How Steward decides what to fix next.

Monthly website maintenance without a system for deciding what to work on is just busywork with a retainer attached. You pay someone every month, they make some changes, the site looks slightly different, and neither of you knows whether any of it helped. This is the standard model for "website maintenance" in most agencies, and it is the thing Haus Steward is specifically designed not to be.

Here is the exact process we use each month to look at real data, find what moved, and decide which fix earns its slot in the queue — before anything is proposed for your approval.

Start with Search Console, not assumptions

The first step every month is pulling the last 28 days of data from Google Search Console. Search Console shows what queries people are actually using to find your site, how many times your pages appeared in search results, how many times someone clicked through, and what your average ranking position is for each query.

This data drives everything else. Without it, a monthly SEO recommendation is an educated guess. With it, the recommendations are specific: this page dropped from position 8 to position 14 for this query; here is what the page currently says and here is what competitors on page one say; here is the proposed change.

Most small business website owners never look at Search Console. Most monthly SEO retainers do not require them to share access. Steward is built around this data — we request read access at the start, and we check it every month before doing anything else.

Identify what moved — in either direction

The most useful signal in Search Console is movement, not absolute position. A page that has been sitting at position 22 for three months is not interesting. A page that moved from position 22 to position 11 last month — appearing in front of twice as many people without any change to the page — is very interesting. Something is working, and understanding what is working tells you what to do more of.

The same logic applies in reverse. A page that dropped from position 6 to position 19 needs attention immediately, before the traffic loss compounds. Drops that happen slowly are easy to miss if you are not looking at monthly data — they just feel like "things slowing down" without a clear cause.

We categorize every page in the top 50 results as gaining, holding, or losing, then focus the month's work on the three or four that show the most meaningful movement in either direction.

Check what Google Analytics shows about behavior

Search Console tells you how people find the site. Google Analytics tells you what happens after they arrive. The combination matters. A page with high clicks but high bounce rate — people click from Google, then leave immediately — is a page that matched the search query but failed to deliver on it. That is a content problem, not a ranking problem.

We also check which pages are getting traffic that was not in the original plan. Sometimes a page ranks for a query no one intended it to rank for, and there is an opportunity to build on that by expanding the content, adding a dedicated page, or linking to related services. Organic search is partly strategy and partly paying attention to what is already working.

Build the proposed queue

With the data reviewed, we build the month's proposed queue. Each item in the queue is a specific action with a specific rationale:

• Content updates — expanding a page that is close to ranking but thin; adding a section for a service that is showing up in queries but not well-covered; updating pricing or seasonal information that may have changed
• Technical fixes — a page returning a slow load time in PageSpeed Insights; a missing meta description; an image without alt text; a redirect that is not working correctly
• New content — a blog post targeting a query with meaningful search volume that the site does not currently answer; a new service page for an offering that keeps coming up in client inquiries
• LinkedIn or social drafts — one or two posts connecting recent business activity to a broader point, drafted for your review and scheduled if you approve

The queue never has more items than can reasonably be done in a month. A list of twenty things is not a plan — it is a way to feel productive while avoiding the harder question of what actually matters most right now. We keep the queue to five to eight items, ordered by expected impact.

The approval step

The proposed queue goes to you via Telegram, described plainly, one item per numbered message. You approve, modify, or skip each item. Nothing goes live without a yes on that specific item. Items you skip stay in the candidate list for the following month — if the data still supports them, they come back. If circumstances changed, they do not.

This is the part of the process that most automated SEO tools skip. They generate the action item and implement it. We generate the action item and ask you. For content that goes under your name and affects your business's reputation in search, that step is not optional.

Measure, then repeat

At the end of each month, we close the loop: which approved items were completed, which queries moved, what the net traffic change was. This is a short summary — not a PDF report with forty slides — delivered the same way everything else is, in Telegram.

The goal over time is a compounding record: month one established baseline, month two fixed the technical issues, month three started adding content, month six we can see which content is moving and which is not, month twelve we can trace a line from the changes we made to the ranking positions and traffic they produced.

This is what ongoing website management should look like. Not "we made some updates last month" — but "here is exactly what we changed, here is what the data showed, and here is what we are doing next."