A Google Ads campaign is running too expensive, so you lower the bids. It doesn't help. You increase the budget. The number of conversions goes up, but so does their cost. You switch bidding strategies, add negative keywords, and after a few days, it's hard to tell which change actually made a difference.
The problem is often not that the account is being optimised poorly. The problem is the order of operations.
If Google receives bad conversion data, even a well-configured Smart Bidding setup will steer the campaign in the wrong direction.
That is why we always start with measurement. Then we check goals, campaign structure, user search terms, and the landing page. Only at the very end do we touch bidding strategies and budgets.
1. First, check what Google Ads counts as a conversion
This is the single most critical step in any optimisation process.
Google Ads has no idea what actually happens inside your business. It only knows what is sent back to it.
If your real business value comes from a submitted form, but you also track visits to the contact page as a conversion, the algorithm will happily find plenty of people who love visiting your contact page. That doesn't mean any of them will ever send an enquiry.
The same goes for an online shop. Adding a product to cart, initiating checkout, and completing a purchase should never be treated by the algorithm as three equal business outcomes.
Before adjusting budgets or bids, audit all conversion actions and answer one straightforward question:
Is Google optimising for the exact event I genuinely want to pay for?
2. Separate primary conversions from secondary ones
Google Ads allows you to mark actions as primary or secondary.
Primary conversions populate the "Conversions" column and directly drive automated bidding. Secondary conversions are primarily for observation.
This lets you measure more user interactions without telling the algorithm that every single one carries the same commercial weight.
You can monitor form starts, phone number clicks, or add-to-carts, while keeping the actual form submission or completed purchase as the sole primary goal.
Pay close attention to custom goals. Review their setup separately, as they can influence bidding independently of standard primary and secondary action assignments.
3. Verify that conversions are actually reaching Google
Seeing a conversion action neatly named in your dashboard does not mean tracking is actually working.
You need to test the entire user journey end-to-end.
In an online shop, place a test order. For a lead generation campaign, submit the form. Confirm that the event fired, that user consent was passed correctly, and that the conversion shows up in the right place.
If you use enhanced conversions, check their diagnostic reports as well.
Enhanced conversions use hashed first-party data provided by users—such as email addresses or phone numbers—to improve matching back to ad interactions.
The less data reaching the platform, the less context your automated bidding strategies have to work with.
4. Consent Mode will not magically recover all missing data
This is especially critical for smaller accounts.
Google can model some conversions that couldn't be observed directly due to a lack of user consent, but the algorithm needs sufficient data volume to do so.
According to Google's documentation, conversion modelling requires at least 700 ad clicks over a 7-day period for a specific domain and country pair.
If your shop generates 300 clicks a week in Poland, you cannot simply assume missing conversions will be filled in by the model.
This has real practical consequences.
If some sales never register in Google Ads, your reported cost per acquisition will look worse than it is, and ROAS will be understated.
More importantly, Smart Bidding also trains on the data visible in the system.
5. For e-commerce, check your product feed
Before you start tweaking Performance Max, ensure Google can actually serve ads for the products you want to sell.
Errors in Merchant Center, missing attributes, price discrepancies, or stock issues can quietly exclude parts of your inventory.
When that happens, the problem isn't your bid. Your product simply isn't entering the auction.
Always review:
- how many products are active,
- how many are disapproved or limited,
- whether prices and availability match the live store,
- whether product IDs remain stable,
- whether titles and descriptions accurately reflect what you sell.
Only evaluate shopping campaign performance once the product feed is clean.
6. Stop changing strategies every few days
Automated bidding strategies need time and data.
Changing strategy settings, major restructuring, or reactivating bids can trigger a new learning phase.
If you tweak your CPA on Monday, adjust the budget on Wednesday, and rebuild ad groups on Friday, it becomes nearly impossible to know what actually drove any change in results.
Post-change calibration can take weeks, especially depending on the length of your conversion cycle.
Effective optimisation is not about making frequent adjustments.
It's about making a change, collecting sufficient data, and only then deciding on the next step.
7. Set Target CPA based on real data, not wishful thinking
Target CPA is not a wishlist where you tell Google what you'd like to pay for a customer in an ideal world.
It is an operating parameter the bidding algorithm uses to make decisions in live auctions.
If a campaign historically brings in conversions at 100 PLN, setting a target of 40 PLN won't magically make the system find those same customers at a 60% discount.
Instead, it will likely restrict your auction participation and slash your volume.
The reverse is also true. If a campaign consistently hits a CPA around 60 PLN while your target remains set at 100 PLN, check whether that setting still aligns with what you need from the campaign.
Your target must stem from actual performance data and unit economics, not from an arbitrary figure entered during initial setup.
8. Don't switch to ROAS if every conversion carries the same value
Value-based bidding strategies only make sense when the system actually receives dynamic values.
In an online shop, this happens naturally: a 300 PLN order and a 3,000 PLN order are completely different commercial outcomes.
In lead generation campaigns, you can do something similar if your business qualifies leads and passes their actual values back from your CRM to Google Ads.
However, if every form submission is arbitrarily assigned a flat 100 PLN value regardless of quality, the bidding algorithm gets zero additional insight.
Switching from maximising conversions to maximising conversion value won't improve campaign performance in that scenario.
9. Don't over-segment campaigns without a clear reason
When auditing incoming accounts, we repeatedly see the same pattern: a modest conversion volume fragmented across too many campaigns.
If an account generates 20 conversions a month spread across six similar campaigns, each campaign receives very little data.
