Extension Ecom LLC

09/04/2026

This vCPM campaign showed a $2.06 ROAS, however, it's actual ROAS from clicks was $.26. 🤯

If the first rule of Amazon is to never run out-of-stock, the second rule of Amazon should be to never launch vCPM.

vCPM = viewable Cost Per Mille (Mille means 1,000)

The attribution model is that you pay for every 1,000 viewable impressions, rather than per click.

It's an attribution model Amazon Ads launched a few years back and when people launch vCPM, and target their own detail pages it gets insane ROAS.

When vCPM first launched I worked with a dog bed brand doing $50M/YR on Amazon spending $250k/mo+ on Amazon Ads.

Their SD ROAS went from $4 to $40 overnight, the only thing that changed was swapping from CPC to vCPM.

Massive win, right?

Well, their sales didn't change.

When you target your own ASINs with a viewable attribution model it cannibalizes nearly every sale that occurs.

Why?

Because it's highly improbable that you'll buy a product on Amazon without viewing the product.

And since everyone's viewing the product, every single sale is now attributed to this vCPM ad.

Most Sellers don't understand this, they just see the beautiful ROAS it produces (relative to the rest of the account), and they never want it turned off.

But they should turn it off.

They should turn it off right away.

And focus on TACOS - because it won't drop.

Because the ads actual ROAS in this case is $.26.

And we all know if we saw a $.26 ROAS campaign that spent $1.2k we'd pause it immediately.

TLDR: never launch vCPM, you're wasting money.

09/04/2026

A returns rate that goes from 5% to 22% in two weeks isn’t random.

Something changed.

For a motorcycle cover brand, the culprit turned out to be one line of text sitting inside a size chart graphic: “size up for a looser fit.”

Shoppers weren’t reading it as optional guidance.

They were reading it as an instruction — landing on the “large” size chart, seeing that line, and buying the extra large instead of the size that actually fit their bike.

The return reason data confirmed it: most returns clustered on the extra large, with customers saying it was too big.

Not too small. Too big.

That’s the signature of people sizing up when they didn’t need to.

The fix wasn’t a redesign.

It was removing one sentence from two images — the main image and the A+ chart — and replacing it with a stationary-use disclaimer that solved a second, unrelated return reason (people trailering products not built for it).

Text sitting inside an image doesn’t feel like copy people scrutinize, but it gets read as an instruction more literally than body copy ever does.

If returns spike right after an image update, check the words in the image before you check anything else.

09/03/2026

"The communication is a million times better here. I feel like there's always someone on top of the account, always doing something, always trying."

That's how another one of our partners described working with us...

A few things he shared:

🔹"I really didn't have the skill set I thought I had — especially with advertising. I was just throwing money into ads and hoping they'd do well, without really knowing. That's something Extension Ecom has taken a hold of, and it's something I don't worry about anymore."

🔹"It's probably exceeded my expectations. I've worked with a couple of agencies before, and in the past they'd promise a lot and nothing would really happen. That has not been the case here."

🔹"It's been a steady win. Comparing to last year's numbers, other than the first month when we were still onboarding, the account has been better every month, year over year."

Check it out here: https://www.youtube.com/watch?v=k7ULAjWLEBg

This video highlights David's experience working with the Extension...

09/03/2026

We reordered a client’s image carousel based on one question: what do shoppers actually hesitate on before buying?

For this grill product, the two biggest objections were always the same.

How long does it burn? And is this actually clean-burning, or am I still cooking over chemicals?

There was an image sitting near the back of the carousel — position 7 or 8 — showing burn time and the hardwood charcoal callout.

Buried where most shoppers never scroll.

We moved it to position 4.

The logic: Amazon’s carousel gets clicked through less the further back you go, and some platforms even cut off the last couple of images entirely. If an image answers your top objection, it doesn’t belong near the end.

Lifestyle images went last instead — they carry the least information density, so they’re the ones you can afford to lose to scroll fatigue.

Before you finalize carousel order, ask what your last two customer service emails or reviews were actually confused about.

That’s usually your real image 4.

09/02/2026

We just showed a client they could run PPC at 1x ROAS and still be HIGHLY profitable. 🤑

After auditing a protein bar brand doing $51K/month in sales, we discovered they were in the 98th percentile for customer retention in their category. However, they aren't leveraging non-branded advertising due to the low ROAS.

Let's talk about why they should spend ALL the money at a $1 ROAS.

Their LTV numbers (over a 12 month period):
- Top 10% customers: $373.33
- All other customers: $68.73

Knowing this data we can identify the lifetime ROAS they'll receive on non-branded sales.

Let's say we spent $8,540.26 on non-branded traffic at a $1 ROAS, that would result in $8,540.26 in ad sales.

We could then take that, and divide it by their AOV ($32.04) to identify the quantity of orders they'd receive on that spend. Next, we can take those orders and break it up into 10%, and 90% buckets and apply the LTV values.

