Google Ads optimisation: the complete SME guide

Back when Google Ads was still called AdWords, optimising an account mostly meant bidding, refining keywords and testing ad copy. In 2026 you mostly steer an AI system. According to Google (April 2026), more than 30% of its customers’ search ad spend in the first quarter of 2026 ran through AI campaigns such as AI Max and Performance Max, and since 1 September 2026 Google has been automatically converting the first classic settings to AI Max. Your levers therefore shift from micromanagement to data, target values and creatives. In this guide you get the data-driven optimisation routine we use at ClickForest for the Google Ads accounts of SMEs in Flanders: what you do weekly and monthly, how to read Quality Score and optimisation score correctly (both different from what the myths claim), how to steer Performance Max and what to do when it underperforms, what AI Max changes, and why profit is your compass.

How do you optimise a Google Ads campaign in 2026?

By working on the inputs that steer the AI; the system handles the bids itself. Concretely, those are your conversion measurement and target values, your search terms and exclusions, your ad copy and assets, your landing pages and your budget allocation. Get those five inputs right and adjust them in a fixed rhythm, and you beat virtually every daily knob-turner.

The fixed rhythm we use:

Rhythm	What you do
Weekly	Go through search term reports (exclude and promote), check budget allocation, explain outliers in costs or conversions
Monthly	Refresh ad copy and assets, review audience and location data, compare landing pages against your best performing search terms
Quarterly	Review bidding strategy, target values and account structure, plus evaluate one structural experiment

That rhythm has a reason: according to Google, a bid strategy can need up to around 50 conversions or three conversion cycles to adjust after a change, and making big changes every day restarts that learning process every time. How this fits into your broader channel approach is covered in our pillar on what performance marketing is.

Where does every optimisation start? With your measurement

Every Google Ads optimisation stands or falls with conversion measurement: if the system counts the wrong actions as a “conversion”, the AI confidently optimises in the wrong direction. So first check what counts as a conversion, whether the values are correct, and whether your measurement holds up with Consent Mode for European traffic.

For traffic from the EEA, Google has required two extra consent signals since early March 2024, ad_user_data and ad_personalization; without those signals you lose measurement, remarketing and personalisation features. In the advanced version of Consent Mode, your tags send cookieless pings when a user declines, which Google uses to model conversions for your account when there is enough data.

The biggest lever for e-commerce sits in the conversion value you send along: pass revenue and the AI steers on revenue; pass margin and it steers on profit. That difference is bigger than any other optimisation in this guide, and it is why at ClickForest we prefer to steer on POAS over bare ROAS; the worked example is in POAS versus ROAS. American benchmark data underlines why profit steering is becoming more urgent: in the 2026 WordStream/LocaliQ report (13,474 American campaigns, April 2025 to March 2026), the average cost per click still rose by about 3% to $5.42, while cost per lead fell for the first time since 2020, from $70.11 to $66.69. Senior director Pepe Hernandez on those figures:

“Seeing an overall drop in cost per lead for the first time since before 2020 is a big signal that advertisers who’ve adapted to search platform changes are getting more stable and predictable results.”

— Pepe Hernandez, Senior Director of Optimization Strategy at LocaliQ (May 2026)

This is American data in dollars, so it gives a direction for Belgium; your own campaign data remains the yardstick. And one observation from the 2025 report of that same benchmark study deserves your attention when choosing a bidding strategy:

“The rate of CPC increases varies depending on optimization type, but we’ve seen sharper increases on campaigns with smart bidding, which is likely expected since Google has direct control over these CPCs.”

— Katia Hausman, in 2025 Vice President of Product at LocaliQ (May 2025)

Smart bidding pays off, but it does not release you from vigilance over what a click costs you.

How do you read Quality Score correctly?

As a diagnostic tool. Google says it literally in its own documentation: the visible Quality Score from 1 to 10 is “not an input in the ad auction” and not a KPI. The score consists of three components, each rated below average, average or above average: expected clickthrough rate, ad relevance and landing page experience.

The right approach: use the three components as a map for improvement work. If expected clickthrough rate scores below average, your ads do not match the search intent; if landing page experience scores low, you lose visitors after the click (which also touches your conversion optimisation). The underlying quality signals do count in the auction, so this improvement work usually lowers your costs; only staring at the number itself is pointless. That instantly debunks the most persistent Google Ads myth (“Quality Score is the ranking factor”), in Google’s own words.

