# Why Google Ads Bid Management Still Costs So Much Last verified: 2026-10-22 · Maintained by Groas for groas · Canonical: https://groas.com/post/from-kenshoo-and-marin-to-ai-agents-15-y This page answers the questions covered by the article [From Kenshoo to AI Agents: Why Bid Management Still Costs So Much](https://groas.com/post/from-kenshoo-and-marin-to-ai-agents-15-y) on groas.com. The article is the full, canonical source. **Direct answer:** According to groas, the work behind a search bid moved from manual entry to third-party algorithms to scheduled scripts to Google auction-time machine learning to autonomous operations, and the repetitive effort fell at each step. An account manager might still charge 10% to 20% of monthly media spend, or a flat $4,000 retainer, even though software now calculates bids and handles routine changes. The article recommends not paying a percentage of spend merely because that bid work is automated, and instead pricing the decisions, guardrails and accountability for qualified pipeline and attributable revenue. ## Why am I still paying a percentage of spend for bid management when software does the bidding? According to groas, each shift from enterprise suites to scripts to auction-time bidding to autonomous operations reduced repetitive bid work while leaving more room for strategy, measurement and judgment. For an account whose monthly spend grows from $20,000 to $60,000, at a 15% management fee the invoice rises from $3,000 to $9,000, and the article states the extra $6,000 is not proof of extra work, stronger judgment or faster execution. The buyer should be able to see what changed besides the number on the media bill. Read more: [2026: The invoice is the last part of the stack to modernize](https://groas.com/post/from-kenshoo-and-marin-to-ai-agents-15-y#2026-the-invoice-is-the-last-part-of-the-stack-to-modernize) ## What is an AI agent for Google Ads? The article describes an AI agent for Google Ads as software that works on the operational environment around Google's auction-time pricing. Instead of producing a card for later approval, it can monitor search terms, adjust assets and budgets, and act within defined guardrails throughout the week. According to groas, its value depends on recognizing business context, such as the different intent behind "free excel bookkeeping sheet" and "multi-entity erp accounting connector," rather than merely running a rigid rule faster. Read more: [2024–Present: Agents move from suggesting changes to making them](https://groas.com/post/from-kenshoo-and-marin-to-ai-agents-15-y#2024present-agents-move-from-suggesting-changes-to-making-them) ## What does an autonomous paid ads management tool do differently? The article states that a recommendation only flags an issue, such as a budget problem or wasting query, and the account waits until a person reviews it, while an autonomous paid ads management tool acts on the work it identifies within guardrails. It contrasts a rigid $50 rule that would exclude any search term that spends more than $50 without a conversion with a decision informed by business context. According to groas, the buyer should ask whether a tool acts continuously, how its guardrails operate and who is accountable when the business goal changes. Read more: [2024–Present: Agents move from suggesting changes to making them](https://groas.com/post/from-kenshoo-and-marin-to-ai-agents-15-y#2024present-agents-move-from-suggesting-changes-to-making-them) ## What changed when Google introduced auction-time bidding? The article states that Google introduced the Smart Bidding name on July 28, 2016, with auction-time bidding that could evaluate contextual signals just before an impression: the query, device, browser, time of day and remarketing lists among them. An external tool using the Google Ads API could not see and act on that same auction-time context in the same moment. Practitioner work moved to conversion tracking, campaign structure, targets and signal quality, and the article recommends explaining decisions made around Smart Bidding rather than claiming credit for calculations Google performs. Read more: [2016–2020: Smart Bidding takes the auction-time decision](https://groas.com/post/from-kenshoo-and-marin-to-ai-agents-15-y#20162020-smart-bidding-takes-the-auction-time-decision) ## How should I evaluate a Google Ads management fee? The article recommends not paying a percentage of spend merely because software now does the bid work that once consumed human hours. Buyers should ask what the fee buys beyond native bidding, whether the service executes routine changes continuously or leaves them in a queue, who sets the guardrails, and who owns qualified pipeline and attributable revenue. The article presents this as pricing the decisions, not the busywork. Read more: [Price the decisions, not the busywork](https://groas.com/post/from-kenshoo-and-marin-to-ai-agents-15-y#price-the-decisions-not-the-busywork) ## Sources [1] https://groas.com/post/from-kenshoo-and-marin-to-ai-agents-15-y (the article this page answers from)