Account structure: the decision you can't undo cheaply
The structural decisions you make when setting up a Google Ads account are the hardest to change later because changing them mid-campaign means losing historical performance data that Google's algorithm uses to optimise bidding. Get it right at the start.
The principle is simple: separate things that need separate budgets, separate bidding, or separate reporting. Brand campaigns should always be separate from non-brand, because ROAS from branded search is always higher, and mixing it in inflates your reported performance while hiding how your non-brand campaigns are actually doing. Shopping should be separate from search. Performance Max, if you use it, should be separate from everything else so you can see its contribution clearly.
For most small brands starting out, the simplest viable structure is: one brand campaign (exact match on your brand terms), one non-brand search campaign segmented by product or service category, and if you're in e-commerce, one Shopping campaign. Start there. Add complexity only when the data tells you to.
Keyword strategy: intent tiers and negative lists
Keywords are not all created equal. A searcher typing "buy blue running shoes size 10" is significantly closer to purchasing than someone typing "are running shoes good for flat feet". Both might be relevant to your business. Only one is worth bidding on at the same price.
Build your keyword list in intent tiers. High-intent commercial terms (searches that directly express purchase intent) get your highest bids and your best ad copy. Mid-intent informational terms that are closely adjacent to purchasing decisions go in a separate ad group with more conservative bids. Informational terms with no purchase signal stay out of the account until you have budget to spare.
Negative keyword lists are as important as the keyword list itself. Every week, pull the search terms report and add irrelevant queries to your negatives. An account with a tightly maintained negative keyword list wastes significantly less budget on irrelevant clicks, which means every pound of budget works harder.
Match types: the argument for starting narrow
Broad match has become Google's default recommendation, and Google is not neutral on this question. Broad match makes Google more money by triggering ads on more searches. Whether those searches convert for your business is your problem, not theirs. Broad match without aggressive negative management is a budget burn, especially for smaller accounts.
Start with exact and phrase match. Add broad match only once you have enough conversion data that Google's algorithm has something meaningful to optimise towards, and only with a well-maintained negative list. The accounts that get into trouble with broad match are the ones that turned it on before the algorithm had any real signal about what a good conversion looks like for that business.
"The accounts that scale profitably are the ones that optimised ruthlessly before they scaled, not the ones that threw budget at the wall to find what worked."
Creative: what actually drives clicks and conversions
For search ads, the copy decisions that move the needle are: headline one (the most-read element), the description (where you address the next objection after the headline catches attention), and the CTA (clear, specific, action-oriented). Run three variants per ad group: one benefit-led, one problem-led, and one feature-led. Let them run for three to four weeks and let performance data, not preference, determine the winner.
Responsive search ads give Google flexibility to combine headlines and descriptions. Use them, but be careful: the flexibility means headlines and descriptions are shown in combinations you haven't explicitly approved. Write every headline and every description to stand alone, not to be read in sequence. If your headline assumes the visitor has already read description one, half your ad combinations will be incoherent.
Bidding strategy: match the strategy to your data
Google's Smart Bidding strategies (Target CPA, Target ROAS, and Maximise Conversions) work better the more conversion data they have. "Better" means more accurate, with less wasted spend. An account with 50 conversions per month sees meaningful Smart Bidding performance. An account with 10 conversions per month might be better served by manual CPC while it builds the data that Smart Bidding needs to function.
A common mistake: switching to Target ROAS when the account doesn't yet have enough conversion volume to support it. The algorithm makes poor decisions with thin data. If you set a Target ROAS and the account can't hit it with its current conversion data, Google either restricts spend dramatically (because it can't find enough auction opportunities it believes will convert at that ROAS) or bids aggressively and burns budget missing the target. Neither is useful. Build the data first.
Optimisation cadence: what to review and when
Weekly: search terms report (add negatives), CTR by ad (kill underperformers), budget pacing (is spend on track?). Monthly: landing page performance (are the pages converting?), quality scores (are they trending up or down?), auction insights (how are you performing against competitors?). Quarterly: account structure review, keyword expansion, budget reallocation based on ROAS by campaign.
The cadence matters as much as the actions. An account reviewed monthly will outperform an account reviewed quarterly. An account reviewed weekly will outperform an account reviewed monthly. Time spent in the account, on the right things, is what separates profitable advertisers from ones burning budget.
When to scale: the signal you're looking for
Scale when your cost per conversion is stable at a level that's profitable, and when increasing budget doesn't immediately degrade that metric. If you add 30% more budget and your CPA increases by 60%, you haven't found the ceiling of your market. You've found the ceiling of your current account efficiency. Fix the efficiency problem before adding more budget to it.