The Helium 10 stack for 7-figure brands, Cerebro, Magnet, and Adtomic in a real workflow

Most operators run Helium 10 like a $97 keyword tool. They pull a Cerebro report, eyeball the top 20 phrases, and call it research. Then they wonder why their Adtomic campaigns plateau at $1.40 ACoS targets and 18% TACoS. The Diamond plan is $279/mo as of April 2026. You should expect more from it than a vanity keyword list.

This is the workflow we run for our 7-figure brands at ClearSight. Three tools, Cerebro, Magnet, Adtomic, chained in a specific order, with specific decisions at each handoff. No fluff, no “best practices,” just the sequence that produces ranked SKUs and a sub-25% TACoS.

Step 1: Cerebro pull, but only on the right ASINs

The first mistake is reverse-ASIN-ing your own listing. You already rank for those terms. The point is to find what your competitors rank for that you don’t.

Pull Cerebro on 8-10 ASINs: the top 3 organic competitors, the top 3 sponsored competitors, and 2-3 adjacent-category SKUs that share a buyer. Filter to Organic Rank ≤ 30, Search Volume ≥ 500, and Competing Products ≥ 200. That’s your serious-intent universe.

Now apply the rule we wish we’d known three years ago: drop any phrase where your competitor’s organic rank is better than position 10 AND their sponsored rank is also top 10. They’ve already paid to own it. You’re not outbidding them in month one.

What’s left is the gap. Phrases where competitors rank organically but haven’t locked them down with paid. That’s your seed list. Expect 80-150 phrases from a clean Cerebro pull on a mid-competition category.

Step 2: Magnet refinement, the long-tail hunt

Cerebro tells you what competitors rank for. Magnet tells you what real shoppers type. The two answers diverge more than people admit.

Take your 5 highest-volume Cerebro phrases. Run each through Magnet. Set Magnet IQ Score ≥ 40, Search Volume between 200 and 4,000, and Title Density ≤ 30. Title Density is the underused filter, it tells you how many of the top 10 organic results have the phrase in their title. Below 30 means the SERP isn’t optimized for that exact string. That’s your wedge.

Magnet will surface 4-7 word long-tails that Cerebro misses because the head term has too much volume noise. “Stainless steel insulated water bottle 32 oz” beats “water bottle” every day for a 7-figure brand that already owns a niche.

Pull these into a sheet. Tag each phrase with three columns: head/mid/long, branded/unbranded, and intent (research/comparison/transactional). You’ll need those tags in Adtomic.

Step 3: Adtomic seed list, the rules that actually matter

Adtomic is bundled in the Diamond plan. Most operators treat it as a glorified bulk file editor. It’s not. It’s a rules engine with bid automation that respects 14-day attribution. Use it that way.

Build three campaign types from your tagged sheet:

• Exact-match harvest campaigns seeded with your transactional long-tails from Magnet. Starting bid: 60% of the suggested CPC floor.
• Phrase-match research campaigns seeded with mid-tail Cerebro gaps. Starting bid: suggested CPC floor.
• Auto campaigns with negative-keyword harvesting seeded against your top 5 ASINs. These exist to find search terms you didn’t think of.

Now the rules. Adtomic lets you set bid-adjustment rules at the keyword level. Most operators set one global rule and forget it. That’s why their accounts plateau. Set tiered rules:

• If 14-day ACoS > 1.5x target AND clicks ≥ 10: reduce bid 15%.
• If 14-day ACoS < 0.5x target AND impressions < 1,000: raise bid 10%.
• If 60-day spend > $50 AND zero conversions: pause keyword, send to negatives.

The 60-day spend gate matters more than the others. It’s the one that protects you from “almost converting” delusion.

Bid-floor decisions: where 7-figure operators leave money

Helium 10’s suggested CPC is a regression on Amazon’s own data. It’s not wrong, it’s just averaged. Your account isn’t average.

For each ad group, calculate your real bid floor: (target ACoS × average order value) ÷ projected conversion rate. If your AOV is $42, your target ACoS is 22%, and your conversion rate on this term is 9%, your floor is $1.03. Adtomic’s suggestion might say $1.45. Bid your floor, not theirs.

For high-intent transactional long-tails, you can usually run 20-30% above the floor and still hit target ACoS because conversion rate is so much higher than the category average. For research-intent head terms, run at the floor or below. You’re paying for first-page impressions, not clicks.

The bid-floor formula extends. Apply a placement multiplier on top of the base floor: top-of-search converts 1.4-1.8x rest-of-search in most categories we audit. So your top-of-search floor is base floor × 0.7, because Amazon will already mark it up via placement bidding. Bid the discounted floor and let placement adjustments do the work. Operators who bid the suggested CPC and then add a 50% top-of-search modifier are double-paying.

This is also where Adtomic’s dayparting matters. If your category converts at 2x rate between 6pm and 10pm, the floor is different in that window. Most operators ignore this. Helium 10’s Diamond tier supports it. Use it.

Dayparting that actually moves the number

Generic dayparting advice is “bid up at night.” That’s wrong half the time. The right framework is conversion-rate-weighted bid scaling, hour by hour, against your category’s buyer.

Pull 90 days of Sponsored Products hourly conversion data. Group by hour. Calculate the conversion rate index for each hour: hour CVR ÷ 24-hour average CVR. Any hour with an index above 1.15 is a bid-up window. Any hour below 0.80 is a bid-down window. The middle 70% of hours stay flat.

