How much should a small business spend on SEO when AI-assisted Search changes the way people encounter web links? The answer cannot be a universal percentage of revenue. A firm with high repeat-purchase margins and a reliable booking process has different economics from a new retailer with thin margins or a consultancy that can serve only a handful of clients. Google AI Search announcements also do not establish that every business has lost the same proportion of clicks. A sound spending decision begins with unit economics, realistic implementation capacity and the quality of the customers the website can attract. It then measures whether the work solves valuable problems rather than simply purchasing more activity.
Begin with the business objective, not an agency percentage
An SEO budget should support a real commercial or operational goal. A local service business may need more qualified enquiries from its existing coverage area. An ecommerce retailer may need discoverable product categories and fewer abandoned customer journeys. A professional services company may need credible information that helps prospects decide whether they are eligible for a consultation. The expected value of each outcome differs, and so should the way spending is evaluated.
Write down the current problem before buying an SEO package. Is an important service page inaccessible? Is the website attracting irrelevant questions? Does the business have suitable content but a broken form? Are staff repeatedly answering a technical question that a reliable guide could resolve? These issues suggest different workstreams and costs. A budget chosen merely because another organisation spends an attractive percentage of turnover can overlook the work that would help most.
The old headline associated with this article references dramatic spending and click figures, but such numbers should not be adopted as benchmarks without the original study, sample, definitions and market. A percentage of revenue can be calculated after choosing a budget; it does not itself prove the budget is suitable. Use the actual business model rather than an unexplained industry average.
Estimate what a genuinely qualified customer is worth
Customer value depends on revenue, gross margin, fulfilment cost, retention and the probability of a lead becoming a sale. A £1,000 sale is not equivalent to £1,000 of profit. A project may require substantial delivery time or materials, while another service may have recurring revenue over several months. Ask the finance or operations team for realistic figures and label unknown assumptions clearly.
A useful planning example can remain hypothetical. Suppose an enquiry has a reasonable chance of becoming a customer, and the expected gross contribution of that customer is sufficient to cover acquisition and service costs. The SEO team can then ask what an incremental qualified enquiry might be worth. But it must not substitute a click, page view or email address for a qualified opportunity. If the business cannot track which enquiries are genuinely appropriate, improving that measurement may be a higher priority than buying more content.
Consider constraints as well as value. A small consultancy with a full diary may benefit more from improving the quality of enquiries than from increasing their number. A retailer with inventory shortages should not fund traffic campaigns to unavailable products without correcting the customer experience. A budget has to fit operational capacity to deliver what the website promises.
Separate SEO implementation costs from ongoing services
SEO spending can include a technical audit, developer implementation, editorial research, subject expert review, media licensing, analytics configuration and ongoing maintenance. An agency's fee may cover only some of these. Ask what the proposal actually delivers and who is responsible for deploying recommendations. An audit document that remains unimplemented can be valuable diagnosis, but it should not be reported as a completed technical repair.
Break the programme into a finite initial phase and recurring work. The initial phase might correct broken navigation, inaccessible content, misleading metadata or an important conversion flow. Recurring work may maintain service information, publish carefully researched guides, monitor search behaviour and test changes after CMS releases. These are different cost patterns. A business should avoid agreeing to indefinite monthly article production when its current bottleneck is a developer task or a faulty form.
Document ownership of accounts, content and technical changes. The business needs appropriate access to its own domain, Search Console and analytics where applicable. Contract terms should explain cancellation and handover. A proposal that claims guaranteed rankings or AI citations is promising an outcome an external agency cannot control. Google Search Central's guidance on hiring SEO help is a useful starting point for these conversations.
Evaluate customer acquisition cost with honest denominators
Customer acquisition cost can be defined by dividing relevant acquisition expenditure by the number of customers acquired during a comparable period. The challenge is deciding which costs and customers belong in that calculation. If a business counts an SEO invoice but excludes implementation time, its estimate may understate the investment. If it counts every form submission as a new customer, it will overstate the outcome. Set definitions before calculating a ratio.
SEO often influences more than one step of a buyer's journey. Someone may discover an educational article and return later through a branded search or direct visit. Attribution systems may not capture every influence, and privacy settings can limit observations. Do not fill these gaps with invented certainty. A practical report can show verified acquisition cost, qualified leads and later customer outcomes separately, with clear notes about incomplete attribution.