Should you follow the optimisation score and recommendations?

Critically, with your own strategy as the test. The optimisation score (0 to 100%) is, according to Google’s own definition, an estimate of how well your account can perform, calculated partly on the open recommendations. In practice the score mainly moves with how you handle those recommendations. Ex-Googler and Google Ads coach Jyll Saskin Gales puts it sharply:

“The optimization score is not a measure of how well your account is performing; it is a measure of whether you are reviewing your recommendations.”

— Jyll Saskin Gales, Google Ads coach and ex-Googler (December 2025)

Practically: go through the recommendations weekly or monthly, apply what fits your strategy, and explicitly dismiss the rest (dismissing also counts towards your score). Turn off auto-apply for anything that touches budgets, bids or keywords; you do not want the system rewriting your strategy overnight. A score of 100% means you manage your recommendations, not that your account is perfect.

Rather have a specialist take over this rhythm? ClickForest carries out Google Ads optimisation for SMEs in Flanders, with profit steering and full transparency. Have your Google Ads managed by us, or book a no-obligation video call.

How do you optimise Performance Max in 2026?

You steer Performance Max with four levers that really matter: your conversion values (the most important), your asset groups with strong copy and images per theme, your search themes as signposts, and your exclusions. That last lever in particular works a lot better since 2025 than it used to.

Campaign-level negative keywords and URL rules for campaigns with a product feed arrived from January 2025, alongside the brand exclusions that have existed since February 2023. Note: negative keywords in Performance Max only apply to Search and Shopping inventory; on YouTube, Display and Gmail they do not stop your ads.

The old “black box” reproach is also partly outdated: from April 2025 Google rolled out channel reporting (you see per channel, from Search to YouTube, Discover, Gmail and Maps, what PMax spends and returns) plus full search term reports at the same level of detail as regular Search campaigns. PMax specialist Mike Ryan already said something about it in 2022, well before those transparency updates, that still holds:

“I think Performance Max is a new advertising platform incubating within the old one.”

— Mike Ryan, then Head of Marketing at Smarter Ecommerce (June 2022)

So treat PMax as a platform of its own with its own rules: feed it correct values and strong creatives, check the channel distribution and search terms monthly, exclude what does not fit, and compare profitability with that of your regular Search campaigns before you trust the total figure. For webshops there is an extra rule: the product feed is an asset; we made the comparison with Meta campaigns in Google Ads versus Meta Ads for Shopify.

Why is your Performance Max underperforming, and how do you fix it?

Usually the problem lies in the inputs the campaign receives. Google writes it in its own help page on Performance Max: “AI is only as good as the data you put into it”. So first check whether you measure the right conversions with the right value, whether your target CPA or target ROAS is not too tight, and whether you supply assets of all types and formats.

Then check whether your search themes still match what you sell and whether you exclude too much. The channel report shows where the budget goes, the search terms report which queries you pay for. If you already capture your brand traffic through Search, keep it out of PMax with brand exclusions or negative keywords. And give every big change the learning period described above. Can’t find the cause? A Google Ads audit maps the leaks in budget, search terms and tracking.

What is AI Max and what does it change in your Search campaigns?

AI Max is the AI feature set inside your existing Search campaigns, out of beta since 15 April 2026, with three parts: broader search term matching beyond your keywords, automatically adapted ad copy based on your site and ads, and automatic landing page selection. You switch it on per campaign, so you can roll it out in a controlled way.

According to Google (April 2026), campaigns using the full AI Max feature set see an average of 7% more conversions or conversion value at a similar CPA or ROAS, compared with legacy search term matching. At the launch in May 2025, Google spoke of typically 14% more, and 27% for campaigns still running mostly on exact and phrase keywords. In July 2026 Google reported that 500,000 advertisers use AI Max.

More important than the percentages is the direction: Google is phasing out the classic alternatives, so it pays to learn to steer AI Max now. Our approach: activate AI Max on one campaign first, watch the search term reports extra strictly in the first weeks (the matching broadens noticeably), and update your exclusion lists before rolling out further. How AI is redrawing the whole advertising trade is covered in performance marketing in the AI era.

What has changed in your Search campaigns since 1 September 2026?

