Every vendor in this category quotes a return. Almost none of them show the calculation, and the ones that do leave out the three costs that decide whether the number is real.
This page is the calculation. Four formulas, two worked examples with unflattering numbers, and the payback periods you should actually expect.
The four numbers you need before you start
Your conversion rate from organic traffic. Not your sitewide rate. Organic blog traffic converts differently from branded homepage traffic, usually much worse, and using the wrong one inflates everything downstream.
Your customer value. Average revenue per customer over the period you are measuring, or lifetime value if you have enough history to know it. Gross, not net, then apply margin at the end.
Your true content cost. Including the review time, not just the production cost.
Your time horizon. SEO returns arrive on a six to eighteen month curve. Any ROI calculated over one quarter will be negative and will tell you nothing.
Formula 1: return per published article
- Monthly value = monthly organic sessions x conversion rate x customer value x margin
- Payback months = total cost of the article divided by monthly value
The number people get wrong is sessions. Use the realistic number for a page ranking on page one for a mid-volume term, which for most B2B queries is 30 to 150 sessions a month, not the keyword's search volume.
Formula 2: the portfolio version
Individual articles are the wrong unit. Content returns follow a power law: most pages return nothing and a small number carry the whole programme.
- Portfolio value = (articles published x hit rate) x average monthly value of a hit
- Portfolio cost = articles published x cost per article
Hit rate is the variable that matters most and it is the one nobody publishes, so here is ours. Across 577 blog URLs on this domain over the last 90 days, 26 percent received at least one click and 6 percent received five or more. The top 5 percent of pages hold 68 percent of all blog clicks.
Model your hit rate at 25 percent and be pleased if you beat it. Any vendor projection that assumes every published article performs is wrong by roughly a factor of four.
Formula 3: cost per acquired link
- Cost per link = (subscription + time spent + assets produced) divided by links acquired
- Value per link = only measurable at the campaign level, never per link
Nobody can price a single link honestly. What you can measure is whether a set of pages moved after a set of links, and what that movement was worth. Treat per-link value claims as marketing.
Formula 4: the payback horizon
| Period | What happens |
|---|---|
| Month 0 to 3 | Cost only, no return |
| Month 4 to 6 | Early rankings, minimal traffic |
| Month 7 to 12 | The curve starts, 20 to 40 percent of steady state |
| Month 13 to 18 | Steady state |
| Month 19 onward | Maintenance cost only, returns continue |
Everything before month seven is investment. A programme judged at month four is always a failure, and this is the single most common reason content programmes get cancelled just before they would have worked.
Worked example 1: the modest one
A B2B SaaS at 200 dollars a month per customer, organic conversion of 1.2 percent, 40 articles published over a year at 120 dollars each including review time.
- Cost: 4,800 dollars
- Hits at 25 percent: 10 articles
- Average traffic per hit: 60 sessions a month
- Total monthly sessions at steady state: 600
- Conversions: 7.2 a month
- Monthly revenue added: 1,440 dollars
- Payback: month 3 of steady state, which is roughly month 16 overall
The uncomfortable part: thirty of those forty articles returned nothing and still cost 3,600 dollars. That is not failure, it is how the portfolio works, and any model that assumes every article performs is wrong by a factor of four.
Worked example 2: the one that does not work
A local service business at 400 dollars per customer, organic conversion of 2 percent, publishing 20 articles a year at 300 dollars each.
- Cost: 6,000 dollars
- Hits at 25 percent: 5 articles
- Average traffic per hit: 25 sessions a month, because local query volume is small
- Total monthly sessions: 125
- Conversions: 2.5 a month
- Monthly revenue added: 1,000 dollars
- Payback: month 6 of steady state, roughly month 19 overall
The honest read: this business would return more from the same 6,000 dollars spent on local search presence and reviews. Content volume is not the lever when the total addressable query volume in your area is a few hundred searches a month.
The three costs vendors leave out
Review time. Every automated article needs a human pass for accuracy. At 20 minutes a page and a realistic hourly rate, that is often larger than the generation cost. Any ROI model that omits it is understating cost by half.
Maintenance. Published pages decay. Budget one update per new page past 50 published, or watch the portfolio value fall while the page count rises.
The cost of being wrong. A fabricated statistic or a recommended product that does not exist costs credibility with readers and with the AI engines citing your domain. It is real, it is hard to price, and it argues for the review time above.
What automation actually changes in the maths
It changes cost per article, not hit rate. A pipeline that produces articles at 20 dollars instead of 200 makes the portfolio maths work at a lower conversion value, which is the entire argument for it.
What it does not change: whether the topic has demand, whether your page is better than the ones ranking, or whether the reader converts. Those are unaffected by how the draft was produced.
Where automation makes it worse: publishing four times as much into a topic that supports a fraction of that produces pages Google declines to index, and a rising unindexed share drags what does rank.
The link side of the calculation
Links are not a separate programme with a separate return. They are what determines whether the content you paid for reaches page one, which means their value shows up entirely in the content numbers above.
The practical model: if your pages sit at positions 11 to 25 with reasonable content, links are your constraint and spending there moves the whole portfolio. If your pages are not ranking at all, links are not the problem and buying them wastes the budget.
A moderated exchange changes the cost side rather than the value side, because the cost per acquired link falls sharply when you are not paying for outreach time. Our backlink exchange works on that basis. Our comparison of backlink exchange platforms sets out who runs them and on what terms.
Honest limitation: exchange links are the cheapest links to acquire and not the best ones. In a model, price them as what they are, a way to reach the ranking threshold faster, not as a substitute for the earned coverage that holds a position afterwards.
The three-question test before you spend
Does the query volume exist? Add up the monthly searches for every term you could realistically target. If the total is under a few thousand, no content budget produces a large return.
Can you hold the quality bar at the volume you are planning? If not, reduce the volume. The portfolio maths gets worse, not better, when hit rate falls.
Can you wait eighteen months? If the answer is no, this is the wrong channel and paid acquisition is the right one. That is a legitimate answer and it is better to reach it before spending.
If the answers are yes, our SEO content platform is built to move the cost-per-article side of the equation.
Related reading
For what a lead should cost you before any of this, see lead generation cost and how to calculate cost per lead. For the link side, see how to build a backlink network.
FAQ
What is a good ROI for content marketing? Three to five times the programme cost by month eighteen is a reasonable target for most B2B businesses. Anything quoted above ten times is either an outlier, a very high customer value, or a model that omitted review and maintenance costs.
How do I attribute revenue to a blog post? Imperfectly, and that is fine. Use last non-direct click as a floor and assisted conversions as a ceiling, then report both. A single attribution model presented as truth is the least honest thing in most marketing reports.
Is automated content cheaper than a freelancer? Per draft, yes, by a large multiple. Per published page that ranks, the gap narrows because review time is the same either way. The real advantage is volume and consistency rather than headline cost.
When should I stop and cut my losses? Month twelve with no page on page one for any target term, having published at least 30 articles. That is a demand or a quality problem, and continuing to publish will not resolve either.