Compare alternative acquisition methods only with compatible assumptions. Paid advertising expenditure, referral fees, email campaigns and SEO editorial investment may have different time horizons and reporting coverage. One cannot simply compare the price of a click with the cost of a completed client engagement. Use a consistent outcome and period wherever possible; otherwise state why the metrics are not directly comparable.
Consider profitability and cash flow before increasing the budget
A growing business may be willing to spend more on acquisition, but its cash position and expected time to recover the cost matter. SEO improvements can take time to be discovered, indexed and reflected in user behaviour. No provider can guarantee the precise date of a search ranking change or the conversion value of a new article. A business that needs immediate cash receipts should not treat a speculative organic traffic forecast as equivalent to signed customer orders.
Set a budget the organisation can sustain whilst testing its assumptions. Include necessary development, maintenance and content verification, not just marketing retainers. For businesses with limited reserves, prioritise repairs that directly prevent customers from completing a valuable task. If a service enquiry page is broken, the business may recover useful functionality without undertaking a large content campaign.
A scenario analysis can be more useful than a single promised return. Describe a conservative case, a plausible working case and a favourable case using explicitly labelled assumptions. Do not publish hypothetical conversion rates as measured results. Include the possibility that a particular initiative produces little observable movement despite correct implementation. That risk should be part of the purchasing decision, not hidden in small print.
Understand the AI Search click question without inventing a trend
Google's AI Overviews and AI Mode provide different ways for people to explore information. Their availability and presentation can vary by query, product and market. A claim that AI Search reduced all small-business clicks by the same percentage is not justified by a general product announcement. A site can see changes in informational queries while commercially important service or product searches behave differently.
Review the relevant Google Search Central AI features guidance and separate it from independent studies or commentary. If a report supplies a click-change statistic, inspect its sample, period, definition of search exposure and coverage of AI features. Do not apply a publisher-wide average to a local business without understanding the differences. A headline about AI can be a reason to investigate data, not a replacement for it.
For a small business, ask which queries matter commercially. A guide answering a broad definition may generate many informational visits without immediate revenue. A page describing a real specialised service may attract fewer but more appropriate prospects. Losing some irrelevant clicks need not have the same meaning as losing enquiries from high-intent searches. Budget decisions should respect those distinctions.
Segment Search Console observations by purpose
Search Console can report impressions, clicks, average position and other dimensions for a verified website, subject to its limitations. Group pages into important services, products, informational resources and other relevant categories. Compare suitable periods and note major site releases or changes in demand. Do not infer one universal AI impact from the whole-site click total. A loss on an old informational article may dominate the graph while the core customer journey remains stable.
Examine query types where reporting permits. A branded search, a broad explanation and an explicit service-selection query can have different intent. Record actual observations rather than inventing data to populate a dashboard. If the website has too few impressions for a meaningful comparison, acknowledge that uncertainty. A small sample can change substantially when just a handful of visitors behave differently.
Keep Google Search metrics separate from analytics events and sales records. A click is not a qualified lead, and an enquiry is not necessarily a customer. Differences between reports may arise from measurement scope and configuration. The spending decision should connect these evidence sources carefully rather than adding incompatible totals together.
Prioritise repairs with observable acceptance tests
Before commissioning speculative AI-optimisation content, inspect whether existing important pages actually work. Broken internal links, inaccurate service descriptions, missing mobile content, incorrect redirects or a failed form can have a direct cost to customers. A technical audit should provide the exact URL, observed defect, affected user task, responsible component and verification method. Work that can be shown to remove a serious barrier is often a sensible early investment.
Content quality also matters. Genuine expertise, clear limitations, accurate sourcing and useful explanations can help people decide. Avoid mass publishing of near-identical pages for every prompt or location merely to claim AI visibility. No markup, word count or content format guarantees inclusion in Google's AI experiences. Commission articles to answer real questions and maintain them as business details and technology change.
Record a separate status for saved database changes and verified public results. A successful CMS update does not establish that an image loads, a table remains readable on mobile or a form works. A renderer that displays internal editorial notes requires its own code-level fix. Budget for these engineering needs rather than allowing them to disappear beneath a high content-production count.
Review provider reports in terms of completed value
Ask agencies and internal teams to distinguish research, recommendations, deployed work and independently tested results. A report may legitimately contain work in progress, but it should not describe a draft technical ticket as a completed production repair. Require appropriate evidence and transparent unresolved dependencies. For content, inspect source accuracy, topical focus and whether the article genuinely helps its intended audience.
