Give me one accounting firm, 90 days, and two owner hours a week, and I would make one bet: about 20 pages answering the questions prospects ask before they hire you, written so Google can rank them and ChatGPT can quote them. I would judge the bet on enquiries. Not traffic.

I used to mine search terms in Google Ads at 1am to find the queries that produced clients. I would choose your SEO pages the same way: from call notes, inbox questions, and Search Console queries that already bring people close to enquiring. A keyword tool can wait. If the phone and contact form do not move by day 90, the plan has failed, even if impressions triple.

The two-hour limit decides what the owner does. You confirm the questions, correct the answers, and count the enquiries. Someone else prepares the pages and handles the fixes; if nobody can do that work, this is not a two-owner-hour plan. Here is what ships, what it costs against that limit, and what gets cut.

1. Days 1–14: Find questions with hiring intent

Owner cost: up to two hours a week. What it buys: a shortlist grounded in what prospects ask, plus a baseline you can use at day 90.

Start with the raw material. Export the last 90 days of Search Console queries. Skim your sent folder for answers you have written twice. Ask whoever answers the phone for the five questions that come up every week. I would sort that pile the way I sorted search terms: which questions sound like someone choosing an accountant, and which sound like someone looking for a quick definition?

The accounting SEO playbooks call this service-plus-place, service-plus-audience and service-plus-outcome targeting. Think tax accountant Leeds, accountant for tradies, or catch-up bookkeeping. I call it enquiry language. A question a prospect asked on a call beats a high-volume phrase with no sign that the searcher intends to hire anyone.

Before a new page goes live, set up enquiry tracking. Otherwise, you will spend day 90 arguing about whether the work helped. The Xero practice checklist recommends connecting GA4 and Search Console and tracking form submissions, consultation bookings and phone clicks. Do that now. Test the form yourself. Record the last 30 days of phone and form enquiries separately.

Use the first fortnight to approve the tracking events and sign off a shortlist of about 20 questions. Do not spend your two hours browsing keyword dashboards. No baseline, no honest verdict.

2. Days 15–45: Publish the first 10 answers

Owner cost: one hour a week approving drafts. What it buys: the first half of a page set built around distinct reasons to enquire.

I used to tell clients that one strong Services page could cover everything. I was wrong. Tax returns, bookkeeping, payroll, VAT and audit are different enquiries with different questions attached. Give each core service its own page. Then build pages for the audiences, software and places you actually serve: accountant for tradies, Xero cleanup for a retail shop, tax accountant Leeds.

That does not mean producing a suburb page for every postcode. The lazy version swaps the place name into otherwise duplicate copy. If two pages answer the same question in the same way, keep one. Spend the saved draft on a question your clients genuinely ask.

Each page needs an answer near the top, before the firm history and the long explanation. Aim for a short, direct block that names the service and relevant place, then explains the price, range or process where you can do so accurately. A clear block gives a search engine or AI system something it can extract. It also lets a human decide quickly whether to keep reading. Neither benefit requires a paragraph about your passion for excellence.

Your job is to read that block as a prospective client would. Correct the fee, qualification, process or exception that the drafter missed. Approve it or send it back. A quotable answer has to be a correct answer first. Ten useful pages by day 45 beat 20 pages you have not had time to check.

3. Days 30–45: Fix the path from page to enquiry

Owner cost: one hour total for access and approvals, taken from the second weekly hour. What it buys: a site people can find and contact you through.

For a small accounting site, keep the technical list short. Make each new service page indexable and link it from a Services hub. Give each page a title that names its service and relevant place or audience. Compress the hero images slowing the mobile page. Add LocalBusiness and Service schema using the firm’s real details. Most urgently, test the contact path on a phone. I have watched firms debate blog design while their contact page made the phone number hard to tap. Fix the tap.

Hand over Search Console and Google Business Profile access once, then approve a straightforward review-request email. Whoever executes the work can check the firm’s name, address and phone against its footer, contact page and relevant listings. Professional directories and review sites matter more here than a pile of unrelated directory submissions. Use the associations the firm belongs to; do not claim a listing it does not have.

This is not a second project competing with the pages. It is the route by which a reader finds a page, trusts the firm enough to act, and reaches the form or phone number. Fix broken contact paths before polishing another draft.

4. Day 45: Make the continue, adjust or stop call

Owner cost: the second weekly hour. What it buys: a decision while there is still time to change the next 45 days.

Open one sheet and compare three things with the starting point: phone and form enquiries, Search Console impressions for the target questions, and Business Profile calls and direction requests. Enquiries make the decision. Impressions and profile actions help diagnose it; they do not win the argument by themselves. SEO may take longer than this to show its full effect, but day 45 is still too late to discover that nobody recorded the calls.

