PPC Optimization
A complete, practical guide to optimizing pay-per-click campaigns, plus a free 42-point PPC audit that scores your account live as you answer it. This covers Google Ads and Microsoft Ads search campaigns. It is not about Amazon PPC, and not about Practitioners Publishing Company audit guides.
What PPC optimization actually means
PPC optimization is the ongoing work of making a fixed advertising budget produce more conversions. That's it. Every genuine optimization does one of three things: it stops you paying for clicks that were never going to convert, it wins more of the clicks that do, or it converts a higher share of the clicks you already paid for.
What makes it hard isn't complexity. It's that the three levers sit in different places and most people only ever pull the third one. They rewrite ad copy, tweak a headline, maybe redesign a landing page, and wonder why cost per conversion barely moves. Meanwhile a third of the budget is going to search queries nobody ever looked at, and the conversion action that Smart Bidding is optimizing toward is a newsletter signup.
There's also a structural reason accounts decay if left alone. Match types drift as Google loosens what counts as a "close variant". Competitors enter and leave the auction, moving your CPCs. Search behaviour shifts with seasons and news. Smart Bidding retrains on whatever signal you gave it, including a bad one. An account that was well optimized in January is not well optimized in July unless somebody looked at it.
Why every optimization starts with an audit
Optimization without an audit is guessing which of the three levers to pull. An audit is just the cheap step that tells you where the money is actually leaking, so you spend your limited attention on the leak rather than the thing that happened to be on your screen.
The order matters more than the checklist. Audit in dependency order, not screen order: measurement first, because every judgement downstream depends on it; then what you're bidding on; then structure; then creative; then post-click; then bidding strategy. Fixing ad copy while your conversion tracking double-counts is work you'll have to redo, because you can't tell whether the new copy helped.
Most published PPC audit checklists, and there are dozens of them, are flat lists of 60 or 100 items with no weighting, which quietly implies a missing callout extension matters as much as absent conversion tracking. It doesn't. The scorecard below weights every check, so a low score means something real.
Free PPC audit scorecard (42 points)
Answer as many as you can. It scores live, nothing is sent anywhere, and unanswered items are not counted against you. Each section maps to a section of this guide, so a low score in one area tells you exactly which part to read.
A note on what this is and isn't: this is a self-reported audit, not an automated one. Tools that connect to your Google Ads account can read your settings but can't tell you whether your conversion value reflects real margin, or whether your landing page keeps the ad's promise. This asks you the things only you know, which happens to cover the half of a PPC audit that automated graders skip.
1. Conversion tracking & measurement
Nothing else on this page can be judged if this section is wrong, which is why it's first and why its checks carry the heaviest weight. A campaign optimizing toward a broken signal doesn't fail loudly. It succeeds at the wrong thing, confidently, for months.
Count one primary conversion per goal
This is the most common expensive mistake. If newsletter signups and purchases are both marked Primary, Smart Bidding sees them as equally valuable and rationally buys the cheaper one. You get a falling cost per conversion and falling revenue at the same time, which reads as success on the dashboard.
Mark only the revenue-bearing action as Primary. Demote soft actions to Secondary. They stay visible in reporting for context but never influence bidding.
Get the counting method right
"Every" conversion on a lead form counts a double-submit twice. "One" on an ecommerce purchase undercounts a repeat customer. Use One for leads, Every for sales. It's a two-click setting that silently distorts every metric above it.
Pass values, and enable enhanced conversions
Value-based bidding cannot function when every conversion is worth zero. If you don't have per-transaction values, assign a modelled value per lead type. Even a rough static figure beats nothing, because it lets the algorithm distinguish a demo request from a brochure download. Enhanced conversions, meanwhile, recovers conversions that cookie loss and browser restrictions would otherwise drop entirely.
2. Keyword research & match types
What you bid on decides who arrives. Most wasted spend is created here and merely discovered later in the search terms report.
Separate intent, not just topics
"What is project management software" and "project management software pricing" are the same topic and completely different buyers. Grouping them means one ad and one bid for both, which underserves the buyer and overpays for the researcher. Split commercial modifiers (buy, pricing, cost, vs, alternative, near me) into their own campaigns so you can fund them differently.
Stop running everything on broad match
Broad match is not inherently bad, but it needs conversion data to steer it. On a new campaign with no conversion history, broad match plus Smart Bidding is close to instructing Google to spend your budget experimentally. Run exact and phrase on your proven converters, and let broad explore only where there's enough signal to correct it.
Know the price before you bid
A keyword with a $48 cost per click needs a very different conversion rate to work than one at $2, and plenty of accounts discover this after the fact. Price your list first; you can check real cost per click, monthly volume and organic difficulty for up to 40 keywords at once with the free CPC Checker, or hand a whole site or business idea to the Keyword Opportunity Finder and get a ranked shortlist with the economics attached.
