The questions a buyer most needs answered are precisely the ones a provider has the least incentive to answer honestly. What does this cost. How long before anything moves. What happens if it does not work. When should you hire nobody at all. Each of those has a real answer, and each answer is either commercially inconvenient, uncomfortably imprecise, or both. So the market has settled on substitutes — it depends, every business is different, results vary, book a free consultation — which are not answers but devices for getting you onto a call before you can evaluate the call. These fourteen pages answer the questions anyway.
The four questions that go unanswered, and why
Cost. Publishing a number is a competitive disadvantage and a half-truth at once, since the same engagement genuinely does cost different amounts depending on site size, whether implementation is included, whether content is included, how many markets are involved, and whether anyone on your side can build what gets specified. But the drivers are knowable, and a buyer who understands them can place themselves before speaking to anyone. Declining to publish a figure is defensible. Declining to explain what moves it is not, and that is what makes most cost pages useless.
Time. The familiar three-to-six-month answer is not a measurement. It traces to an opinion poll of practitioners — a survey of what people in the industry believe, not an observation of what happens. Google declines to give a timeframe at all, and explicitly declines to promise that changes will produce a noticeable effect. Different work also runs on different clocks: a technical fix can take effect within days of a recrawl, a migration takes weeks for most pages on a small or medium site by Google's own estimate, and authority work is measured in months.
What a guarantee really means. Nobody outside the company operating the ranking system controls its output, so a promised position is either false, quietly written against phrases nobody searches for, refundable only as credit against more of the same service, or backed by methods that put the domain at risk. The offer is diagnostic: it tells you about the method before you have seen the method.
When to hire nobody. There is no commercial incentive whatsoever to write that one down, which is why it is the most useful page here.
Written to be useful to someone who never makes contact
These pages are written so that reading them can be the end of the transaction. Someone who reads what an audit covers and concludes their own developers can handle it has been well served. So has someone who takes three better questions back to a provider they were already talking to, even though the value lands somewhere else entirely.
That is a commercial position as much as an ethical one. A buyer who understands the work is a materially better client: scoping takes a fraction of the time, expectations start out calibrated, and nobody spends month three discovering what should have been settled in week one.
Most of these answers will be read somewhere other than here
Of the keywords researched for this site, 69% already return an AI Overview — the generated answer block above the results. That is not a forecast about where search is heading. It is the current state of the queries this material is written for. For most of these questions, a machine-written summary is the first thing a person reads, and the page it was assembled from sits one click further away than it used to.
Alongside that, SparkToro's analysis of Similarweb clickstream data found roughly 68% of US Google searches ended without any click in January to April 2026. Two caveats have to travel with that figure every time, and any page quoting it without them is misrepresenting it:
- The panel excludes the Google mobile app. It covers desktop and mobile searches made in a browser. SparkToro notes zero-click behavior is likely more pronounced inside the app, so the figure is not a ceiling — but it is also not a measurement of all searching.
- SparkToro states long-term comparisons are not like-for-like. The multi-year trend stitches together data from different panel providers in different years, and the source says plainly these are not the same users or devices. The number describes a period. It must never be published as a measured change over time, and you will see it used that way constantly.
One more thing belongs next to those, because the loudest version of this story is not the best-supported one. The largest published study of AI Overview prevalence found the trigger rate rising through mid-2025 and then falling through the back half of the year. Prevalence is volatile rather than a line heading straight up, and any figure you are quoted needs a date attached.
What that changes about how these pages are written
If a large share of readers meet the answer without ever meeting the page, the answer has to be correct when it is lifted out and quoted with no context around it. That is the whole job here.
Answer first, then the reasoning
Each page settles the question in its opening rather than warming up to it — no throat-clearing about how the field has evolved, and no definition of the title term on a page already titled that.
Every paragraph has to survive extraction
No dangling antecedents, no as mentioned above, no claim whose caveat lives three paragraphs away. If a number is only true with a qualification attached, the qualification sits in the same sentence. The zero-click figure above is written that way on purpose.
Figures carry their source and their limits
Numbers here trace to something published, and the ones that could not be traced to a primary source with a stated method were left out, even where they would have made a more dramatic point. A widely circulated figure for how far AI Overviews cut click-through appears in a great many articles and traces to no study. It is not here.
How the fourteen are grouped
What this work is defines the terms in a buyer's language rather than a practitioner's: what a consultant does day to day, what search engine marketing covers and why two people using the phrase often mean different things, what a technical audit examines, and what the word enterprise is doing when a vendor uses it. Choosing someone is the evaluation material — how to run a hire, what to ask before signing, how consultant, agency and in-house genuinely differ, what a proposal reveals, and why a promised ranking is a warning rather than a benefit. Cost and commitment covers what drives the number, how retainer, project and hourly arrangements each fail, and how long the work takes. Measurement and 2026 covers what each reporting instrument can and cannot see.
A fair amount of this will contradict something you have been told by someone with a quota. Where it does, the reasoning is on the page rather than in a footnote, so you can weigh it yourself instead of picking whichever source sounded more confident. The questions are answerable, and you should not have to buy anything to get the answers.
Published