AI Optimization Checklist: The Setup for Every New Account

An AI optimization checklist is the list of setup steps that make automated marketing systems work on accurate information. For a new account it covers access and ownership, analytics and conversion tracking, consent handling, conversion values, product or lead data, content and structured data for AI search, automation settings and a review routine. Completing it before optimising prevents weeks of algorithms learning from wrong signals.

Why a checklist instead of best guesses?

Automated systems in Google Ads, Meta, Microsoft Advertising and analytics tools behave predictably when their inputs are clean. When something goes wrong, the cause is rarely obvious from the dashboard. A missed duplicate tag or a wrong currency can look like a bidding problem for a month.

A fixed checklist makes the setup repeatable across accounts and team members. It also gives clients a clear record of what was checked and when.

What has to be in place before any optimisation?

1. Access and ownership

Start with the boring part, because everything else depends on it.

• The client owns the ad accounts, analytics properties, tag manager containers and Search Console properties. The agency has user access, not ownership.
• Admin access is limited to people who need it; others get standard or read-only roles.
• Billing is set up in the client’s name, with payment alerts going to someone who reads them.
• Business Manager or Meta Business Suite, Google Ads manager accounts and Merchant Center are linked correctly.
• Two-factor authentication is enabled for every user.

2. Analytics that measures the right things

Set up or audit Google Analytics 4, or the analytics tool the client uses. Define key events that reflect business value: purchases, qualified form submissions, booked calls, trial starts. Remove page-view-based “conversions” from reporting unless they genuinely matter.

Check cross-domain tracking if checkout or booking runs on another domain, exclude internal traffic and spam referrals, and link GA4 to Google Ads and Search Console. Set data retention to the longest available period for explorations.

3. Conversion tracking in each ad platform

Every ad platform needs its own tracking for its algorithms. Confirm each one works and reports values that match reality.

For Google Ads, Use the Google tag or GA4 imports for conversions, enable Enhanced Conversions, and choose primary goals carefully. Primary goals drive bidding; secondary ones are for observation.

For Meta, install the pixel and the Conversions API, ideally through a server-side setup or a platform integration such as Shopify’s. Check event match quality in Events Manager and deduplication between browser and server events.

Microsoft Advertising uses the UET tag, TikTok its pixel and Events API, LinkedIn its Insight Tag and Conversions API. Use the same event names and values across all of them where possible.

Finish by making a test purchase or form submission and checking that each platform records exactly one conversion with the right value.

Data inputs the algorithms learn from

4. Consent and privacy settings

In regions with consent laws, such as the EU, UK and Turkey under KVKK, tags must respect visitor choices. Implement a consent management platform and Google Consent Mode v2, so Google can model conversions for users who decline cookies. Confirm that tags fire only after consent where required, and document what each tag collects in the privacy policy.

5. Conversion values for automation

Value-based bidding performs best when values reflect what matters to the business. Decide with the client:

• Revenue or gross margin for e-commerce. Margin is better if product margins vary widely.
• Lead scores or stage values for lead generation, imported from the CRM as offline conversions.
• Subscription value, such as expected first-year revenue, for SaaS and memberships.
• Whether to exclude returns or cancelled orders through conversion adjustments.

6. Product and lead data quality

For retailers, the product feed is a primary input for Shopping, Performance Max, Meta Advantage+ catalogue ads and marketplaces. Check Merchant Center diagnostics, fix disapproved items, write clear titles with the attributes people search for, and add custom labels for margin or season so campaigns can be split by business priority.

For lead generation, check that the CRM captures source, campaign and click IDs (gclid, fbclid and similar) so offline conversions can be matched back.

8. Automation settings and guardrails

Review every automated feature and make a deliberate choice for each:

• Google Ads auto-applied recommendations: usually turned off, with recommendations reviewed manually.
• Automatically created assets and final URL expansion: on or off depending on brand rules and site structure, with page exclusions set.
• Broad match: used only with Smart Bidding and solid negative keyword lists.
• Meta Advantage+ creative enhancements: reviewed for brand compliance.
• Brand exclusions and placement exclusions for brand safety.
• Budget caps and automated rules that alert when spend rises unexpectedly.

9. Baseline and review cadence

Before any optimisation, record the starting point: spend, conversions, cost per conversion, ROAS or lead quality by campaign for the last 30 and 90 days. Store it in a shared document.

Then agree on a review rhythm. A common pattern is a weekly check on tracking health and spend pacing, a two-weekly review of performance and tests, and a monthly report that reconciles platform data with business data. Keep a change log so every shift in results can be linked to a cause.

Mistakes that undo a good setup

A checklist completed once can still drift. Website redesigns, new checkout apps, theme updates and tag manager edits often break tracking quietly. A new team member may switch on a recommendation that the original setup turned off. Customer lists go stale when nobody refreshes them.

Protect the setup with a few habits:

• Re-run the test conversion after every site release that touches checkout, forms or tags.
• Keep the tag manager container versioned and document what each version changed.
• Review Google Ads change history and Meta activity history monthly for unexpected edits.
• Refresh customer list uploads at least quarterly, or connect them through an automatic integration.

Where to start this week

Take one account and work through sections 1 to 3 in a single session: confirm ownership, verify analytics key events and run a test conversion through every ad platform. Note every discrepancy you find in a shared document. The remaining sections can follow over the next two weeks, but these first three usually reveal the problems that matter most.

Keep reading: AI Optimization · AI Optimization