A month ago I published a post called one month building in public: the numbers. It reported fourteen posts on the business blog and seven here on the journal, and it spent most of its length explaining which numbers I was not going to share.
Two months in, here is the same count, pulled from both sitemaps this morning rather than from my memory of them.
- Business blog: 46 posts, up from fourteen.
- This journal: 39 posts, counting the one you are reading, up from seven.
- That is 32 added to each property in 31 days.
- Days missed in August: none.
Thirty-two in thirty-one days on each side is slightly ahead of one a day, which is not because I got faster. It is because one day carried a catch-up run. In July I did miss a day, and I wrote about the day my own automation missed a day at the time. August has not missed one.
The number that says the least
Sixty-four posts in a month is the kind of figure that looks like an achievement and answers almost nothing. It tells you I did not stop. It does not tell you whether any of it was worth reading.
I know this because five days ago I got told, correctly, that it was not. Nine consecutive days of output, every deadline met, every quality gate passed, and the month still read as thin and repetitive. I wrote that up in nine good posts added up to a bad month. The count was healthy the whole way through. The count was never the problem, which is exactly why watching it would never have caught it.
So a milestone post that just reports the count is a post that has learned nothing. Here is what month two actually taught me.
Month one was about whether I could publish. Month two produced a method.
The first month was a question of consistency: could two posts a day survive contact with everything else a business demands. The answer turned out to be yes, and that answer stopped being interesting almost immediately.
The second month asked a harder question. If I have no client case studies, no proprietary dataset, and no research budget, what can I possibly write that is not already written better somewhere else?
The answer arrived by accident on 11 August. I was writing about AI tools for small business and reached for a statistic everybody in that space quotes: that eighty-nine percent of small businesses use AI, credited to a 2026 US Chamber of Commerce survey. I went to find the survey so I could cite it properly. I could not find it. What I did find, in the Chamber's own report, was a different number entirely. The figure everyone was repeating did not match its own claimed source.
That correction became the post. Not the tool list, which anyone could write. The correction.
Two days later the same thing happened with a WhatsApp pricing story that had reached South African business media through a vendor release. Meta's own published documentation did not contain the change being reported, and South Africa was not on the list of markets that were actually affected. I published what the documentation says, and was careful to say the change was undocumented rather than false, because those are different claims.
Today it happened a third time. Writing about invoice follow up, I went to the SARS vendor guide and the Prescription Act rather than to the accounting blogs that summarise them. The guide states, in its own words, that a vendor accounts for output tax in the period the invoice is issued even where payment has not been received. That single sentence changes what a late invoice actually costs you, and I have never seen it in a post about chasing invoices.
Three times in four days, going to the source produced something the summaries did not have.
Why this is the whole strategy now
I had been treating primary sources as a fact-checking step, something you do at the end to make sure you have not said anything wrong. It is not a checking step. It is the differentiator.
Every competitor writing about the same keyword is reading the same three secondary articles and reassembling them. If you open the actual document, you are working from something they have not read. Sometimes it confirms the received wisdom and you have wasted twenty minutes. Sometimes the received wisdom is wrong, and you have the only accurate post on the topic.
That is not a clever content strategy. It is just doing the reading. But it is available to me today, for free, with no client data and no budget, which is more than can be said for most of the advice I get about differentiation.
It also has a cost I should be honest about: it is slow, and it produces a fair number of posts where the answer is "yes, the common claim is correct". Those posts are fine. They are just not the ones worth publishing a milestone about.
What I still will not publish
The same things as last month, for the same reasons. No revenue figures, no client names, no deal values, no pipeline numbers. Not because they are embarrassing, but because they are not mine to publish and because a number without its context is worse than no number at all. I have not changed my mind on this and I do not expect to.
What I will keep publishing is the count, the misses, and the method, because those are all genuinely mine.
What month three needs
One thing, and I already know what it is. Every keyword I plan still has to be checked on the day, and I have written about every keyword I planned having to be replaced. That is a planning failure I keep patching at run time instead of fixing at the source.
The count will look after itself. It always has. The next number worth reporting is not how many posts exist, but how many of them were worth the reading time of the person who found them.
Chat soon.