Our Method
The mechanics behind the KPIs — not just the vocabulary.
Most agencies can name Quality Score and ROAS. Fewer can explain what actually moves them — across Search, Shopping, Demand Gen, Video, and Performance Max. Here's the version we'd give another specialist.
The Big Picture
One ecosystem. Many opportunities.
Different entry points, different signals — but every campaign type runs through the same auction system underneath. This is the map before the mechanics.
At a Glance
Seven ways to see the difference before the mechanics behind each one.
The same auction ecosystem sits underneath every campaign type. What changes is where intent comes from, what information the system can use, and how much of the execution you actually control.
First, what is a signal?
A signal is information Google can use to decide who's likely to respond, what they might be looking for, or which opportunity is worth pursuing. Search queries, product data, audience data, conversion history, creative interactions, and business goals can all act as signals.
Signals inform the system. They don't instruct it.
Signal ≠ instruction
You provide the signals. Google decides where the opportunity actually is.
A signal doesn't tell Google what to do — it gives the system information to decide with. Performance Max makes this easiest to see, since every input you give it becomes exactly that: a signal, not a command.
Example: Performance Max
You provide
Purchase goal, conversion data, product feed, audience signals, creative assets.
Google uses these as signals
Not instructions — inputs the system weighs alongside everything else it already knows.
Google predicts
Which users, searches, placements, and moments are most likely to convert.
The system acts
Bids, inventory, delivery, and audience selection — automatically.
| Mechanism | Search | Shopping | Demand Gen | Video | Performance Max |
|---|---|---|---|---|---|
| Primary signal | Query | Product data + query | Audience / context | Audience / context | Goal + signals |
| User intent | High | High | Lower | Variable | Variable |
| Main advertiser input | Keywords + assets | Product feed | Audience + creative | Audience + creative | Goals + signals + assets + feed |
| Auction automation | High | High | High | High | Very high |
| Creative importance | High | Product-dependent | Very high | Very high | High |
| Data dependency | Medium | Very high | High | High | Very high |
| Advertiser control | High — direct | Moderate — via feed | Moderate-to-lower | Moderate-to-lower | Low — steered via inputs |
Campaign type 01 / 05
Search
Ad Rank isn't just "bid × Quality Score." That's the textbook version.
01 — The Auction
The formula is a simplification. The auction weighs more than two numbers.
The simplified formula is useful for learning the concept, but the actual auction weighs more than two numbers. We manage every input we can, not just the bid.
Your bid
What you're willing to pay — the one input you set directly, and the least interesting lever once Smart Bidding is running.
Auction-time quality
Expected CTR, ad relevance, and landing page experience — assessed fresh at each individual auction, not as a fixed account-level score.
Ad Rank thresholds & auction competitiveness
How many other advertisers are bidding on that exact query, at that moment, sets a variable bar — the same bid can win easily in one auction and lose in the next.
Search context
Device, location, time of day, the exact search terms, and even other ads and results on that page — all factored in per auction.
Expected impact of assets & formats
Sitelinks, callouts, structured snippets, and image or video assets widen your ad's footprint and raise its expected performance — Google factors that in before you outbid anyone.
02 — Ad Quality
Ad relevance is scored algorithmically now — message match still wins.
Expected CTR and ad relevance used to be closer to keyword-matching exercises. They're increasingly assessed by Google's own models reading intent and content quality — which shifts the work from "insert keyword into headline" to genuinely relevant assets.
Full asset coverage
Headlines, descriptions, sitelinks, callouts, images, and video assets all feed the relevance and expected-CTR signals — a thin ad with three headlines is competing against a fully-built one before the bid is even compared.
Landing page experience
Relevance to what the ad promised, mobile load speed, and how easily a visitor can actually complete the action — the one Quality Score component that's entirely on your side of the auction, no bidding involved.
Consistency over tricks
Exact-match keyword stuffing moves the needle less than it used to. Consistent message match across ad copy, extensions, and landing page moves it more.
03 — Bidding & Budget
Broad match + Smart Bidding has been the default for two years. The job changed, not the goal.
The algorithm sets the bid at every individual auction in real time. What we actually manage is the target, the signal quality feeding it, and the budget architecture around it.
Signal quality over match-type tinkering
With broad match reading intent rather than keyword strings, conversion tracking accuracy and audience signals matter more than which match type a keyword uses.
Target CPA / Target ROAS, set and re-set
These are targets we set for the algorithm to hit — and they need revisiting as account data, seasonality, and Google's own bid-target recalibrations shift what's realistic.