Be wary of dashboards that show only a proprietary authority score or an estimated AI share of voice without definitions. Such metrics can inform hypotheses, but they are not direct revenue or customer counts. The business owner should understand the measured population, date, tool and limitations. Reports should explain what decision follows from the observation rather than merely offering more charts.
Compare costs against outcomes the team can influence, such as repairing a form, clarifying a service page or producing a verified technical guide. Monitor search exposure as an external observation, not an agency-controlled deliverable. A supplier that refuses to distinguish those categories makes it harder to manage financial risk.
Use a staged investment decision
Begin with a short diagnostic phase focused on high-value pages, existing reporting and real conversion barriers. Then approve a small set of improvements with clear owners and acceptance tests. Review whether those actions were implemented correctly and whether the business has enough observations to evaluate customer outcomes. Expand recurring investment only when the next tasks are justified by remaining opportunities and resources.
A staged programme can include an explicit stop or redirect condition. If a proposed content series repeatedly attracts the wrong audience, reconsider its purpose. If the business has insufficient development capacity to implement technical recommendations, address that constraint before buying another audit. If Search clicks fluctuate but qualified leads remain healthy, avoid reacting to the headline figure alone. These are practical budget controls, not admissions of failure.
Keep the unit-economics assumptions under review as margins, capacity and services change. SEO is not separate from the business's profitability and operational ability to serve customers. The best spending level is one the organisation can justify with a coherent opportunity and maintain without neglecting essential work.
Make the final review defensible
A useful SEO budget review states the business objective, qualified-customer definition, real costs, existing performance evidence, technical barriers and uncertainties. It also records the source of any study about AI Search rather than treating its headline statistic as a universal rule. A table of scenarios can be useful when denominators and assumptions are clear, but fictional results must remain explicitly hypothetical.
The next recommendation should be actionable: fix a verified defect, improve one important page, measure qualified leads more reliably or monitor a credible trend until sufficient data exists. Do not promise a fixed Google ranking or AI citation date. Read Google's guidance on hiring an SEO, AI features documentation and Search Console Performance reporting as complementary sources. They explain SEO practice and measurement boundaries, not a mandated revenue percentage or a guaranteed return.
How should an owner choose between fixing the site and buying more SEO content?
Compare opportunity and evidence, not arbitrary percentages of revenue. Both activities can be useful, but the first investment should match the actual business constraint.
| Budget question | Repair verified customer barrier | Expand researched content programme |
|---|---|---|
| Problem addressed | Existing important visitor task fails or misleads | Genuine customer question lacks a useful answer |
| Cost components | Diagnosis, development, release and regression test | Research, expert review, writing and maintenance |
| Direct acceptance test | Public customer action now works as intended | Published answer is accurate, complete and discoverable |
| Main uncertainty | Commercial recovery may depend on further factors | Search demand and traffic are not guaranteed |
| Reason to prioritise | Confirmed loss of functionality on valuable route | Verified information gap for eligible customers |
Source A: Google guidance on hiring an SEO addresses realistic provider expectations and site improvements. Source B: Search Console Performance report explains the scope of search clicks and impressions, not profit or customer acquisition cost. Both are general sources reviewed 30 October 2026, not evidence for any universal spending percentage. Decision: use real unit economics and verified customer barriers to prioritise investment, then measure appropriate outcomes over a sensible period.
Five questions about SEO spending and AI Search
Should every business spend the same percentage of revenue on SEO?
No. Margins, growth goals, customer value, capacity, existing technical problems and cash flow differ. Evaluate the opportunity and costs of actual proposed work.
Does an AI Search announcement prove our clicks will fall?
No. Investigate the site's own relevant Search observations and customer outcomes; broad product news does not establish a universal effect.
What is the difference between a click and an acquired customer?
A click records a search interaction; acquiring a customer requires an actual commercial outcome. Qualified leads and sales need separate trustworthy evidence.
Should a business fix a broken enquiry form before commissioning new articles?
Usually prioritise a confirmed high-value customer barrier. Assess the cost and impact, then fund additional content where a genuine information gap remains.
How can we judge whether an SEO agency's work was completed?
Distinguish recommendations, deployed changes and public acceptance tests. Monitor rankings and revenue separately without treating either as a guaranteed agency-controlled result.