Skip to content
Curious Context Labs

case study · Marketing · Agent

Growing a service business website with pages people actually search for

How search data, staff know-how and a careful review step can produce new and refreshed pages every month.

business
Accounting and law firms
build
Agent
timeline
3 weeks to first pages
read
3 min

01 the business

Accounting practice

30 staff, two offices

The practice's website has eight thin service pages written five years ago. Rankings have slipped, and the partners only notice when enquiries slow down. Nobody has time to research keywords or write.

02 before

The problem

Most small business websites have a handful of thin service pages written years ago. Agencies are expensive and in-house staff have no time to research keywords, write, and keep pages current. Rankings fade slowly and nobody notices until the phone goes quiet.

  • No page answers common questions people search for, such as how to register for a tax or change a year end.
  • Old pages quote outdated thresholds and deadlines.
  • The last agency quote for content was more than the marketing budget.

03 the build

What the AI does

  1. Step 1: Reviews Google Search Console and Semrush data to find pages losing clicks and searches you do not yet have a page for.
  2. Step 2: Proposes a short list of new or refreshed pages, each with a target search term and the question it should answer.
  3. Step 3: Drafts the page from your service details, FAQs and team knowledge, with headings, internal links and a meta description.
  4. Step 4: Rewrites outdated sections of existing pages and notes exactly what changed and why.
  5. Step 5: Loads each draft into WordPress or Webflow as an unpublished revision for review.

paused: waiting for a person to approve

How it was put together

  • Reviewed Search Console and Semrush data to find falling pages and missing topics.
  • Proposed a monthly list of pages, each with a target search and the question it answers.
  • Drafted from the practice's own service notes and short interviews with the partners.
  • Loaded drafts into WordPress as unpublished revisions, with a note of exactly what changed.

04 human in the loop

Where a person stays in charge

You approve the page list before any writing starts. A person who knows the service checks every draft for accuracy and publishes it, especially pages that mention fees, regulated services or medical and legal advice.

  • The partners approve the page list before writing starts.
  • A qualified accountant checks every figure, deadline and piece of advice before publishing.
  • Pages are written for readers first, in line with Google's guidance on helpful content.

05 rollout

From kickoff to live in 3 weeks to first pages

  1. Week 1Audit the site and search data; agree the first ten pages.
  2. Weeks 2–3Draft, review and publish the first batch.
  3. Month 2 onwardsPublish four to six new or refreshed pages a month and track results.

06 after

What changes

The website grows steadily with useful pages that answer what customers search for. Old pages stay current, and staff contribute their expertise by reviewing rather than writing from scratch.

worked example · your numbers will differ

Assume a new page takes 4 hours to research and write by hand and 1 hour to review as a draft, at six pages a month.

6 × 4 hours = 24 hours of writing a month, against 6 × 1 hour = 6 hours of review. The partners give 6 hours instead of 24.

What to measure

  • Organic clicks from Search Console.
  • Enquiries from organic search.
  • Pages refreshed per quarter.
  • Reviewer time per page.

07 evidence

What the research says

  1. Google rewards original, high-quality content however it is produced, but using automation mainly to manipulate rankings breaks its spam policies.

    Google Search Central, Google Search's guidance about AI-generated content (opens in a new tab) · 2023

  2. In a study of 453 professionals doing writing tasks from their own jobs, access to ChatGPT cut the time taken by 40% and raised quality, as rated by evaluators, by 18%.

    Science / MIT, Experimental evidence on the productivity effects of generative AI (Noy and Zhang) (opens in a new tab) · 2023

08 lessons

Mistakes to avoid

  • Publishing in bulk. Dozens of thin pages look like spam and help nobody.
  • Skipping expert review on regulated topics.
  • Never updating. Tax and legal pages go stale every year.

09 questions

Common questions

Does Google penalise AI-assisted pages?
Google says it rewards helpful, original content however it is made, and treats content made mainly to manipulate rankings as spam. Expert review is what makes the pages helpful.
How soon do results show?
Search results usually take weeks to months to move. Track clicks monthly, not daily.

keep reading

All case studies ->

Tell us how your business runs. We'll tell you where AI fits.

A free 30-minute call. You talk, we ask questions, and you leave with a few concrete ideas — or an honest answer that AI isn't the right fix yet.

Working with teams around the world.