Budget pacing
Spend that exhausts by noon is missing the rest of the day's high-intent traffic. Spend that never reaches its cap is capital sitting idle. We watch pacing daily, not monthly.
Rank-loss vs. budget-loss diagnosis
Two different problems with two different fixes, both showing up as "we're not getting enough impressions" if you don't separate them.
04 — AI Max
The layer Search is moving toward — not a new campaign type, an automation layer inside the one you already run.
AI Max expands search-term matching using broad match and keywordless technology, optimizes creative per searcher, and routes clicks to whichever page on your site is actually most relevant via final URL expansion. You keep control over required disclaimers and can pin specific text so it always shows.
Search-term matching
Broad match and keywordless technology reach queries you never typed as keywords — expanding reach beyond what a keyword list alone would capture.
Creative & landing page optimization
Creative is optimized per searcher, and final URL expansion routes the click to the most relevant page on your site — not always the one you'd have picked manually.
What you still control
Required disclaimers always display, and specific text can be pinned so it shows regardless of what the system generates — automation with guardrails, not a black box.
Campaign type 02 / 05
Shopping Campaign Mechanics
The product is the keyword. There's no list to write — Google reads it off your feed.
The Mechanism
In Search, you pick the queries. In Shopping, Google picks them for you.
In Search, you tell Google which queries you want to compete for. In Shopping, you describe the products you sell, and Google decides which queries those products are relevant to. That reversal is the whole mechanism.
Your feed is your product's information layer
| Feed field | What it tells Google |
|---|---|
| Title | What is it? |
| Description | What does it do, and why is it relevant? |
| Product type / category | Where does it belong? |
| Brand / GTIN | Which specific product, exactly? |
| Price | What does it cost? |
| Availability | Can the customer actually buy it right now? |
| Image | What does the customer see before they click? |
The Levers
Where you actually get control.
The bid still matters — but it operates inside an eligibility-and-relevance system it can't override.
Product groups
Organize by brand, product type, category, item ID, or custom labels — the same bid control keywords give you in Search, so 10,000 products don't all compete at one flat bid.
Price competitiveness
Matters more here than almost anywhere else, since Google shows your price right inside the ad — next to whichever competitor undercuts you.
Query mix
Branded vs. generic search demand shifts constantly, and changes which products actually earn impressions week to week.
The bid, in context
Raising it doesn't help a product that isn't eligible or relevant in the first place — the auction only sees products that already cleared those gates.
The Diagnosis
Don't reach for the bid slider first.
Work through it in this order — each scenario points to a different fix.
- Are the products you expect to advertise actually eligible? Check Merchant Center before anything else.
- Did the feed stop updating, or did availability or attributes change upstream?
- Did campaign settings change, or did demand and query volume simply shift?
- What is the shopper actually seeing? Product image, title, price, and brand all compete in the same results.
- Is the offer competitive against what's shown right next to it?
- Shopping sells the product before it sells the click — a weak proposition won't be fixed by a higher bid.
- Move downstream: product page, price match, availability, shipping terms, checkout friction, conversion tracking.
- The auction can deliver a perfectly relevant shopper. It cannot fix a broken ecommerce experience once they arrive.
Shopping performance isn't created inside Google Ads alone — it's the combined result of feed data, eligibility, relevance, the auction, and what happens after the click. Most "Shopping bidding problems" are feed problems wearing a disguise.
Campaign type 03 / 05
Demand Generation Mechanics
Search captures demand. Demand Gen has to create it.
The Mechanism
There's no query here — the system has to decide the opportunity exists at all.
Google began migrating standard Display campaigns into Demand Gen in June 2026 — new standard Display campaigns can no longer be created, and existing ones migrate automatically and permanently. Migrated campaigns gain YouTube, Discover, Gmail, and Maps inventory alongside the traditional Display Network, and lose manual CPC bidding, viewable-impression optimization, and brand lift measurement. What follows describes the system you're moving into.
The Levers
Signals, not just targeting.
Audience signals
Your data segments, custom segments, and in-market/affinity audiences are a starting point for Google's own expansion — a hint to the system, not a hard fence around who sees your ad.
Creative
Carries more weight than in Search: image and video combinations, Shorts, and visual storytelling decide placement quality as much as your bid does.
Available surfaces
YouTube, Discover, Gmail, and the Display Network each behave differently — the same creative doesn't necessarily work across all of them.
The Diagnosis
Reach is up, conversions are down — what to check.
- Audience expansion — Google broadened beyond your original signal further than expected.
