Google Ads experts

Structure your campaigns and steer your bidding on business outcomes.

We build and run your Google Ads accounts, and we feed the data that matters to your business all the way into the bidding.

Book a discovery call

Learn more about: our SEA and Google Ads agency

Partner Premier

EdgeAngel Google Ads expertise

Building the account, steering the bidding, securing the signal

Account structure and campaigns

The account tree decides what you can arbitrate: a budget can only be steered once it is isolated.

Architecture by intent: campaigns and ad groups split along your objectives, your catalogue and your markets, rather than along the account history
Manager accounts: MCC hierarchy, labels and access, up to six levels between the root account and the final one
Keywords and match types: broad match, phrase match, exact match, and the upkeep of the exclusions that keeps an account clean
Responsive search ads: headlines and descriptions written on your real search terms, three active ads per group at most
The formats that have their own page: Performance Max, Demand Gen and Google Shopping each follow their own rules
Microsoft Ads alongside: the same setup carried over to Microsoft Ads, with its account import and its LinkedIn audiences

Bidding strategies and steering

Automated bidding optimises towards the objective it is given. We build that objective on your real costs.

tROAS and target CPA: built on your margin and your costs, rather than on the revenue the platform reports
Bid portfolios: one strategy shared across campaigns when a single one lacks the volume to learn
New customer acquisition: the distinction between a new and an existing customer fed into the bidding objective
ROAS tolerance: since version 25 of the API, an accepted degradation margin to widen traffic, set to the thousandth of a percent
Budget arbitration: what goes to growth, what goes to profitability, reviewed as the signals come in

Measurement and signal quality

Campaigns learn from what comes back to them. That is where the performance of an account is decided.

Enhanced conversions: first-party data normalised and SHA-256 hashed, sent through the tag, through Google Tag Manager or through the API up to 24 hours after the conversion
Offline conversions: calls, store visits, CRM signatures, sent back with their value and their click identifier
Google tag gateway: the tag served from your own domain, through your CDN or your server, for measurement that survives blocking
Consent Mode: the seven consent types wired to your CMP, with collection continuing cookie-free for refusals
Data Manager API: audiences and conversions sent to several Google products in a single call, the migration from Customer Match included
First-party audiences: Customer Match, value segments and exclusions, fed from your warehouse

Automation, API and agents

What repeats gets coded. The time freed goes into analysis and trade-offs.

Google Ads API: one major version every three to four months, a lifespan of about twelve months, so at most two upgrades a year to keep up with
Google Ads scripts: alerts, delivery checks and bulk imports, in JavaScript inside the account, with no infrastructure to operate
MCP server: Google's official server, read-only, and our own servers to query your accounts from a language model
Google Ads Editor: bulk edits and offline work, on high-volume accounts
Pre-launch checks: the actual settings read back through the API before activation, never on declaration

The tool

Google Ads, our reading

Google Ads is Google's advertising platform: Search, Shopping, YouTube, Display, Discover and Gmail, run from a single account. Its hierarchy is stable, account, campaign, ad group, ad, and its documented limits give the measure of what an account can carry: 10,000 campaigns per account, 20,000 ad groups per campaign, five million criteria, an 80-character keyword, a 30-character headline.

What has changed over two years comes down to one name: AI Max. Google itself describes it as « an optimization layer » rather than a new campaign type. Out of beta on 15 April 2026, it has been rolling out automatically onto dynamic search campaigns since September 2026, and the API has exposed the fields that track that migration since version 25.1, on 19 August 2026.

Updated September 2026

A platform that executes very well against what it is told to aim at

The strength of Google Ads is the reach of its inventory and the maturity of its automated bidding. The strategies carry precise names and do precise things: maximise conversions with an optional target CPA, maximise value with a target ROAS, reach a target impression share. What most accounts miss lies elsewhere: these algorithms optimise towards the objective they are handed. An account that sends a purchase along with its margin asks Google to go after profit, and gets it.

An AI layer taking over, and guardrails that stay explicit

With AI Max enabled, the campaign-level match type setting is deprecated and every keyword is treated as broad match, unless it is switched off at ad group level. Ad text and landing page can be picked by the model. Google documents three guardrails alongside, each answering a distinct problem: a bad query is handled by a negative keyword, a bad destination by a URL exclusion, brand cannibalisation by a brand list. This layer moves the configuration work, it does not remove it.

Where the platform stops, and what takes over

Three limits are documented by Google itself and deserve to be known before arbitrating a budget. The search terms report is truncated: queries below a volume threshold are removed from it, while their clicks still count in the totals. Attribution stays inside Google: the models offered observe the clicks Google sees. And a successful conversion import does not mean an attributed conversion, as Google states plainly. To know what a channel truly creates, you have to step outside the tool: incrementality testing, geo-lift, marketing mix modeling.

Advanced Media Measurement, our advertising effectiveness offering

Sources : AI Max for Search campaigns, developer documentation · System limits, Google Ads API

Our expert notes : Google Ads: optimise your tracking and capitalise on first-party data · Google Ads: the wbraid and gbraid parameters for conversion tracking · Customer Match: migrating to the Data Manager API · Dynamic remarketing and Google Analytics

Frequently asked questions about Google Ads

Should you switch AI Max on for your Search campaigns?

It is worth testing, campaign by campaign. AI Max widens keyword matching, may rewrite ad text and pick the landing page. On an account whose exclusions and brand lists are maintained, the gain is real. On a loosely framed account, that widening costs. Google has in any case started migrating dynamic search campaigns automatically since September 2026.

Which bidding strategy should you choose?

The one that matches what you can measure. Without a reliable conversion value, a target ROAS optimises on a wrong number. So we start with measurement: what comes back, with what value, and within what delay. The strategy is decided afterwards, and it is built on your real costs rather than on reported revenue.

Why do campaign totals not match the search terms report?

Because Google removes queries below a volume threshold from the report, for privacy reasons, while still counting their clicks in the campaign totals. The gap is structural and documented. It does not signal a tracking problem, but it does rule out rebuilding spend per query from that report alone.

Is Google Ads enough to know whether your campaigns are profitable?

It reports what it observes, that is the conversions tied to the clicks it served. It does not say what would have happened without the campaigns. To establish the share genuinely created, you need an incrementality test, a geo-lift or marketing mix modeling. We frame those tests before running them, and we say when they are not worth it.

Can you automate the management of a Google Ads account?

Partly, along three routes. Scripts, in JavaScript inside the account, cover alerts and checks with no infrastructure to operate. The API covers the rest, with a major version every three to four months. Google's official MCP server finally allows querying an account from a language model, read-only.

Contact

You leave your details, we get back to you within 24 hours.

A first call to understand your context, and tell you what is feasible and under which conditions.

Contact details
Defining your need
Then we schedule a call

Your details are used to reply to you on this topic. Data protection policy

Or call us directly: +33 1 84 16 42 20

Paul Schmitt

Consulting Director

"Our goal is to make your data actionable to generate concrete value, quickly."

The team