Their top 10% of customers will spend $9,951.11 on them in a year, and the remaining 90% will spend $16,487.98.

This puts their lifetime ROAS at $3.09 ($26,439 lifetime sales / $8,540.26 ad spend).

Understanding this ^ is how CPG companies with a GREAT product blitz scale.

09/02/2026

When we took over this account, 53% of ad spend was going toward defending branded keywords the brand already owned.

Only 47% was going toward offensive, non-branded targeting — the spend that actually finds new customers.

The brand already converted well on its own name.

It didn’t need that much protection.

It needed more dollars finding people who’d never heard of it yet.

Three months in, that ratio has moved to roughly 59% offensive / 41% defensive.

The target: 90% offensive, 10% defensive.

That doesn’t mean abandoning branded terms — 10% is still real coverage.

It means recognizing that once your branded conversion rate is already strong, every additional dollar defending it is a dollar not spent finding your next customer.

If you don’t know your offensive-to-defensive spend ratio right now, that’s worth pulling before your next budget conversation.

09/01/2026

A prospect we audited last week had a 12-month LTV of $576.99 on top Shoppers & Subscribe & Save for a product they've been spending years developing.

The category top had $536.14.

I've done hundreds of audits.

It's EXTREMELY RARE to see an LTV above the category top.

This prospect is sitting on a winning lottery ticket...

But they only have 56 shoppers that are top 10%/SnS, while the category top has 18,394.

They have an 8.32% TACOS.

29% of that spend is on branded search.

50% of it is on vCPM targeting themselves.

The ROAS looks amazing.

But they aren't growing.

They're underspending (8.32% TACOS for a supplement brand is far too low).

Non-branded KWs result in a $.80 - $1.20 ROAS, which they thought was way too low (comparing to their branded/vCPM).

I reframed it for them - it it worth it spend $40 to make $576+?

That's a $14.4 LTV:CAC.

If they get to 18,500 shoppers spending $570 on them (like the category top) that's $10.5M annually in EXTREMELY profitable sales.

We agreed - it's time cut vCPM, pull back on branded search, and spend on non-branded.

It's time to cash that winning lottery ticket in.

09/01/2026

If your keywords are underperforming, don’t just look at ACOS.

Look at where they break in the funnel.

Pull your SQP report and track three numbers per keyword: brand impression share, click share, purchase share.

Watch where the drop happens:
• Impressions are fine but clicks aren’t coming → CTR problem, likely image or price
• Clicks are coming but purchases aren’t → CVR problem, likely listing or reviews
• Both are healthy but both are shrinking → look at search volume and CPC, not your listing

One account we audited had a keyword behave completely differently across two quarters.

Early in the year it was a converting keyword — solid purchase share relative to clicks.

By midyear, it had quietly shifted into a pure consideration keyword: impressions fine, clicks fine, purchase share gone.

Same keyword.

Same spend logic would’ve applied.

Completely different problem to solve.

Don’t fix a CTR problem with a CVR solution, or vice versa.

Break the keyword down by funnel stage before you touch the bid.

08/31/2026

PPC specialists consistently fail this technical interview question, yet it costs Amazon sellers that employ them THOUSANDS each month.

The question: "When should you pause, or negate a keyword?"

Their answer: "10 clicks without a sale."

My follow-up: "Why 10?" :thinking_face:

Their response: Silence.

Here's the costly truth they don't know:

If your product costs $1,000 with a 2% conversion rate, you need 50 clicks to even get a sale. Realistically, you should give it 75 clicks before making a decision.

Pausing at 10 clicks means you'll never identify winning keywords.

But if you sell a $20 product with 29% conversion rate? You convert every 3-4 clicks. Here, 10 clicks makes sense.

Calculate your pause/negation threshold with this 4-step process:
1️⃣ Find your conversion rate
2️⃣ Calculate average clicks per sale
3️⃣ Add 25% buffer for statistical significance
4️⃣ That's your actual pause point

One-size-fits-all PPC rules are costing you money.

Your pausing/negation strategy should match your product's economics.

08/31/2026

Ranking for a keyword doesn’t mean bidding on that keyword.

We had a client staring down a high-volume head term — bully sticks — that was going to cost a fortune to push to page one.

The instinct is to throw dollars straight at the generic term.

We didn’t.

Instead we mapped it to the closest lower-cost variant: bully sticks for small dogs.

Same root phrase, meaningfully lower CPC, still highly relevant.

Here’s the mechanism: if “bully sticks for small dogs” sits at the front of the title and gets pushed through PPC, Amazon’s algorithm still credits the listing for “bully sticks” organically.

Two birds, one cheaper stone.

Rank-push campaigns are inherently inefficient by design.

Inefficient doesn’t have to mean careless.

Before you fund one:
• Map the head term to its lower-cost siblings
• Confirm the head term is already embedded in the listing copy
• Check competitor sales velocity actually justifies the push
• Run it exact match only — phrase match opens combinations you can’t track

The generic keyword isn’t always the right place to spend to rank for the generic keyword.