The Consultant Who Had No Time to Publish

An independent property consultant needed a visible professional presence and had no time to build one.

Publishing consistently on a professional network is a known lever in that trade. It is also a weekly commitment that competes directly with the client-facing work that actually pays.

The ask was straightforward: make the publishing happen without it costing her a working day.

What we built

A weekly content pipeline that runs end to end on our infrastructure.

Three phases: research, drafting, then scheduling. Posts land in a PostgreSQL queue. A cron job on our VPS checks that queue every 30 minutes and publishes whatever is due, handling text, image and video posts.

Research is AI-assisted, restricted to a whitelist of sources, and governed by a hard rule against inventing data. Every factual claim in a post has to come back to a real, dated source.

Every post passes a human review gate before it is queued. Three of them, in fact: text, images, schedule, each requiring an explicit approval before the next step runs.

We want to be precise about who approves. That reviewer is us, not the client. The pipeline is automated. The judgement is not.

Then there was the website, and its problem was invisible

We rebuilt her site too. The old one had a specific, measurable defect.

It was a single-page React app. To a browser, fine. To a search crawler, it was an empty <div id="root">.

Zero pages were indexed. Not ranked badly. Not present.

The fix was to prerender at build time, emitting 87 KB of static HTML so that a crawler receives actual content instead of an instruction to run JavaScript. Combined with a sitemap, the site went from invisible to indexable.

This is the kind of defect that no amount of content strategy solves, because the content was already there. It was being served to the one visitor who could not execute it.

The numbers

From the performance report we delivered to the client for June 2026:

	April	May	June
Posts	22	21	22
Impressions	2,186	2,459	5,124
Mean per post	99	117	233
Median per post	66	57	118.5

Impressions more than doubled at unchanged posting volume. Twenty-two posts in April, twenty-two in June, and 108% more reach.

The median matters more than the mean here, and it is the number we would point at. Median per post also roughly doubled, 66 to 118.5. That means the improvement was spread across the month's posts rather than carried by one lucky item.

As a check on exactly that: June's best single post was 20% of the month's impressions, where in May one post had been 44% of the total. Growth got broader, not spikier.

Across the whole engagement, 191 posts published since November 2025, at roughly five per week.

What we are not claiming

We are not claiming she never touches anything. Our own strategy document said that, and when we checked it this week we could not support it. The publishing path does not require her, but our database only knows about posts we queued, so "never" is an absolute we cannot prove. She has also been an approver on the website work.

The provable version: the weekly publishing runs end to end on our infrastructure, and the client does not operate the queue, the scheduler, or the analytics.

We are also not claiming continuous measurement. The per-post metrics collector is manual and sparse: seven collection runs spread over about five months. Where we quote numbers above, they come from the report we actually delivered, on the date we delivered it.

The honest shape of it

This is not a case study about AI writing posts. It is about removing an operational burden: the queue, the scheduler, the token handling, the analytics collection, the monthly report assembly, and a search-engine defect that made the website unfindable.

The consultant's job was to be good at property. Ours was to make sure that showing up online did not require her to become good at infrastructure.

If this sounds familiar

The shape repeats in any small practice where visibility is a known lever and nobody has a spare day a week. The work that gets dropped first is the work with no client waiting at the end of it.

If that is the shape of your week, it is worth twenty minutes of conversation.

This case study is published anonymised. No client, brand or location is named.