  Then you don’t incur the falloff penalty of 99% of readers not visiting some other site, you can handle your own marketing automation, and you can sell something with much higher upside than $10.

If you can entice 10% of your own visitors into some kind of nurture sequence and convert 10% of those eventually, you’re now converting something like 250 people per post, over the lifetime of your post.  And if you’re selling something that has an LTV of, say, $100, you’re now paying back $25,000 on a $500 investment.

Granted, those conversion rates are extremely good, and you likely don’t have a site with DaedTech’s DA (or Hit Subscribe’s institutionalized SEO-fu), but this should give you a sense of how penny-ante affiliate marketing is.

Modeling and Projecting Traffic In Earnest

I think the only remaining question that I can answer in this runaway dump-truck of a “business 101” post is how you go about projecting traffic to your content.  Ideally it’s more accurate than “assume 1K visitors” or “let’s improve that to 25K visitors.”

I’m going to show you the basics of how we do this, modulo a finely-tuned sigmoid function that I’ve developed over the years.  It’s not exactly a trade secret, but I don’t have a strong incentive to just dump it here, either.

Our Standard Traffic Modeling

Take a look at the screenshot of the base here, wherein I’ve entered 3 potential DaedTech keywords.  One just seemed fun, given the subject of this post, but the other two would make excellent affiliate magnets, given that they have commercial search intent.  I could just make these hypothetical posts round-ups of affiliate links to CRMs and “apps,” whatever that means.

Here’s what you’re looking at, in terms of the columns (and ignoring synonyms and parents, which I talked about previously).

  • Volume is the global monthly search volume.
  • Difficulty is the keyword’s difficulty, as reported by ahrefs.
  • The elided formula field is DaedTech’s domain authority.
  • Projected rank is where our function projects that DaedTech would rank if I targeted this term with a post.
  • Projected traffic is the click-through I’d expect for the projected rank and this keyword’s type of search, based on standard click-through rates.  It’s also the projected traffic to my site that I would earn, monthly, for this keyword.  (It likely understates actual traffic potential for a variety of reasons beyond the scope of this post)

Building Out and Refining End to End ROI Modeling

If I were so inclined, I could add columns for conversion rates down the funnel.  I could even add budget for the content, rounding out CAC, and profitability or LTV for a given offering.

For instance, I could model ROI as it pertains to my book, Developer Hegemony, which I think you can still buy from the side gutter (or get for free by joining our community).  If I started from scratch, this modeling would rely heavily on hypotheticals.

However, I have years of data that I can draw on.  I know how many visitors I’ve had over the last 5 years and how many book purchases.  Even though that’s not the most precise data, it’s a start.

And, if I were really getting serious about this, I could take all manner of steps to run tests and improve the data’s precision.

Assume, Measure, Verify, Feedback

ROI modeling for SEO will be highly assumption-based and speculative, especially in the beginning.  And that’s true even when you’re drawing on thousands of historical examples and millions upon millions of measured impressions over the years.

For this reason, it’s not a static game or a one-and-done thing.  You begin with initial assumptions, document the uncertainty, and then just start executing.  As you execute, your data improves and you must, critically, measure that data and feed it back into your modeling for refinement.

When you do this, content creation stops being some inscrutable dark art, and it stops drawing questions like “so what do you even measure anyway, Twitter likes?”

Instead, it becomes something that you can predict, measure, and reason about.  And that reasoning can feed back so far into your core value proposition, in that it gives you insight into whether your business model itself is valid.

And with this, I bring the series to a soft end.  I may yet do a case study, and I may also do an advanced topics addendum.  But for me to have motivation to do this, I’ll rely on your collective desire (or lack thereof) to ask clarifying questions.

Either way it goes, thank you for sticking with me through this SEO for non-scumbags series.  It’s been an interesting ride.