3. Negative keywords & search terms
If you do one thing on this page, do this one. The search terms report is where the account shows you what people actually typed (not all of it, since Google withholds terms it considers too rare), but enough, and in most unaudited accounts a meaningful share of spend is sitting there in plain sight, attached to queries that were never going to buy.
Read it on a schedule, not when something feels wrong
Sort by cost, filter to zero conversions, and work down the list. The judgement is simple: could this person plausibly have bought from me? If not, negate it. Thirty minutes a month is usually enough once the obvious waste is gone.
Use one shared list at account level
Per-campaign negatives are forgotten the moment somebody launches a new campaign. Build one shared list of universal waste: jobs, career, salary, free, tutorial, template, reddit, any competitor you don't want to pay for. Apply it everywhere. Apply that list to new campaigns as you create them. Shared lists are not inherited automatically, which is exactly why the one-list habit beats scattering negatives per campaign.
Three lookalikes that drain budget quietly
These three are the ones that cost the most in aggregate, because each one looks like a real buyer at a glance. They aren't. All three convert far below your account average, and most of them never convert at all. Recognising them in the search terms report is the difference between reading the report and actually using it.
- Job seekers. "[your category] jobs" and "[your category] salary" are topically perfect and convert at zero.
- DIY researchers. "template", "checklist", "how to do X yourself": these people are explicitly not buying a service.
- Your own other campaigns. Two campaigns bidding on the same query means you're bidding against yourself. Cross-campaign negatives give each query one home.
4. Campaign & ad group structure
Structure isn't an aesthetic preference; it's what makes an account steerable. Budget, bids, ad copy and reporting are all applied at campaign or ad group level, so a structure that lumps unrelated things together removes your ability to treat them differently.
Tight ad groups exist so ad copy can be specific
The test is simple: could every keyword in this ad group share the same headline without the headline becoming vague? If not, split it. A sixty-keyword ad group forces generic copy by construction, and generic copy is what specific copy beats.
Allocate budget by performance, not evenly
Even splits feel fair and are quietly expensive: they subsidise your worst campaign with your best one's money. Rank by cost per conversion, move budget up the list, cap the tail.
Never leave a profitable campaign budget-limited
"Limited by budget" on a campaign converting below target cost per acquisition is money you've decided not to make. Either raise the budget or move it from something weaker, but don't leave it sitting there, which is the default outcome because nothing in the interface nags you about it.
Check the two settings nobody revisits
Location targeting defaults to "Presence or interest", which also reaches people merely interested in your area. Worth knowing before you change it: this is Google's documented default and its recommended option, and Google reports better conversion volume for it. For a local service business that can only serve people physically nearby, "Presence" is still usually the better trade, but but it is a trade of reach for control, not a settings error. And ad scheduling defaults to always-on, including hours when nobody can answer the phone.
5. Quality Score & Ad Rank
Quality Score isn't a vanity metric, but it isn't a lever either. It's a 1-10 diagnostic of the same relevance signals the auction uses, not a discount applied to your bid. Ad Rank is roughly your bid combined with expected click-through rate, ad relevance and landing page experience, meaning a more relevant advertiser can outrank a higher bidder and pay less per click for the privilege.
The practical read: Quality Score is a diagnostic, not a target. Its three components tell you where to look.
- Expected CTR below average: your ad isn't compelling against what else is on the page. This is a copy and offer problem.
- Ad relevance below average: the query and the ad text don't align. This is usually an ad group that's too broad, not bad writing.
- Landing page experience below average: the page doesn't deliver what the ad promised, or it's slow. This is the section-8 problem below.
Chasing the number itself leads people to pause keywords with low scores, which just hides the diagnostic. Fix the component instead.
6. Ad copy & RSA assets
This is where the auction becomes a human decision. Everything upstream got you onto the page; the copy decides whether the click happens and, increasingly, whether it's the right person clicking.
Feed responsive search ads properly
A responsive search ad with three headlines has almost nothing to combine, and Ad Strength reflects that. Write toward the full 15 headlines and 4 descriptions, but make them genuinely different angles, not synonym swaps. Google can only test variety you actually supplied. You can lay out headlines, descriptions and sitelinks and see the result rendered as a real Google ad, with live character limits, in the free Google Ads Preview Tool.
Put the search term in a headline
Query-to-headline match drives both click-through rate and ad relevance. It doesn't need to be clever; it needs to be the words the person typed, so the ad visibly answers their search.
Say something a competitor couldn't paste into their own ad
"Quality service, competitive prices, trusted experts" is what everyone else in the auction also says, which makes it invisible. Replace adjectives with facts: a number, a timeframe, a guarantee, a named integration, a specific inclusion. If your competitor could put your headline in their ad without lying, it isn't a differentiator.
