The Keyword Triangulation Cheat: How to Replace BMM in 2026

Google wants you to fish with a net. I'd rather fish with bait.

That's the difference between broad match and triangulation. Broad match gives Google permission to serve your ad against anything the algorithm thinks is "related." You sell custom dog tags. Someone searches for dog collars. Google decides that's close enough, serves your ad, you pay for the click, they don't buy a dog tag.

Triangulation is the correction. Three match variants of the same keyword — broad, phrase, exact — all running in one ad group. Google's auction system picks the match type that serves the query most efficiently, which usually means the lowest qualifying cost per click. Blended CPC drops. Wasted spend on loose broad-match variants drops. Your budget lasts longer and buys better traffic.

I've been running triangulation since 2006 or 2007. For most of that time, Google reps would try to talk clients out of it. Then in late 2024, during a strategy meeting on a high-spend account where performance was flat-lining, a Google high-spend team rep approved it — with one requirement: put all three match types in one ad group, not three. That shift is the modern version. It's cleaner to manage and the algorithm responds better.

This post is the complete operator view: what triangulation actually is, why it matters more after Google killed broad match modifiers, how to set it up, when it breaks, and what to watch for in the first 30 days.

The 2021 change that made this matter

In February 2021, Google retired broad match modifier — the "+keyword +variant" syntax that let advertisers force specific words to be present in queries without locking into phrase match. BMM was surgical. You could require "accounting" and "software" to appear together without dictating word order.

Google merged BMM's behavior into phrase match. On paper, that was supposed to preserve the advertiser's control. In practice, phrase match loosened over the following years and broad match became the default Google pushes at every opportunity — through account recommendations, through auto-apply, through campaign templates.

The net effect: advertisers lost a precise tool and gained a blunter one. Triangulation doesn't replace BMM exactly. What it does is force Google's auction logic to do the filtering that BMM used to do manually — by giving it multiple match candidates for the same query and letting it pick.

What triangulation actually does

When a search query comes in, Google's auction looks at every keyword in your account eligible to serve that query. It picks the one with the highest Ad Rank — a combination of bid, quality score, ad relevance, and context.

If you've triangulated a keyword — broad, phrase, and exact variants all active — and the user searches the exact query, the exact-match keyword typically wins the auction. Exact match keywords on their literal query have the highest quality score because the relevance is perfect. High quality score means the bid doesn't have to be as high to win the position. So you pay less.

If the query is a variant Google's system thinks is close but not exact, phrase match can win. If it's looser, broad wins. The point isn't which match type "wins" — the point is that when all three are eligible, Google is forced to serve the most precise match available, not just the broadest one it can justify.

That's the cheat. It's not a loophole and it's not a hack. It's using Google's own auction logic to do the tightening that broad match by itself doesn't do.

The old way vs the new way

The textbook approach, pre-2024, was to run three separate ad groups — one for broad, one for phrase, one for exact — each with the same keyword list. I ran it that way for close to twenty years.

The problem with three ad groups was impression share fragmentation. Google would allocate impressions across them based on its own opaque logic. Management overhead was real. You had three sets of ads to optimize, three bid strategies to maintain, three performance reports to reconcile.

One ad group is better. All three match types under one roof, one set of ads, one set of bids, one performance report. The algorithm handles the selection; you handle the structure.

The only time I'd recommend splitting back into separate ad groups is when you genuinely need different ad copy for different match types — for example, if your broad match traffic is qualitatively different from your exact match traffic and needs different messaging. That's rare. In most accounts, one ad group does the job.

When it works

Triangulation is vertical-agnostic. I've used it in B2B, ecommerce, medical, lead gen, and local service accounts. The structural benefit — forcing Google's auction to serve the most specific qualifying match — applies everywhere paid search runs.

But it's not the first move I make on a new account. I start with broad match in most cases, watch the search terms report for 2-4 weeks to understand what queries are actually flowing in, then add phrase and exact variants deliberately as the data reveals which queries are worth owning at a tighter match. Triangulation is a structural decision you make once you understand the search landscape, not a setup decision you make blind.

The exception is accounts with what I call a "stress budget" — very low budget, very high expectations. On those, I triangulate from day one because the account can't afford the broad-match learning period. It's a forcing function that cuts the initial waste.

When it breaks

Three scenarios where triangulation doesn't help or actively hurts:

Low impression volume campaigns. Scientific and highly specialized B2B verticals often run on keywords with genuinely low search volume — specific instrument names, niche compound classes, narrow professional terminology. In those accounts, there's no meaningful broad-match leakage to filter out, because the search universe is already small. Triangulation adds overhead without proportional benefit.

Campaigns already hitting Target CPA or Target CPL and performing cleanly. If the account is healthy and the automated bidding is converging, don't restructure the keywords. Let it run. Triangulation is a corrective tool; applied to a healthy account it's a solution looking for a problem.

Verticals where cross-category matches are genuinely valuable. If you sell iPhones and Android phones and a search for "best smartphone" could legitimately convert for either product line, broad match isn't a bug — it's feature. Don't triangulate yourself out of traffic you actually want.

The ~20 keyword ceiling

Google's own guidance — and mine — is to keep ad groups under 20 keywords for ad relevance reasons. With triangulation (three match types per keyword), that translates to roughly 6-7 root keywords per ad group as the default. That's the recommendation to start with.

I've seen accounts hold together at 100 keywords per ad group without ad relevance collapsing. Every vertical, account, and campaign is different. But "it sometimes works" is not a planning recommendation. Gambling is fun but not everyone wins. Plan around the 20 ceiling. If an account later proves it can hold more, push it — but never structure a new account expecting to live at 100.

