The exhaustive Google Ads FAQ.

Enough for your target CPA to gather data — a rough guide is at least 15–30 conversions/month for automated bidding to learn. Start where you can sustain a few weeks of learning, not a single test day. It's better to run one focused campaign properly than three underfunded ones.

For most businesses, Search first. It's higher-intent, more transparent, and teaches you what actually converts. Layer Performance Max on later, once you have conversion tracking and data — not as your untested first campaign.

Expect a 1–2 week learning period where performance is noisy and you shouldn't judge it. Meaningful signal usually takes 2–4 weeks. Changing settings constantly resets learning and is the most common self-inflicted wound.

A dedicated landing page matched to the ad almost always converts better than a generic homepage. A healthy click-through rate with a poor conversion rate usually means the page, not the ad, is the problem.

Broad match shows your ad for loosely related searches (widest reach, most waste); phrase match requires your term's meaning to be included; exact match is tightest, only close variants of your exact term. More reach means more irrelevant clicks to filter out.

A common approach: start tighter (phrase/exact) when budget is limited and you can't afford waste, and use broad match only once you have Smart Bidding and solid conversion data guiding it. Broad match without good conversion signals is how budgets evaporate.

Negative keywords stop your ads showing for searches you don't want — 'free', 'jobs', 'cheap', competitor DIY terms. They're the single biggest lever for cutting wasted spend. The risk is cutting real customers by mistake, so they should be reviewed, not blindly automated.

Keywords are what you target; search terms are what people actually typed to trigger your ad. Broad match means those can differ wildly. Reviewing the search-terms report is how you find both new negative keywords and new keywords worth adding.

Fewer, tightly-themed keywords per ad group beat cramming in dozens — it keeps your ads relevant to each keyword, which lifts Quality Score. If keywords need very different ad copy, they belong in separate ad groups.

With little conversion data, Maximize Clicks or Manual CPC gets you traffic to learn from. Once you have ~15–30 conversions/month, move to a conversion-based strategy (Maximize Conversions, then Target CPA or Target ROAS). Moving too early gives the algorithm too little to work with.

Target CPA tells Google the cost-per-acquisition you want and it bids toward it; Target ROAS does the same for return on ad spend. Both are only as reliable as your conversion tracking — set a target on bad data and Google confidently optimises toward the wrong thing.

Smart Bidding needs a learning period and enough conversion data; early on it can overspend while it calibrates. It also silently ignores manual bid adjustments for device, location and schedule — a common surprise. Give it 1–2 weeks before judging.

Google can spend up to 2x your daily budget on high-opportunity days, balancing out over the month so you don't exceed your monthly average. So a single day over budget isn't a bug. What matters is the monthly total and the cost per result.

It's the percentage of available impressions you missed — split into 'lost to budget' (you ran out of money) and 'lost to rank' (your Ad Rank was too low). It's an abstract percentage until you translate it into the actual budget or quality improvement needed to recover it.

Google's 1–10 rating of a keyword, built from three parts: expected click-through rate, ad relevance, and landing page experience. A higher score means cheaper clicks and better positions for the same bid.

Fix the specific weak component, not the blended number. If landing page experience is 'below average', improve page speed and relevance; if ad relevance is low, tighten your ad copy to the keyword; if expected CTR is low, write more compelling ads. Guessing wastes effort on the wrong lever.

Ad Rank is roughly your bid multiplied by your Quality Score, plus context and expected impact of assets. That's why a competitor with better relevance and a smaller budget can appear above you — you can't always outbid a quality gap.

Ad Strength (Poor to Excellent) rates how many varied, relevant headlines and descriptions you've given a responsive search ad. A 'Poor' rating quietly limits how often and where your ad shows — it's worth fixing even though it's not the same as Quality Score.

It can be very effective with good tracking, strong assets and enough conversion volume — and wasteful without them. The real problem is visibility: it won't tell you where spend went or why. It rewards active oversight, which its native reporting doesn't provide.

It can. Left unchecked, PMax often claims credit for brand and existing Search demand you'd have won cheaply anyway, inflating its apparent performance. Watching for this cannibalisation is essential and easy to miss manually.

Look past the headline ROAS at whether it's finding new customers or just harvesting existing demand, whether assets are rated well, and whether any asset groups have meaningful spend with zero conversions. These signals are buried and rarely surfaced by default.

Common causes: conversion tracking not firing correctly, the wrong attribution window, iOS privacy and ad-blockers hiding conversions, or double-counting. An account optimising on inaccurate conversions will confidently make the wrong calls — measurement is the foundation everything else sits on.

It reports conversions to Google from your server rather than only the visitor's browser, catching the ones ad-blockers and iOS privacy settings hide — often 20–40% more. If Smart Bidding is optimising on an undercount, fixing this is one of the highest-impact things you can do.

A conversion is any tracked action; a good lead is one that becomes a customer. Optimising toward raw conversion volume can flood you with cheap, low-quality leads. Feeding quality signals back — which leads actually closed — points bidding at the right outcome.

Google defaults to data-driven attribution where available, which spreads credit across the path rather than only the last click. The key is consistency — comparing two campaigns or platforms on different models or windows is comparing apples to oranges.

Running broad match without negative keywords or good conversion data — it spends fast on searches that never convert. Close behind: judging performance during the learning period, and changing settings so often that learning never completes.

Usually one of: the landing page doesn't match the ad's promise, the traffic is low-intent (wrong keywords or match type), tracking isn't firing so conversions aren't recorded, or the offer itself isn't compelling. Check tracking first — you may be converting and not seeing it.

Not during the learning period, and not on too little data — pausing prematurely throws away the learning you paid for. Give campaigns a fair window (usually 2–4 weeks) and a meaningful number of clicks before judging, unless spend is clearly running away with zero results.

Often yes — it's cheap, protects against competitors bidding on your brand, and controls the message. But if no competitors are bidding and you rank #1 organically, it can be spend you'd have won for free. It's account-specific; test it.

Less than you'd think. Frequent changes reset the learning phase and make it impossible to tell what worked. A steady cadence — review regularly, change deliberately, then let it run — beats daily tinkering. This is exactly the discipline automation should enforce, not undermine.