Copy is also the one part of this audit where the diagnosis genuinely benefits from an outside read. The free Google Ads audit on this site runs your ad through six lenses (behavioral psychology, platform and media buying, copywriting, offer, creative craft and conversion funnel) and ranks every finding by impact against effort.
7. Ad extensions
Extensions cost nothing to add, though clicks on them are billed like any other click, and most accounts leave half of them empty. They don't just add information; they make the ad physically larger, which pushes competitors down the page.
- Sitelinks are the biggest single available lift. Google recommends at least 6 with descriptions for high-volume campaigns and reports up to 3.5% more conversions at a similar cost per conversion for advertisers who get to 6. You need a minimum of 2 before any can show. Point them at genuinely different pages rather than anchors on one page.
- Callouts carry the trust signals that don't fit in a headline: free shipping, 24/7 support, no contract, established date.
- Structured snippets list your categories or services in a format Google understands as a set.
- Image assets make search ads visibly bigger on mobile, and a logo asset can be supplied alongside them. Google reports a 6% average lift in click-through rate when image assets show with a Search ad. Note the eligibility bar: the account needs roughly 60 days of history, recent Search spend and a clean policy record.
- Call and location assets can convert high-intent mobile traffic without a landing page being involved at all.
The one caveat: don't add an extension type just to fill the slot. A sitelink pointing somewhere useless is worse than no sitelink, because it spends a click you paid for.
8. Landing pages & post-click
By the time this page loads, you've already paid. Everything that happens next is either recovering that cost or wasting it, and this is the part of PPC that most advertisers treat as somebody else's job.
Message match is the whole game
The most expensive bounce in paid search is the one where the visitor can't tell within two seconds that they're in the right place. Put the ad's promise in the page's main heading, near-verbatim. If the ad said "Free 30-day trial, no card", the page should not lead with "Enterprise-grade workflow automation".
Don't send paid traffic to your homepage
A homepage makes the visitor go and find the thing your ad already promised them. Every ad group should land on the page matching its intent, even if that means building a few pages.
Speed is a conversion setting
Mobile bounce climbs sharply past a few seconds and you're billed for the click regardless. Test the actual landing URL (with its tracking parameters) rather than the homepage, and fix largest contentful paint before anything cosmetic.
Because the ad and the page have to work as one thing, the audit on this site reads them together: give it your landing page URL alongside the ad and it will flag promise mismatches between the two as findings in their own right.
9. Bidding & budget
Which bid strategy you choose matters far less than whether it has the data it needs to work. Most bidding problems are really data problems wearing a strategy costume.
Match the strategy to your conversion volume
Be clear that this one is a judgement call, not a rule: Google publishes no minimum and says you can start Target CPA with no conversion history. In practice a campaign getting two conversions a month has very little for the model to learn from, so below roughly 15–30 conversions a month many practitioners get better results from maximize clicks or manual CPC with tight keyword control. Treat that band as caution rather than a threshold, and if you disagree, the disagreement is with us, not with Google.
Set targets from margin, not ambition
An unreachable target CPA doesn't make Google try harder. It can throttle delivery until the campaign is barely serving at all. Set the target near what you're currently achieving, then tighten in steps of 10–15%.
Stop churning
Every significant change to a bid strategy or target starts a new learning period, during which performance is unstable and shouldn't be judged. Google frames the length in conversion cycles rather than calendar days, so it varies with your volume. There's no official "wait two weeks" rule, despite how often it's repeated. The practical point stands regardless: changing targets weekly keeps the algorithm in learning far more of the time than it needs to be.
Check the networks spending your search budget elsewhere
Search Partners is on by default for new Search campaigns; Display expansion is a separate toggle whose default has moved around between interface versions, so check yours rather than assuming. Segment your reporting by network. If partner or display traffic converts materially worse, switch it off. You didn't choose it, and it's competing with the search traffic you did choose.
10. Audiences & targeting
Audience settings are the ones nobody revisits after launch, and one of them can silently shrink your reach to almost nothing.
The critical distinction is Observation versus Targeting. Observation lets you read how an audience performs while still showing to everyone. Targeting restricts delivery to only that audience. Setting an audience to Targeting by accident is a common cause of "my campaign suddenly stopped getting impressions".
Beyond that: attach remarketing and customer-match lists, because past visitors are the cheapest converting traffic you have access to, though that is worth confirming in your own numbers rather than assuming. And exclude existing customers from acquisition campaigns if your product is a one-off purchase. Paying to re-acquire someone who already bought is pure waste.
Know who you're bidding against
Most advertisers write ad copy without ever reading their competitors'. You can pull the advertisers showing on your keywords at the time of the scan, how often each appears by position, and what their ads actually say, with the Google Ads Benchmark, which is the difference between positioning your offer against a real auction and guessing at it.