For most CPG categories the bid-up window is 7pm-11pm Eastern. For B2B-leaning categories it’s 9am-noon weekdays. For kids’ products it’s 9pm-midnight. The point is to measure, not assume. Adtomic’s scheduler accepts hour-level multipliers; set them and walk away.

One detail nobody talks about: weekend dayparts are different from weekday dayparts in 80% of categories. Build two schedules. The lift on a single SKU portfolio is usually 6-9% lower TACoS within a quarter, just from this.

Search-term harvesting cadence

Auto campaigns and broad/phrase research campaigns generate search terms. Most teams harvest them quarterly, which is too slow, or weekly, which is too noisy. The right cadence is every 14 days, aligned with Amazon’s attribution window.

On day 14, pull the Search Term Report for every research campaign. Filter to terms with ≥ 7 clicks. Branch into three buckets: (a) terms with ≥ 1 conversion at sub-target ACoS, promote to exact-match harvest campaign at the calculated floor. (b) terms with ≥ 7 clicks and zero conversions, push to a campaign-level negative. (c) terms in between, let them keep running another 14 days, then re-evaluate.

The trap operators fall into is harvesting only the winners. The negatives matter just as much. A research campaign without aggressive negatives becomes a slush fund for irrelevant traffic. Half your harvesting cadence should be subtraction, not addition.

What we strip out from the H10 default workflow

Helium 10’s onboarding pushes operators toward Frankenstein, Scribbles, and the Listing Optimizer as part of the keyword-to-listing pipeline. For 7-figure brands, we strip most of that out.

Frankenstein is fine for a first pass on 1,000 keywords. We don’t use it after that, it dedupes too aggressively and loses singular/plural variants that Amazon treats as different terms. Scribbles is a checklist tool that adds friction without adding accuracy. The Listing Optimizer’s scoring rubric optimizes for keyword density, which Amazon stopped rewarding around 2022. Density above 2-3% is now a downgrade signal.

We also skip the Index Checker for established SKUs. Index Checker is useful for a launch, not a maintenance audit. If your SKU has been live 6+ months and converting, it’s indexed. The 10 minutes spent running Index Checker is better spent on a Cerebro reverse-ASIN of your top organic ranker.

What we keep: Cerebro, Magnet, Adtomic, Profits (with caveats), and Keyword Tracker. What we drop or deprioritize: Frankenstein at scale, Scribbles, Listing Optimizer scoring, Index Checker for mature SKUs, and the “Trending Products” feed. The Diamond plan still pays for itself even if you only use 5 of the 12 tools, because those 5 are the load-bearing ones.

Why the order of operations matters

Operators who run Adtomic without first cleaning their Cerebro and Magnet output produce campaigns that look fine in week one and rot by week six. They’re optimizing inside a polluted keyword set.

The Cerebro filter strips bad-faith competitor copies. The Magnet pass strips low-intent volume that converts at 1%. By the time keywords land in Adtomic, you’ve already removed 80% of the spend leak. The bid rules then handle the remaining drift.

This is what the stack is for. Each tool removes a class of error before the next tool runs. Skipping a step doesn’t save time, it just defers the loss into your ad spend.

One adjacent point: most brands reading this also have an A+ content problem feeding the same ad spend. We covered the Cosmo A+ rebuild separately, and it’s a prerequisite for the Adtomic numbers to actually compound. If your detail page converts at 6%, no bid rule saves you. Ad-driven traffic also runs into how Rufus is reranking results, which changes what your ranked terms are even worth. Both posts pair with this one.

What this looks like at month three

Run the stack on a real catalog and the pattern is consistent. Week 1-2: campaign structure rebuilt, 30-50% of legacy keywords paused. Week 3-6: Adtomic rules trigger their first major bid adjustments. Week 7-12: TACoS drops 4-7 points while units sold hold or grow.

That’s not a promise. It’s what the math says when you stop using a $279/mo tool like a $29 one.

What this looks like at $20M GMV vs $5M GMV

The same stack scales differently at different revenue tiers. The mistake is running the $20M playbook at $5M, or vice versa.

At $5M GMV, you’re typically running 30-80 active SKUs, $40-120K/mo in ad spend, and one or two analysts. Adtomic’s rule engine carries the weight. You’re mostly in Sponsored Products with limited Sponsored Brands and Sponsored Display. The Cerebro-to-Magnet-to-Adtomic loop runs every 14 days on a tight set of priority SKUs. Search-term harvest is manual but bounded. Total tool stack outside H10 is small: Sellerboard for true margin, a single competitive intel layer, no DSP. Diamond at $279/mo is the right tier.

At $20M GMV, the picture is different. You’re running 200-600 SKUs, $400K-1.2M/mo in ad spend, a 3-5 person ad team, and you’re into Sponsored Display retargeting and DSP. Adtomic alone stops being enough, not because it breaks, but because the rule complexity outgrows what one tool can hold. We typically pair Adtomic with a layer like Pacvue or Quartile for DSP and AMC, and let Adtomic own Sponsored Products only. Cerebro and Magnet still run, but the cadence shifts: priority SKUs every 14 days, mid-tier SKUs monthly, long-tail SKUs quarterly. You start needing Helium 10 Elite at $1,597/mo for the team seats and training, not because Diamond is missing features, but because the org around it has grown.

The other shift at $20M is governance. At $5M, one operator can hold the entire bid-rule logic in their head. At $20M, you need a written bid-rule policy, a quarterly audit, and a change log. The tool doesn’t change. The operating discipline around it does.

If you’re running Helium 10 and your TACoS hasn’t moved in 6 months, the problem is almost never the tool. It’s the workflow. Get a free audit and we’ll show you where the leak is in your specific account.

---