You paid for the click.
Then sent it to a page with nine exits.

You can, and it is the most common way paid budget is wasted. A homepage is built to let anyone go anywhere, which is correct for organic traffic and wrong for a click you bought against one specific query. The visitor arrived asking one question and lands somewhere answering forty. Our observation, labeled as observation since no authority publishes attention data, is that the first thing checked is whether the promise from the ad appears on the page — not whether the company looks credible. Measurement of the difference belongs to conversion rate optimization. Google's own framing of content built for people rather than for a mechanism is the closest published standard.

The fewest that let you respond usefully — and understand that each additional field is a deliberate trade of volume for qualification, not a free addition. The faults we find most are fields collected because a CRM has a column and required fields asking for something nobody has to hand. Whether a specific field is worth its cost is a testable question, and Google's conversion documentation covers how the outcome should be counted. Test design sits under CRO rather than here.

It depends on the page's job, and it should be decided deliberately rather than by default. A page built only to receive paid traffic and duplicating an existing service page is usually better noindexed, and thank-you states always should be. A page carrying substantial unique content may deserve to be indexed and earn organic traffic as well. What matters is that the directive is chosen, not inherited — Google's technical requirements set the baseline. Related: technical SEO.

Producing hundreds of near-identical pages, one per keyword, can. Google's spam policies define scaled content abuse as generating many pages primarily to manipulate rankings rather than to help people, and its guidance separately notes that a high quantity of pages does not make a site higher quality. A handful of pages for genuinely distinct offers is not that. The test is whether the pages differ in substance or only in a phrase. Consolidating an existing set is usually the first fix — see content strategy.

Faster than a site page, because you are paying for every visitor who waits. Core Web Vitals are reported from real-world field data rather than a lab score, Largest Contentful Paint defines what is being timed, and Cumulative Layout Shift covers the page moving under a thumb mid-tap. Google's page experience guidance frames these as part of a good experience rather than one ranking dial — which on a paid page is the point. Related: website design.

Yes, within published conditions. Google's website testing guidance states that running A/B and multivariate tests is not itself a problem, and sets out the requirements: do not cloak, use temporary rather than permanent redirects for split tests, use canonical tags correctly on variants, and run tests only as long as necessary. The failure is showing crawlers one version and users another, which the spam policies name as cloaking. Whether you have the volume to resolve a test is a separate question, covered under CRO.

Almost always because the phone number is not attributed. On service offers a large share of conversions are calls, and a call from an untracked number is a conversion that happened and was never counted — so a working campaign looks dead and gets paused. Google's conversion documentation covers the counting mechanics; the definition of what counts is yours to set. Tracking numbers also need implementing without fragmenting your canonical number across listings, which is why it coordinates with directory management.

When contact will happen, by which channel, and what to do if that is too slow. "We'll be in touch" answers none of those. Technically it matters more than it looks: it is usually where conversion tracking fires, so a broken thank-you state makes a campaign appear to produce nothing, and it should sit on a state that cannot be reached without converting. It should also carry noindex, since it exists only for people who already converted. Related: website maintenance.

Nobody can tell you honestly. No issuing authority publishes conversion benchmarks by industry, so every figure in circulation traces back to a vendor survey of that vendor's own customers, usually undated and without methodology. What is real is your own baseline, measured on your own traffic, and whether a specific change moved it. Google's guidance on content made for people rather than for a mechanism is the closest thing to a published standard here, and it is about usefulness rather than a number. Baseline setting runs through the audit.

One per genuinely distinct offer or audience, not one per keyword. If two pages would say the same thing with a phrase swapped, they should be one page — and building them separately edges toward what Google's spam policies describe. If two offers have different buyers, different objections and different proof, they need different pages, because a page serving both serves neither. In practice most accounts need fewer pages built better rather than more. Start with an A.R.C. Report or a paid search review.