What we would have to see before publishing a thousand pages

We intend to publish programmatic pages. It would be dishonest to write a series of posts criticising how other people do it without saying that.

What we have tried to do is write down, in advance and in public, the conditions under which we would keep going. Committing to them before we have any traffic is the only time such a commitment costs nothing to make and means anything.

The schedule

Week 0    1 page. The landing page. Let it sit seven days.
Week 1    +5   prompt pages
Week 2    +8   competitor pages
Week 3    +6   comparison pages
Week 4    == GATE ==
Wk 5-8    tutorials ramp: 15, 25, 40, 60 per week
Wk 9-12   hold 60-80/week
Wk 13+    80-100/week, while indexation holds
Ceiling   ~1,200 pages in year one

Never more than 150 URLs in a single deploy.

The gate at week four

By week four there will be roughly nineteen live pages. Before publishing a twentieth, one condition has to hold:

At least 70% of those pages are actually indexed, and none are sitting in bulk as “crawled, currently not indexed”.

If they are not, the template is thin and scaling it multiplies a problem instead of a result. The rule is: fix it, do not proceed. No amount of “let’s see if volume helps”.

That second failure state is the important one. A page that is crawled and not indexed has been seen and judged not worth storing. Nineteen of those is a diagnosis.

The ongoing conditions

After the gate, three standing rules:

Indexation stays above 60%. Measured weekly, as a curve rather than a snapshot.

If it drops below 50% for two consecutive weeks, stop. Not slow down. Stop publishing, spend two weeks improving what exists, then resume at half the previous rate.

Prune quarterly. Pages with zero impressions in ninety days become candidates for removal. Removing them is a positive signal, not an admission of failure, and the same research that says so is why we are willing to do it.

The condition that actually binds

None of the above is the real constraint, and it is worth being clear about that, because the schedule above reads like the plan and it is not.

The real constraint is human review time. At ten to twenty minutes per page, and about five focused hours a week, sustainable throughput is twenty to thirty pages a week. The ramp above already sits at the edge of that.

Going faster requires either hiring reviewers or lowering the bar. There is no third option, and the second one removes the only property that would make our pages worth reading. So the honest version of our ceiling is not 1,200 pages. It is however many pages we can actually review, and if that turns out to be 300, then it is 300.

The measurement we care about

Not page count. Indexation rate, tracked weekly.

Page count is an input. It measures how busy we were. Indexation measures whether anything we published was judged worth keeping, which is the only question that matters, and it is the number we will publish when we have it.