Managing Many Google Ads Accounts with AI: My Weekly Routine

MCC or direct access

Some of my clients are under my manager account (MCC). Others added my Google login directly to their account, usually because they want to keep ownership simple. In the Google Ads interface that means switching between views. Through MCP Ads it does not matter: I connect each Google login once, and google_ads_list_customers lists every account those logins can reach, expanding any manager account into its client accounts. If you have the choice, an MCC is still tidier for billing and access. But you do not need to move a client into it to manage them from Claude.

List all my Google Ads accounts with their customer ID, currency, time zone and which manager account they sit under, if any. Then show last week's spend for each, highest first.

Step 1: campaign performance, completed weeks only

I start every account with the same view: spend, conversions and cost per conversion by campaign, for the last completed Monday-to-Sunday week against the one before. A week that includes today is not a fair comparison, so I give the dates. google_ads_campaign_performance returns these per campaign, including search impression share, which I use in step 4.

Step 2: wasted search terms

Search terms are where the money leaks on most accounts I take over. google_ads_search_terms returns the actual queries that triggered ads, with spend and the campaign and ad group they came from. For Performance Max, Google only gives search categories without spend, so this step is mainly about Search and Shopping. I ask for the terms grouped by theme, because a list of 300 queries is unreadable and a list of eight themes is not. Then I read it. This is the step where context matters: a term that looks irrelevant to the AI may be exactly what a client sells.

For the plumbing client's Google Ads account, show search terms from the last 30 days that spent money with zero conversions. Group them into themes (jobs, DIY, competitor names, locations we do not serve, and so on), show total spend per theme, and list the terms under each. Do not add anything yet.

Step 3: negatives, after review

Once I have crossed out the terms I want to keep, I ask for the rest to be added. google_ads_add_negative_keywords adds negatives at campaign level, and google_ads_bulk_add_search_terms_as_negatives takes search terms straight from the report and excludes them where they came from. I prefer exact match for single odd queries and phrase match for a whole theme like "jobs" or "salary". For accounts where the same junk appears every month, I keep a shared negative list and apply it to new campaigns instead of adding the same words again.

Add the terms from the "jobs" and "DIY" themes as phrase-match negatives, and the competitor terms as exact-match negatives, each to the campaign it came from. Skip anything I crossed out above. Show me the full list before you add them, then wait for my OK.

Step 4: budget pacing and impression share

Pacing is a mid-week check. google_ads_budget_pacing shows how each campaign in one account is tracking against its budget for the month, and google_ads_mcc_budget_pacing runs the same check across every account under a manager account. The question I ask is not "what did we spend" but "where will we end the month at this rate".

Impression share tells me whether a campaign is limited by budget or by rank. If a campaign that converts well is losing impression share to budget, that is a conversation with the client about moving money. If it is losing to rank, that is ads, keywords or bids. google_ads_auction_insights adds who else is showing in the same auctions.

• Which campaigns in all my Google Ads accounts are on track to overspend or underspend this month?
• For the campaigns with the best cost per conversion last month, how much impression share did they lose to budget and how much to rank?
• Who showed up most often in the same auctions as the Brand campaign in the last 30 days?

Step 5: change history

With many accounts, some of them shared with the client's own team, I need to know what changed and who changed it. google_ads_change_history returns recent changes with the time and the user. Google keeps change history for 30 days, so this is a weekly habit, not a quarterly one. When performance moves for no obvious reason, this is the first place I look. A budget edited by someone at the client, or a paused ad group nobody mentioned, explains more drops than any algorithm theory.

Step 6: building campaigns, paused

I build new Search campaigns from Claude when the structure is clear: a budget, a campaign, ad groups by theme, keywords and responsive search ads. google_ads_create_campaign creates the campaign paused by default, so nothing in it can spend until the campaign is enabled. I then open the campaign in Google Ads, check the settings I care about (locations, networks, bidding), and enable it myself.

The time saving is not in the clicking. It is in going from a brief to a reviewable draft in one conversation, and in having the AI check the draft against the account's existing negatives and campaigns before I look.

In the plumbing client's Google Ads account, build a paused Search campaign for "emergency plumber" in the three cities they serve. Daily budget: for example 40 in the account currency. Two ad groups (emergency, boiler repair), phrase and exact match keywords, one responsive search ad per ad group with 10 headlines and 4 descriptions. Add the account's shared negative list. Show me the full plan first and wait for my OK before creating anything.