Choose one of three moves:

  1. Continue if enquiries from the new work are appearing. Ship the remaining questions and keep the tracking intact.
  2. Adjust if the pages are getting seen but enquiries are flat. Check the top answers, call buttons and form. Make it easy for a stranger to enquire before asking for more content.
  3. Stop the page queue if there is no useful signal at all. Check indexing, the questions you chose and the tracking before publishing page 11. That is a pause to find the fault, not a declaration that SEO must work in 45 days.

You still have two owner hours this week. Use the approval hour to settle any disputed answers and the numbers hour to pick one move. What the deck calls “patience” can sometimes mean paying for more pages before anyone checks whether the first ones are reachable.

Accountant reviewing an enquiry tally on a notepad beside a laptop, with a phone nearby

5. Days 46–90: Finish the pages, then check AI visibility

Owner cost: one hour a week on answers and one on measurement. What it buys: about 20 approved pages by day 60, followed by time to find and fix weak spots before day 90.

If the day-45 decision is continue, ship questions 11 to 20. If it is adjust, spend a fortnight repairing the best-seen pages and their contact paths, then resume the queue. If it is stop, resolve the tracking, indexing or intent problem first. Twenty is the working target, not a reason to publish pages you know are wrong.

Once the pages and firm details are in place, check what a prospect might see in AI search. Write 10 hiring questions, not 10 prompts designed to flatter an SEO report: best accountant for a tradie in Leeds, how much does catch-up bookkeeping cost, Xero accountant for a dental practice, do I need an audit. Run the same questions in ChatGPT, Gemini and Google AI Overviews once a week. Record whether the firm is named.

That log is a diagnostic, not a second definition of success. A mention that produces no enquiry does not rescue the plan. A missing mention tells you where to look: the answer block, the Business Profile, the consistency of the firm’s details, relevant third-party listings, or whether AI crawlers can access the pages. Check those before buying another visibility dashboard or writing page 21.

A firm owner typing a hiring question into ChatGPT and checking whether their firm is named

Do not expect every prompt to name you on a schedule. Use the weekly checks to find avoidable gaps, then return to the enquiry sheet. The test is useful only if it changes what you fix.

6. For all 90 days: Cut work that cannot earn its time

Owner cost: zero hours on the cut list. What it buys: room for the questions, approvals and measurement that fit the constraint.

I would cut weekly tax-news recaps and Budget reactions. They may attract attention briefly, but this plan needs answers to the questions people ask before hiring. I would cut suburb pages made by swapping place names into the same copy. I would cut link packages promising 50 links a month; relevant association and review listings are a better use of this limited effort than links no prospect will see.

Concretely, none of this ships in my 90 days:

  • No rebrand, rebuild or new blog theme. Fix the clicks and mobile load on the site you have.
  • No keyword-volume chase. A large search count does not turn an informational query into a hiring question.
  • No traffic report presented as the result. Use impressions to diagnose. Count calls, forms and booked consultations to judge the work.
  • No AI visibility subscription yet. Run the 10-prompt log first; this 30-day check shows the kind of manual test I mean.

Cuts are not an apology for a small plan. Every hour spared from generic content can go into an answer that helps someone enquire.

7. Each week: Spend the two owner hours on truth and counting

Owner cost: two hours, not two hours plus a meeting. What it buys: accurate pages and an enquiry count you can act on.

The allocation stays simple:

  1. 30 minutes, twice a week: approve answers. Read the drafts waiting for you. Fix facts, fees and process steps. Approve or reject; do not let a month’s worth pile up.
  2. 30 minutes on Friday: count enquiries. Put phone and form enquiries beside the baseline. Note which question or page brought someone in when you can tell.
  3. 30 minutes on Friday: investigate one signal. Early on, use it to check tracking, the contact path or a recurring call question. From day 60, use it for the 10 AI prompts and send any relevant misses to whoever owns the pages.

A page getting impressions but no enquiries may need a clearer answer or an easier contact path. A question that keeps coming up on calls belongs on the next list. Owners trying to do all of SEO in two hours usually lose those hours inside tools. Spend them on what the firm knows and what the firm can count.

When the constraint lifts at day 90, use the enquiry record to choose what happens next. If enquiries rose, keep the system running and add the next 10 questions. Revisit the review and association work you could not finish. If the work is producing enquiries but you want your two hours back, hand the same question list, approval standard and enquiry sheet to someone who can run it daily. That is what hands-off SEO should mean: the owner sets the truth and the target; someone else ships and fixes the work.

groas can run that SEO and AI search work for a flat fee, with a named strategist accountable for direction rather than another monthly report about clicks. It is the same rule I would apply to a 90-day professional-services lead-quality fix: keep what produced qualified enquiries, cut what did not, and spend the extra time on the next question a client asks.