A proposal can contain three subscriptions, dozens of articles, and a weekly report without saying who will fix an incorrect service page. Start your 2027 budget with the decisions and changes you need someone to own.
This guide offers a way to scope the work. It is not a market price survey, an Appear price list, or a forecast of returns. The right spend depends on your website, available staff, business economics, and the work already covered by other suppliers.
Write down what you already pay for
List your current SEO, content, website maintenance, analytics, and agency commitments. Put a person's name beside each responsibility. Include internal review time, since your subject experts still need to confirm facts even when an outside team writes the page.
Ask whether existing contracts already cover technical fixes, service-page updates, customer studies, and reporting. Google describes its AI search optimization as part of SEO. A supplier should explain what additional work its AI visibility fee buys.
Price the diagnosis before the publishing volume
Choose one revenue-relevant offer and inspect the path from discovery to enquiry. Review a few important pages, the questions sales hears, and the evidence currently available. A broken contact form deserves a different response from an unanswered comparison question.
Ask for a written diagnosis with example URLs and recommended actions. Each action should have an owner, an estimated effort, and a way to check completion. This lets you compare proposals that otherwise use different names for similar work.
- Access problem. Pay for a specific website fix and verify the affected page afterwards.
- Information gap. Commission the missing facts and an appropriate page update or new article.
- Weak evidence. Fund a customer interview, demonstration, or documented project before paying for more general copy.
- Measurement gap. Define a repeatable baseline and connect enquiries to your existing reporting.
A worked example using 100 hours
Suppose a small service business can allocate 100 combined staff and supplier hours to a first 90-day cycle. The following is an illustrative allocation, not a recommended industry percentage or a promise of results. Replace the estimates after inspecting the site.
- 15 hours for the baseline. Review buyer questions, current pages, search data, and the contact path. Agree the questions and outcomes to track.
- 20 hours for priority fixes. Correct access failures, inaccurate business details, and problems on the enquiry path.
- 35 hours for useful content. Improve core service information and produce one substantial proof asset. Include interviews, fact review, editing, and publishing.
- 15 hours for measurement. Repeat checks, inspect answers and citations, and review qualified enquiries.
- 15 hours held back. Use this for findings from the first checks, or leave it unspent if further work has no clear justification.
Turn that allocation into a cash budget
Multiply each supplier's agreed hours by its quoted rate, then add any fixed fees and approved usage charges. Track internal hours separately so a low external bill does not hide a large burden on your team.
For each subscription, ask what limits apply: websites, users, questions, engines, check frequency, content, and exports. Ask what happens at the limit and what would trigger extra spend. A quote that excludes publishing or review may cover less work than it first appears.
If a supplier sells a monthly page allowance, specify what qualifies as a new page, what counts as an update, who approves facts, and what happens to unused capacity. Keep the publishing commitment separate from claims about ranking or lead growth.
Choose software, services, or a combination
Software can be appropriate when you already have people who can investigate findings, edit the website, and review content. Ask the team to demonstrate that workflow on one real issue before buying a large annual commitment.
A service can be useful when ownership or execution capacity is missing. Check whether the scope includes research, fact gathering, editing, publication, and verification. If you use both software and a service, agree which report is the record you use to make decisions.
Compare the total work completed and retained assets. You should know where the published pages live, who owns the content, and what evidence you can export if the relationship ends.
Set a review decision before the first invoice
At the end of the first cycle, review what shipped, whether priority pages work, whether answers became more accurate, and whether qualified enquiries changed. Save the underlying records so you can question the explanation.
Continue when there is useful progress and a credible next action. Change the plan when the diagnosis was wrong or execution failed. Pause expansion when the only reported gains are more pages or more bot requests.
For business planning, calculate how many incremental customers would cover the cost using your own contribution margin. Treat that as a break-even scenario. It is not a forecast that AI visibility work will produce those customers.
Common questions
What percentage of marketing spend should go to AI search?
There is no universal percentage supported by the evidence in this guide. Start with a defined business need and the cost of addressing it. Protect channels that already produce profitable customers while you evaluate new work.
Should SEO and AI visibility have separate budgets?
You can report them separately, but avoid paying twice for the same page edits, technical fixes, and evidence gathering. Ask each supplier to identify the work it owns.
Can I begin with a small budget?
Yes. Pick one offer, improve the most relevant existing pages, and establish a modest repeatable measurement process. Expand only when you have capacity and a specific reason.