Most accounts treat match type as a single lever and let one campaign mix exact, phrase and broad. That hands the algorithm the steering wheel and blurs your data. RGM does the opposite: we control which queries we pay for, and we expand deliberately.
The model is a grid. Every keyword gets a search-intent tier and a match type, and each cell runs as its own campaign with its own budget and bids. Exact leads, phrase supports, and broad runs last as a small prospecting engine that feeds the other two.
Two questions decide how a keyword is treated. How much intent does the query carry? And which match type are we using to reach it? Cross those two axes and you get a grid that drives budget, bids and structure for the whole account.
The principle is control with deliberate expansion. You bid hardest where intent is highest and the match is tightest, and you use looser matches as a small, fenced way to discover new winners — not as the main event. Done well, you always know which queries you are paying for, your data stays clean, and you still grow.
Before you pick a match type, classify the demand. Intent is how close the searcher is to the action you want. We sort keywords into four tiers and launch each tier separately so its budget and bids reflect how valuable that demand is.
| Intent tier | What the query looks like | Example (a plumber) | Budget & bid stance |
|---|---|---|---|
| Very high | Ready to buy or book now; explicit, commercial, often local or branded | “emergency plumber near me” | Full saturation — bid to win, never capped by budget |
| High | Clear commercial intent, comparing or shortlisting | “water heater replacement cost” | Strong budget, competitive bids |
| Medium | Researching a problem; intent forming | “why is my water heater leaking” | Moderate budget, efficiency-first bids |
| Low | Top-of-funnel, informational, weak commercial signal | “how does a water heater work” | Small budget — learn only, or skip |
Tiering is a judgement call you make per keyword, informed by the words themselves, the landing experience and — once you have data — conversion rate. The point is that a very-high-intent term and a low-intent term should never compete for the same budget. Separate campaigns let the high-intent demand take all the budget it can convert.
To grow, brands must build mental and physical availability — be noticed by, and easy to buy for, more people in more situations.
— Byron Sharp, How Brands Grow (2010)
Google’s match types in 2026 all expand through close variants, so “exact” is not literally exact and phrase reaches further than it used to. That makes control the scarce resource. We treat the three match types as a tier system from tight to loose.
| Match type | How it behaves in 2026 | RGM tier | Role |
|---|---|---|---|
Exact [ ] | Your term plus very close variants and same-meaning queries | Tier 1 | Own the query. Highest bids and budget. |
Phrase " " | Queries that include your term’s meaning in order | Tier 2 | Controlled expansion. Mine it, promote winners to exact. |
| Broad | Any query Google judges related, steered by signals and bidding | Tier 3 | Learn & expand. Small, fenced prospecting only. |
Each match type runs in its own campaign. That is the heart of the method. When exact, phrase and broad share a campaign, they share a budget and a bid strategy, and you lose the ability to fund the tight, high-intent traffic differently from the loose, exploratory traffic. Splitting them out is what gives you control over exactly where the money goes.
Now combine the two axes. Budget and bid intensity is highest where high intent meets a tight match, and lowest where low intent meets broad. The interactive grid above is the reference: very-high-intent exact gets full saturation; low-intent broad runs only to learn.
In practice you will not build all twelve cells. You build the cells that matter for your account — usually exact and phrase across your top two intent tiers, plus one small broad prospecting campaign — and you let the matrix tell you how to bid and budget each one.
Build a campaign for each intent-tier × match-type cell you actually want to fund differently. A common, manageable starting structure for a focused account is five campaigns: very-high exact, very-high phrase, high exact, high phrase, and one broad prospecting campaign covering your priority themes.
Resist the urge to build all twelve. Over-fragmenting splinters conversion data and makes Smart Bidding learn slowly. The test for a new campaign is simple: do I want to fund or bid this traffic differently from everything else? If yes, split it; if no, keep it together.
Name campaigns so the cell is obvious at a glance — for example Search | NonBrand | Exact | T1 | tROAS. The fixed order (channel, brand split, match type, intent tier, bid strategy) keeps the whole account filterable.
Run your own seed list through the tool below to see the tiering applied — it sorts each term by intent, assigns the match type and campaign tier, and flags negatives.
We do not avoid broad — we fence it. Broad is the only match type that finds queries you did not think to target, which makes it a genuine discovery tool. The discipline is in how you run it.
Broad only behaves when Smart Bidding has real conversion values to steer by, a deep negative list to fence it in, and tracking the model can trust. Miss any one and broad is just spend leakage.
So we quarantine broad in its own campaign on its own small budget — never mixed with exact — and harvest its winning search terms up into exact and phrase every week.
The tiers are not static buckets — they feed each other. Broad and phrase are search-term factories; exact is where you cash in. The loop runs every week:
That last step is what keeps the system clean: a query you have promoted should be paid for once, from the campaign you intended, at the bid you chose.
With several campaigns chasing related demand, negatives are not just for junk — they are how you route each query to the campaign you intended. The rule of thumb: a query should be eligible in exactly one of your campaigns.
Junk, competitor, and irrelevant terms get added once to master negative lists applied account-wide, so a lesson learned in one campaign protects all of them.
And we mine search terms by spend-at-risk, not on a flat schedule — the campaigns burning the most budget get checked daily; the long tail waits for the monthly pass.
Bids and budgets follow the matrix, not the keyword count. Exact in your top intent tier gets the highest target and the most budget because that is the demand most likely to convert profitably. Phrase sits a notch lower; broad lowest of all.
Run a bid strategy per campaign so each tier optimises on its own merit — high-intent exact can chase efficiency hard, while the broad campaign optimises gently for discovery. Tier-1 intent should rarely be budget-capped; if a very-high-intent campaign is limited by budget, you are leaving profitable demand on the table and should move money from lower tiers.
Keep brand and non-brand in separate campaigns, always. Brand converts cheaply because you already earned the demand; letting it sit inside a prospecting campaign flatters the numbers and misdirects budget. A small, protected brand-exact campaign defends your terms; everything else is non-brand, tiered by intent and match as above.
Brand defence and conquesting both start with knowing who you’re up against. Drop your Auction Insights numbers into the tool below to see which rivals actually threaten you, the move for each, and where the messaging white space is.
Brand terms convert cheaply and make a blended account look brilliant. Leave them mixed with non-brand and you can’t see whether your real prospecting is working or quietly bleeding.
We isolate brand, judge it on its own, and hold non-brand to its honest number — the one that tells us if we’re actually growing.
Order matters, especially on a new account or a tight budget:
On a small budget, you may never reach step three until the core is profitable. That is fine. Broad earns its place once you can afford to learn; it is not the foundation.
The recurring theme: people hand control to the algorithm and then wonder why spend leaks. Lead with broad, mix match types in one campaign, skip the routing negatives, ignore intent — each one trades control for convenience, and the bill arrives at the next budget review.
Most accounts arrive as one of two extremes: a broad-and-Smart-Bidding blob, or a pile of exact-only SKAGs. Both migrate the same way — carefully, without losing conversion history.
Pass at 90% to unlock the completion passcode for this module. You can retake it as many times as you like.