Playbook 6: Growth, what one month of work looks like
One real month of Growth work: what we found, what we fixed and what the numbers did
- For
- Business owners weighing monthly website work
- You get
- The monthly routine, and one client month as it happened
After launch a website starts drifting: content goes stale, links break, a new page cannibalises an old one, a form stops being counted. Growth is the monthly routine that catches this and turns what the data shows into changes on the site. Below is how a month runs, followed by one real month as it happened, with the client's name left out.
How a month runs
The site runs in Orbit, the software we built for this work. Twice a week Orbit sweeps the site's search data, visitor and lead data, technical health, backlinks and presence in AI answers, and turns each finding into a card on a board with its evidence attached. The client can see the board and act on cards too.
Each month we go through every open card and decide. A card we work on gets fixed and closed with a one-sentence reason. A card that is a duplicate, purely informational, without search volume, or outside what the client sells gets dismissed, with the reason written on it. Cards never disappear silently. When work starts on a card, Orbit records the number it is meant to move, and weeks later shows what changed against that starting point.
The month ends with a written report to the client: what was done, what we found on the way, what the numbers did including the ones that went down, what we need from them, and what we are watching next.
One real month
The client is a Belgian sustainability group with three brands on one multilingual site, in Dutch, English and French. Its marketing lead had already handled the cards she could fix herself in the CMS and assigned the rest to us, together with six requests of her own: a missing video, a service to swap on the homepage, missing meta fields, a duplicate button, a translation bug in job titles, and documents still to be uploaded.
Working through those and the remaining cards, we found and fixed the following.
What we found | Why it mattered | What we did |
|---|---|---|
Several pages showed the group's generic title and a Dutch description in all three languages | Probably part of why some English and French pages were not indexed | Gave every page its own title and description per language, and added the fields to the CMS so the client can edit them |
More than 1,000 links inside articles carried tracking parameters | To Google each one pointed at a different URL, so the links passed almost no value to the service pages | Removed the parameters and repaired links that showed as raw code |
Related articles under every post were the same three newest posts | Readers and crawlers got no topical path to follow | Related articles now share the post's topic |
Archive pages 2 to 4 were skipped by Google | Older articles were hard to reach | Made the paginated archive crawlable |
FAQs on service pages had no markup | Search engines and assistants could not read them as questions and answers | Added FAQ markup |
Google did not know one brand was separate, or that it used to have another name | The brand was described inconsistently | Declared both in structured data and redirected the old URLs |
A long-read vision page had no next step | Readers stayed and then left | Added a call to action to the contact form |
Seven articles appeared in Google but almost nobody clicked | Their titles did not match the words people search | Rewrote the titles to the searched phrasing |
English contact requests were not counted | Button text differed from what the tracking expected | Fixed the tracking so English leads count |
Some cards claimed that no article linked to certain service pages. That was a measurement error caused by the tracking parameters: up to 20 articles linked to them. We said so in the report instead of quietly closing the cards.
The numbers went in both directions, and the report said so. Leads from one brand fell to 15 in the last four weeks against 35 in the four weeks before. We tested every form and download, found them working, and traced the drop to fewer requests on one service page while visits stayed level. Leads for a second brand rose from 14 to 22, and clicks from Google rose from 31 to 62, which is still a small number.
One recommendation went back to the client instead of being done by us: renaming a page to match the phrase people actually search, because it was her copy and her call.
What this month shows
Most of the value came from things nobody had asked for, found by looking at the data with the site's goals in mind: a thousand links passing no value, translated pages under the wrong title, leads that were never counted. The report covers what went down as carefully as what went up, because a client who only hears good news stops trusting the good news. Results take two to eight weeks to appear in search, so every change made this month is tracked against its own starting point and reported on when it lands.
Do you have a project in mind?
Let's talk about how to apply this to your own site.