The reason the ceiling matters is simple. A low-performing or low-quality-score keyword inside a triangulated ad group can drag down the quality score of everything else in the group. One bad keyword is a black hole. It affects ad relevance scoring for the whole group, which affects your CPCs across all the match types you've carefully structured. The tighter the group, the less exposure you have to that drag.

How to set it up

Six steps for implementing triangulation on an account:

1. Pick the root keywords. Start with your highest-intent, highest-relevance terms — the ones where you'd bid aggressively if you knew the search was qualified. 6-10 roots is the sweet spot for a first ad group. Don't include your tangential or exploratory keywords here; those live in separate ad groups where you can test more loosely.

2. Create all three match variants. For each root, add broad, phrase (in quotes), and exact (in brackets). If the root is marketing audit, you'll have three keywords in the ad group: marketing audit, "marketing audit", and [marketing audit].

3. Set bids manually. Start conservative. I don't turn on automated bidding on a new triangulation ad group. Manual CPC, start with a bid slightly below the suggested opening bid Google gives you, and watch it for a week. If the exact match isn't winning impressions, inch the bid up. If it is and the broader match types are burning budget, nudge them down slightly. You and Google are both performance-based — let the data train the bid, don't set and forget. Once the ad group has 90+ days of clean conversion data and the numbers are holding, you can experiment with automated bidding, but treat it like a defibrillator: don't use it on a healthy account, use it on one that's flat-lined.

4. Write many ads. Ideally 20 or more RSA variants. The more headline and description variations you feed the ad group, the more dynamically Google can assemble relevant ads per query. Triangulation pulls varied query intent into one ad group — the ad copy has to cover that range. This is the one place where more is better.

5. Pull search terms reports weekly for the first 30 days. Because broad is in the mix, you'll see query variants you wouldn't have seen with phrase and exact alone. Add negatives aggressively during this window. After the first month, the account settles and search term hygiene moves to monthly.

6. Watch quality score by match type. The tell that triangulation is working in your favor is that exact match's quality score holds at 7-10 and broad match's quality score stays in a defensible range (5+). If exact match's quality score starts dropping or broad match is under 5 and pulling costs up, that's your signal to dig in — usually one keyword is underperforming and needs to be paused before it drags the group.

What to watch in the first 30 days

The signal it's working: CPL or CPA starts trending down within 2-3 weeks. Blended CPC drops as more of your clicks get served at exact match's lower price point. Search terms report shows cleaner alignment with intent — fewer "what was this click doing here?" moments.

The signal it's not working: total cost per acquisition rises without lead volume rising proportionally. Quality scores on exact match start declining. One keyword in the group is absorbing disproportionate spend without converting.

If costs are rising but lead volume is also rising, that's often fine — the client just has to recalibrate their spend expectations. But if costs are rising and conversions are flat, the account is telling you something. Pause the worst-performing keyword (don't delete it), watch the next week, see if the ad group recovers.

If pausing doesn't fix it, copy those paused keywords into a separate campaign with tighter ad copy specifically written for them, and see how they perform in isolation. Sometimes the issue isn't the keyword — it's that the ad group's overall theme drifted, and the keyword was carrying relevance weight it couldn't support. In a tighter campaign built around its exact theme, it might work fine.

The senior care account: triangulation in context

I described this account in the pillar post on Google Recommendations: a 65+ primary care network that scaled from 33 to 136 clinics across five states under my paid media leadership, with monthly spend growing from $75K to $600K over 18 months.

Triangulation was one of the first structural changes I made when I took over. It wasn't the only change — we also killed Performance Max after $20K in testing showed a <0.5% lead-to-appointment rate, disabled auto-apply, restructured campaigns around individual clinic geography, and blocked every country and state we weren't operating in.

But triangulation was the foundation the rest of the structure sat on. By forcing Google's auction to serve the most precise match available for each query, the account stopped bleeding budget on loose variants and had room to invest in the campaign structure and geographic targeting that actually drove patient volume.

Over 18 months, cost per lead moved from $200-$250 to $100-$125. Cost per actual patient — the metric the business cared about — moved from ~$1,500 to under $300. Triangulation didn't produce those numbers by itself. It made those numbers possible.

The multi-platform version of the same problem

Triangulation is a Google-Ads-specific technique. It works because of how Google's auction logic evaluates competing match variants for the same query. It doesn't solve the bigger question: is Google the right place for your next marketing dollar at all, versus Meta, LinkedIn, Microsoft, or somewhere else?

That multi-platform budget question is structurally invisible to any single-platform optimization. It's the problem I started Campaign Budget Optimizer to solve — an AI-native cross-platform allocation tool that connects to Google Ads, Meta, Microsoft Ads, LinkedIn, TikTok, and Google Analytics to surface where the next dollar should actually go. It launches in May 2026.

For the manual version of the same thinking, see Where should the next dollar go?.

If you take one thing from this post

Triangulation is not a universal fix. It's a structural correction for the specific problem of broad match spending your budget on queries that don't convert. On a healthy, mature account running target CPA or target ROAS cleanly, don't touch it. On an account that's flat-lining, bleeding budget, or can't hold an efficient CPL, triangulation is probably the first structural change to make before you touch bidding, copy, or audience targeting.

Start with 6-10 root keywords. Three match types each. Manual bids, held conservative. 20+ RSAs. Weekly search terms review for the first 30 days. Watch quality score by match type. Pause underperformers before they drag the group.

Google wants you to fish with a net. The net catches more fish, most of which you can't sell. Bait catches the fish you actually want.

---

If your Google Ads account is flat-lining and you want a senior operator to look at it, SEM consulting and marketing audits are both services I run. Engagement pricing is on the pricing page. First conversation is always free.