This starves automated bidding strategies and makes learning far harder.
A separate campaign only makes sense when there is a clear business reason, such as:
- a distinct budget,
- a different market or language,
- different profit margins,
- a distinct conversion goal,
- the need to isolate branded traffic from generic search.
Creating a separate campaign for every product category just to make the account look tidy only fragments your data unnecessarily.
10. Open the search terms report
This is one of the easiest places to uncover wasted ad spend.
Look at the actual queries users typed before landing on your site.
Don't just look for weird one-off queries. Look for patterns.
If an intent you don't serve appears regularly, add an appropriate negative keyword. If the issue affects multiple campaigns, use a shared negative keyword list.
Keep in mind that the report doesn't display every single query. Low-volume searches are hidden by Google.
Treat the search terms report as a representative sample of user behaviour, not an exhaustive query log.
Performance Max can report impressive numbers in the interface while driving far less incremental sales than the figures imply.
One key element to verify is your own brand traffic.
Someone searching for your brand name already knows your business and is much closer to buying than someone searching generic product terms.
If Performance Max captures a large share of that branded traffic, campaign stats will look stellar, but they won't tell you how much net-new demand your ads actually generated.
Review brand exclusions, the search terms report, and the relationship between your standard Search campaigns and Performance Max.
Performance Max is heavily automated, but that doesn't mean your job ends once you submit a budget and a few creative assets.
You can steer it using:
- negative keywords,
- brand exclusions,
- search themes,
- the search terms report,
- creative assets,
- final URL expansion settings.
Always review these levers before assuming the campaign simply "decides everything on its own".
A new campaign needs a ramp-up phase, and performance during the first few days can be erratic.
Google recommends factoring in an initial learning period and evaluating Performance Max across an adequately long date range. When analysing results, look at at least several weeks of data, provided campaign volume is sufficient.
If you tweak budgets, targets, assets, and campaign structure every few days, you are constantly assessing a campaign operating under different conditions.
That makes it impossible to separate genuine performance gains from normal baseline volatility.
14. Inspect your landing page
Not every campaign problem can be solved inside Google Ads.
If visitors land on a slow website, can't find the offer promised in the ad, or struggle to submit a contact form on mobile, ongoing bid tweaks won't fix the underlying issue.
Landing page experience also directly affects your ad quality scores and auction eligibility.
During optimisation, verify:
- whether the ad directs users to the most relevant landing page,
- whether visitors immediately see the offer promised in the copy,
- whether the page loads fast on mobile devices,
- whether the form successfully transmits submissions,
- whether the form asks for unnecessary fields,
- whether submissions trigger conversions reliably.
If CPC remains stable but conversion costs climb, pay close attention to your landing page conversion rate.
You can also check our guide on how to speed up a WordPress site.
15. Only now should you increase the budget
A bigger budget won't fix a campaign that tracks conversions incorrectly, attracts the wrong audience, or sends traffic to an underperforming landing page.
It will simply burn through more budget, faster.
Scaling budgets makes sense once:
- conversion tracking is rock-solid and verified,
- the campaign meets your target CPA or ROAS,
- the campaign is genuinely budget-constrained,
- there is sufficient historical conversion volume,
- your team can handle additional sales volume or inbound leads.
At that point, the question is no longer "is this campaign working?", but "how much further can we safely scale it?".
In what order should you optimise Google Ads?
If we had to distill the entire process down to a single checklist, we start with the things that could invalidate every downstream insight.
- Measurement. Verify that conversions are tracking accurately.
- Goals. Define which actions should steer bidding decisions.
- Consent and enhanced conversions. Check the quality and integrity of data flowing into Google Ads.
- Product feed. In e-commerce, check for errors, disapproved items, price mismatches, and stock status.
- Structure. Consolidate campaigns that needlessly fragment data, and split only those requiring dedicated budgets or targets.
- Search terms. Identify irrelevant intent and build out negative keyword lists.
- Performance Max. Audit brand traffic, queries, search themes, and creative assets.
- Landing page. Test speed, messaging, and form functionality—especially on mobile.
- Bidding strategy. Align CPA or ROAS targets with historical performance and real unit economics.
- Budget. Scale spend only once the foundation is operating cleanly.
After any major change, give the campaign time to collect fresh data.
The worst way to optimise is to alter multiple variables at once and attempt to evaluate the outcome three days later.
The core rule: the algorithm executes the goal you set for it
Google Ads features more automation every year, but automation cannot fix a poorly defined goal.
If you tell the system that visiting the contact page is your primary event, it will hunt for users who love looking at contact pages.
If every lead has an identical nominal value, the algorithm cannot know which one turned into a 500 PLN sale and which became a 50,000 PLN contract.
If a portion of sales never makes it into tracking, the algorithm won't know they happened either.
That is why the most impactful optimisation work often happens outside the campaign settings screen: in your analytics, CRM, product feed, or landing page.
When is it worth outsourcing your campaigns?
At DOCK, we manage Google Ads and performance campaigns and Meta Ads alongside complete analytics setup and technical website development.
This matters most when the root cause of poor performance lies outside the ad dashboard itself.
Whether you need to fix tagging, implement Consent Mode, configure CRM integrations, refine a product feed, streamline forms, or improve page speed, we take you straight from campaign diagnosis to technical remediation.
If you want to know what's holding your campaigns back, get in touch with us. We can audit your tracking, ad account, and landing pages, and give you a clear roadmap of what to fix and in what order.
