Skip to main content

Checklist

Is AI SEO worth it for your business? A readiness checklist

How to decide whether AI search visibility work is worth investing in now: the businesses it tends to suit, the ones it does not, what has to be in place before starting, and a six-point self-score. Written to disqualify as readily as to qualify.

Author
Dorian Menard, founder of Search Scope
Published
Reading time
6 min

The short version

TESTED AI-assisted search is a measurable share of how people search: Pew found 18% of US Google searches produced an AI summary in March 2025, and fewer clicks when one appeared. INFERRED Whether that makes AI SEO worth paying for depends less on the trend than on the business: whether people research it before contacting it, whether it has something independent sources can corroborate, and whether someone inside it can supply real expertise. This page is written to disqualify as readily as to qualify, because the common way budgets get wasted is a business starting layer six work with layer one broken.

Who it tends to suit

  • Established businesses that are researched before contact. Professional services, health and allied health, finance, B2B consulting, considered purchases. The categories where a buyer asks an assistant to shortlist before making a call.
  • Businesses with a real footprint to corroborate. Registry records, association memberships, some independent mentions. Corroboration work needs something to work with.
  • Businesses whose facts are inconsistent. Mismatched names, addresses and service lists across the web are the most fixable problem in the field and often the highest-return one.
  • Product businesses with genuine expertise. Stores that can say something original about how their products are used, rather than repeating supplier copy.

Who it does not suit yet

  • Pre-revenue businesses with no public footprint. Independent sources need something to corroborate.
  • Businesses running generic supplier or template content. Restated content gives a system no reason to prefer it.
  • Anyone wanting guaranteed citations or fast results. DOCUMENTED No platform documents a way to guarantee inclusion; Google says outright that serving is not guaranteed. A guarantee is a sign of a loose definition or a different metric being billed.
  • Businesses that want volume AI content. Generic material produced at scale gives an assistant nothing to cite and is the pattern Google’s spam policies describe.
  • Businesses with no one to answer questions. If nobody internally can spend an hour a month on what the work is actually like, the content will be indistinguishable from every competitor’s.
  • Businesses whose organic foundation is broken. On Google, AI features draw on the Search index. A site that does not rank has an ordinary SEO problem to fix first.

What has to be in place before starting

Score one point for each you can honestly claim.

  1. A budget for sustained work. Entity and corroboration work surfaces over months, not weeks. A three-month trial cannot tell you whether it worked. Perth price points are on the cost page.
  2. A named subject-matter person. Someone who knows how the work is done and will answer detailed questions.
  3. Access to the website. A developer available, or full control of the CMS, so technical changes do not queue for months.
  4. Access to Search Console and analytics. Without them there is no baseline and no measurement.
  5. An existing public footprint. Registry records, some listings, some mentions, an association or licence.
  6. Tolerance for honest reporting. Including “not observable” as an answer, and directional rather than precise numbers.

Five or six: the sensible next step is diagnosis, not building. Three or four: fix the missing items over the next month first. Two or fewer: the timing is not right; spend the year on the offer, the footprint and the foundations, and the platforms will still be there.

“Not yet” is a real answer

INFERRED Being told to wait is cheaper than six months of work on foundations that were never going to hold, and it is a legitimate outcome of any honest first conversation. If you score well, the audit checklist describes what a credible diagnosis looks like before any content gets commissioned.

What is documented and what is inferred

  • Documented: Google’s no-guarantee statement; the Search index as the substrate for its AI features.
  • Tested: Pew’s US frequency and click figures.
  • Inferred: the fit and non-fit lists; the six-point score.
  • Conflict of interest: the publisher of this site sells the work this page helps you decide about.

Sources

  1. AI features and your website, Google Search Central (read 9 September 2026) DOCUMENTED
  2. Google users are less likely to click on links when an AI summary appears in the results, Pew Research Center, 22 July 2025 (read 9 September 2026) TESTED