11. Automation, Smart Bidding & Performance Max
Automation is not a shortcut past the rest of this page; it's an amplifier. Smart Bidding applied to clean data, tight keyword sets and accurate conversion values is genuinely better than manual management. Applied to a broken conversion signal, it finds the cheapest route to the wrong outcome faster than any human could.
Which is the whole argument for auditing before automating: every prerequisite for automation working well is in sections 1 through 4.
The Performance Max caveat worth knowing
Performance Max will happily take credit for branded searches you would have won for free, which makes its reported performance look excellent while adding little incremental revenue. Add brand exclusions, or run a dedicated brand search campaign alongside it and compare. Judge PMax on incrementality, not on its own dashboard.
Keep a human in the reporting loop
Automated rules and scripts are useful for enforcement (pausing runaway spend, flagging disapprovals, catching a broken landing page. They're poor at judgement. Automate the alarms, not the decisions.
12. What to check weekly, monthly, quarterly
Optimization fails more often from lack of cadence than lack of knowledge. A workable rhythm:
Weekly (about 30 minutes)
- Search terms report: negate this week's waste.
- Spend pacing against budget, and anything newly "Limited by budget".
- Disapprovals, broken landing pages, and any campaign that stopped serving.
Monthly (about two hours)
- Cost per conversion by campaign; reallocate budget up the ranking.
- Ad performance: retire the losing variant, write a new challenger.
- Quality Score components on your top spend keywords.
- Extension coverage on any new campaign.
Quarterly (half a day)
- The full 42-point audit above, scored, compared to last quarter.
- Conversion tracking re-verification. Tags break silently after site changes.
- Bid strategy fitness as conversion volume changes.
- Competitor ad copy: what's new in the auction.
The quarterly audit is the one people skip, and it's the one that catches the expensive, slow-moving problems: the tracking that broke in a redesign, the campaign that's been budget-limited since spring, the bid strategy that outgrew its data.
How to use the score (and where to push back)
The 42 points in this checklist are not equal. Some of them are things Google has documented in its own help center: sitelink counts, image-asset lift, location-targeting defaults, the recommended minimums for Smart Bidding. If you fail one of those, you are almost certainly leaving conversions on the table, and the scorecard shows you which page of Google's documentation says so.
Others are practitioner judgement. Google documents the feature but prescribes no threshold, so the rule of thumb is ours: 8+ headlines per responsive search ad, 15–30 monthly conversions before target-based bidding starts to pay off, sitelinks pointing at genuinely different pages rather than anchors. Marked as such on every one. If you disagree with a judgement call, you are reading the score correctly: it is a prompt to decide, not a verdict.
A small number of checks are the other way around. We disagree with Google, the scorecard flags the disagreement, and the page explains both sides. Broad match with Smart Bidding on a new campaign, "Presence or interest" location targeting, consolidated campaign structures: Google actively recommends these. We mark them down when they're applied without thinking. Failing one of those is a prompt to make a decision, not evidence you've made a mistake.
Practically: the score is most useful as a before/after, not a single number. Run it this quarter, fix the cheapest high-impact items, run it again next quarter. A score that moves is a working account; a flat one is a checklist you skimmed.
Sources
Primary references used to verify the checklist and this guide, in the order they first appear:
- Google Ads Help — Conversion tracking diagnostics
- Google Ads Help — Primary and secondary conversion actions
- Google Ads Help — Conversion counting: One vs Every
- Google Ads Help — About enhanced conversions
- Google Ads Help — Set conversion values
- Google — Data-driven attribution is the new default
- Google Ads Help — About keyword matching options
- Google Ads Help — Keyword Planner forecasts
- Google Ads Help — Campaign structure
- Google Ads Help — Low search volume keywords
- Google Ads Help — Search terms report
- Google Ads Help — Add negative keywords
- Google Ads Help — Ad group structure
- Google Ads Help — Exclude placements and content
- Google Ads Help — Limited by budget status
- Google Ads Help — Location targeting settings
- Google Ads Help — About responsive search ads
- Google Ads Help — About Quality Score
- Google Ads Help — Responsive search ad best practices
- Google Ads Help — About sitelink assets
- Google Ads Help — About callout assets
- Google Ads Help — About image assets for Search campaigns
- Google Ads Help — About call assets
- Google Ads Help — About landing page experience
- Google Ads Help — About Smart Bidding
- Google Ads Help — About Target CPA bidding
- Google Ads Help — About the learning period
- Google Ads Help — About Google Search Partners
- Google Ads Help — Brand exclusions in Performance Max
- Google Ads Help — About audience targeting
- Google Ads Help — Targeting vs observation
- Google Ads Help — About auction insights