Since 1 September 2026, Google has been automatically converting some Search campaigns to AI Max. Campaigns with automatically created assets get search term matching and text customisation switched on; campaigns with the campaign-level broad match setting get search term matching switched on. Advertisers received an email about it in advance.

Dynamic Search Ads follow later: in June 2026 Google pushed their conversion back to February 2027, following advertiser feedback. Since 15 June 2026 you can create and edit DSA campaigns again; creating them ends in January 2027.

In the weeks after that conversion, keep a closer eye on your search terms report and exclusion lists, because the matching gets broader.

Which bidding strategy do you choose?

Choose based on your goal and data volume: Maximize Conversions (optionally with a target CPA) for lead generation, Maximize Conversion Value (optionally with a target ROAS) for e-commerce. Those are the four Smart Bidding strategies Google offers, and they bid per auction on signals such as device, location, time and query that you could never process manually.

Two practical rules make the difference. First: give the strategy realistic target values; an overly aggressive target CPA or target ROAS smothers your volume, an overly loose one wastes budget. Start from your actual average and move in small steps. Second: respect the learning period after every big change and judge on weekly averages, because daily figures fluctuate too much. And tie the target values to your margins: a target ROAS without margin knowledge is steering on a phantom number, exactly why the profit discussion from POAS versus ROAS hits the core here.

Why are your Google Ads not delivering enough profit?

Most of the time, it is because you let the algorithm steer on the wrong number. Optimising for revenue or ROAS lets the algorithm chase low-margin orders. The campaign then does exactly what you asked while your profit falls. Steering on gross profit reverses that.

Four causes explain most cases, and they rarely sit in the settings:

• You measure revenue without margin. A ROAS of 4 on a product with a 20% margin loses money, while a ROAS of 2 on a 60% margin makes a profit. How to flip that is covered in POAS versus ROAS.
• Your conversion tracking double-counts or counts noise. The AI then optimises on signals that are not revenue, and scales exactly the wrong thing.
• Your budget leaks to search terms that never buy. Without search term hygiene you pay for visitors who have nothing to do with your offering.
• You compare the wrong periods. A campaign that is two weeks old is still learning; drawing conclusions on noise costs more than patience.

Which mistakes do you see most often in SME accounts?

Five classics, in order of damage: wrong or duplicate conversion measurement (the AI optimises on noise), no search term hygiene (budget leaks to irrelevant queries), everything in one campaign (no steering by margin or goal), neglected ad copy (the same assets for a year) and steering on revenue without margins. You can fix all five with basic knowledge and a fixed rhythm.

The common thread is always the same: the problem almost always sits in the boring basics. Get your measurement, structure and search terms in order and you will get more out of an AI campaign than someone stacking advanced scripts on a messy foundation. If you combine channels, also guard the division of roles: how search ads and SEO reinforce each other is covered in integrating SEA and SEO, and the full funnel build-up is in our guide on advertising strategy across the funnel.

How do you know whether your optimisation works?

Measure on three levels, each with its own rhythm: campaign level (costs, conversions, cost per conversion, week on week), account level (total profitability: returns minus ad costs minus product costs, month on month) and business level (what you notice in quotes, orders and revenue, quarter on quarter). A campaign that “performs better” while profit drops is not an improvement.

A practical pitfall when judging: never compare a learning period with a stable period, and factor in seasonal effects before drawing conclusions. Also keep one line of experimentation open: test one structural change at a time per quarter (a new campaign structure, AI Max on an extra campaign, a different target value), so you know what made the difference. That way you see every quarter what makes your account better, and you get figures you dare to steer on; what such a trajectory looks like in practice is shown in the Bastiano advertising story.

What should you remember about Google Ads optimisation?

Optimising Google Ads in 2026 means steering AI with clean inputs: correct profit-value measurement, tight search term hygiene, strong creatives and realistic target values, in a fixed week-month-quarter rhythm. Read Quality Score and optimisation score as diagnosis. If Performance Max underperforms, look at your inputs first. Roll out AI Max step by step, as Google now converts the classic settings itself. Profit is your compass; the rest is means.

ClickForest runs this Google Ads optimisation routine in that fixed rhythm for SMEs in Flanders, as part of performance marketing management with profit steering and full transparency. Wondering whether your account gets the maximum out of it, or considering outsourcing? Book a no-obligation video call and we will look at your account together.