- Inventory expansion — new placements (Discover, Gmail, Maps) diluting where your budget lands.
- Weaker placements or creative fatigue on assets that were fine a month ago.
- Conversion signals too thin for the system to optimize toward, or a bidding objective mismatched to what you actually want.
A Search campaign asks: did someone searching for this convert? Demand Gen asks: did this audience become interested enough to engage, consider, and eventually convert? Judging both by the same number is the mistake.
Campaign type 04 / 05
Video
A view is not intent. A click is not attention.
The Mechanism
The mechanism doesn't change across objectives — what Google optimizes toward does.
Video campaigns range from pure reach and awareness objectives to view-driven and conversion-oriented configurations. Same mechanism underneath; different stage of it being optimized.
The Levers
Three control layers — creative is doing more work than you think.
Layer 1 — Audience
Who the system is showing the creative to in the first place.
Layer 2 — Creative
The hook in the first few seconds decides whether the rest of the mechanism gets a chance to work at all; retention and CTA placement matter as much as the message itself.
Layer 3 — Bid & goal
What action you're actually asking the system to optimize toward — awareness, views, or conversions pull the same creative in different directions.
The Diagnosis
Views are up, conversions are flat — what to check.
- Objective or bid strategy mismatched to the actual goal — awareness bidding won't optimize for conversions.
- Weak CTA or a landing page that doesn't follow through on the video's promise.
- Conversion window too short for a consideration-length purchase decision.
- Attribution model over-crediting view-throughs that weren't really driving the outcome.
Video optimizes for attention first and action second. If you're only measuring the second, you're missing why the first one failed.
Campaign type 05 / 05
Performance Max
If Google runs the auction, the placement, and the creative mix — what's actually yours to control?
The Mechanism
Search and Shopping run one auction and stop. PMax doesn't.
It continuously re-weights every input against the last conversion signal it received. The system you're actually managing isn't the auction — it's the loop.
The Levers
The six inputs that are actually yours.
What you control
- Goals & conversion actions
- Conversion values
- Budget
- Bidding target
- Creative assets
- Audience signals
- Search themes
- Product data (feed)
- Landing-page inputs
- Brand exclusions, where applicable
What Google controls
- Auction-time bids
- Placement selection
- Audience expansion
- Channel allocation
- Creative combination testing
- Ongoing optimization decisions
Campaign goal
What the system is optimizing for — not a filter, a target.
Bid strategy
How the system values one opportunity against another.
Budget
A daily ceiling on average — not a fixed amount spent evenly every day.
Audience signals
A starting point Google can expand beyond — not a strict targeting rule.
Asset group
Raw text, image, and video material the system assembles — not a fixed, finished ad.
The Diagnosis
Performance drifted — what's actually diagnosable.
- Since Google controls execution, diagnosis narrows to your six inputs, not the auction itself.
- Is a conversion action mis-valued, pulling optimization toward the wrong outcome?
- Is one creative asset dragging down the whole combination mix?
- Is the feed (for e-commerce PMax) degraded the same way it would hurt a standalone Shopping campaign?
- Is the budget too tight for the system to gather enough signal to learn from?
PMax isn't a black box you feed and hope. It's a system you steer through six inputs — everything past that line is Google's auction to run, which is exactly why those six inputs deserve more attention than most advertisers give them.
The Framework
The Google Ads Decision System
Every section above answers the same seven questions, just with different inputs. That's the actual thesis of this page: we don't teach five campaign types. We teach one decision system that happens to run five different ways.
What does Google know?
The intent or context signal the system starts with.
What makes it eligible?
The gate before relevance or bid matter at all.
How does the auction decide?
Bid combined with quality, relevance, and context.
What does the user actually see?
The creative, format, and information at the moment of decision.
What happens after the click or view?
The downstream action that counts as a result.
What data goes back into the system?
The signal Google learns from.
How does that change the next auction?
The optimization step most advertisers never think about.
The same lens, five campaign types
Most Google Ads content teaches you where to click. This is where we teach how the system decides — and how to influence that decision, campaign by campaign.
Diagnosing Performance
CPC increased. That's a symptom, not a diagnosis — find the actual cause before you touch the bid.
The same surface-level number can mean three completely different problems underneath, each with a different fix.
Impressions ↓
- Higher competition — more advertisers bidding
- Query mix changed — different searches triggering
- Device or location segments narrowed
Conversions ↓
- Landing page experience dropped
- Offer or price changed
- Tracking issue broke silently
CPA ↑
- Auction got more competitive
- Quality Score dropped
- Seasonality or market shift
See it applied to your account