Service 12 · Paid Search
Google automated the bidding; it didn't automate the judgment. We manage the inputs the machine learns from, conversion truth, value signals, negatives, structure, across Search, Performance Max, Demand Gen, Shopping, and YouTube, so spend answers to revenue instead of click counts.
Modern Google Ads is an automation you feed, and most accounts feed it garbage: duplicate conversions, form-fill counts with no quality signal, budgets fragmented across ad groups that starve Smart Bidding of data. Our first weeks fix the diet, enhanced conversions recovering matches lost to cookie decay, server-side tagging, Consent Mode v2, and offline conversion import so the algorithm learns from what actually closes, not what merely clicks.
Then we teach it what revenue looks like. Tiered conversion values by funnel stage, an MQL worth $100, an SQL $900, an opportunity $3,000, closed-won at deal value, mean one real opportunity outweighs thirty cheap leads in the auction. Without value data, Smart Bidding optimizes for the cheapest lead, which in B2B is usually the worst one.
Structure follows intent, not keywords. Single-keyword ad groups are dead; close-variant matching collapsed their precision years ago. We consolidate into intent-themed campaigns, research queries separated from commercial ones, each with distinct messaging, full RSA asset coverage, account-level negative lists from day one, and the 30+ monthly conversions Smart Bidding needs to learn.
Performance Max is a strong scaling layer and a poor first campaign. We deploy it once tracking is verified: asset groups split by product or service theme, first-party audience signals, brand exclusions, and campaign-level negative keywords, up to 10,000 per campaign since 2025, a control worth using now that it exists.
Google's newest features get tested, never toggled. AI Max for Search, Smart Bidding Exploration, broad match, every one runs through the Experiments tool as a 50/50 split judged on pipeline, not click volume. Google reports around 7% more conversions from AI Max at similar CPA; we verify whether that holds in your account before it earns budget.
Weekly cadence does the unglamorous work: search-terms triage with phrase-match negatives, budget shifts toward proven value, and a standing target of cutting bad-intent spend in half within the first 30 days.
Enhanced conversions, server-side tagging, Consent Mode v2, offline import, verified before spend moves.
Search terms tagged by intent; the cost of junk queries computed as a share of total spend.
Consolidated campaigns by query intent, sized for the 30+ monthly conversions bidding needs.
Tiered values per CRM stage feed Target ROAS, so bidding chases revenue, not lead count.
Broad match, AI Max, and bidding changes run as 50/50 splits judged on pipeline.
PMax with brand exclusions and negatives, Demand Gen, and weekly triage that keeps waste cut.
Modern Google Ads is an automation you feed, so the program manages the inputs: conversion truth, value signals, negatives, and structure. Nothing structural changes without an experiment. The first weeks go to making sure the machine learns from revenue instead of noise.
What we do
What you get
What we do
What you get
What we do
What you get
What we do
What you get
What we do
What you get
What we do
What you get
Weeks 1–2
Every week
Every month
Every quarter
The stack we run for this
Google Ads
Google Analytics 4
Looker Studio
HubSpot
SemrushCross-industry studies put the average CPC near $5.42, ranging from under $2 in restaurants and travel to $8+ in legal and home services. Average cost per lead sits around $70 but varies four to five times by industry. The practical budget floor is whatever buys roughly 30 conversions a month; below that, Smart Bidding never gets enough data to learn.
Yes, once the foundations exist: verified conversion tracking, 20 to 30 or more monthly conversions, quality assets, brand exclusions, and negative keywords. It is a poor first campaign and a strong scaling layer. Channel-level reporting, added in 2025, finally shows where its budget goes, and we review it like any other placement decision.
Google's campaign enhancement, launched in 2025, that adds keywordless matching and AI-generated text to Search campaigns. Google reports roughly 7% more conversions at similar CPA. We test it through a 50/50 experiment judged on pipeline rather than flipping it on, and for B2B accounts we often disable asset optimization to protect messaging precision.
Expect two to four weeks of learning after launch or any major change, a first reliable read at 30 to 60 days, and compounding efficiency over 90 days as negatives and value data accumulate. Frequent bid and budget changes reset learning, which is why we adjust weekly at most and judge performance on monthly windows.
No. Close-variant matching collapsed the precision SKAGs promised, and fragmenting conversions across dozens of ad groups starves Smart Bidding of learning data. We build consolidated, intent-themed campaigns: exact and phrase cores, distinct messaging per intent stage, and broad match introduced only through experiments once value data is clean.
Adjacent disciplines
