In most accounts Echelonn audits, 50 to 75% of the conversions in a "prospecting" campaign come from people typing the brand's own name. Here are the three checks and five setup levels that fix it, and how to tell whether the new customers are worth what you paid.
On this page
This post is the written version of Scale to Profit, the webinar we co-hosted with Echelonn on 12 November 2026, about setting up Google Ads for ecommerce so that it wins customers you don't already have.
Most advice on the channel is about which campaigns to run. The more useful question is who those campaigns are reaching, because in a lot of accounts a large share of the "prospecting" budget goes on people who were already going to buy.
Jackson Blackledge and Filippo Caroli from Echelonn, an agency that only does Google Ads for ecommerce brands and has managed $185m of spend so far this year, walked through three checks you can run in your own account today, then a five-level setup for making Google find genuinely new customers.
My co-founder Karl O'Brien then added the profit maths, which is at the end. Prefer to watch? The full 60-minute replay is here.
Is Google Ads only a bottom-of-funnel channel?
A lot of brand owners believe Google only catches people who were about to buy anyway, so spending more on it just pays for sales Meta had already won. You may have seen it yourself: you turn the budget up, ROAS looks fine, total revenue barely moves, and the next dollar goes back into Meta.
Jackson's answer is that this is the setup, not the platform. Left on its defaults, Google looks for the easiest conversions available, and those are branded searches and warm audiences. With a different setup the same platform reaches people across the whole funnel, and that holds whether you spend $5,000 a month or $1m.

What is incrementality?
An incremental sale is one that would not have happened without the ad. Attribution tells you which channel got the credit, while incrementality tells you what you would have lost by switching the ad off, and the two often disagree, which is why we've argued before that attribution alone is a poor source of truth. A campaign can report a strong ROAS and add almost nothing, if the people it reached were going to buy from you regardless.
Three checks to run in your Google Ads account today
Check 1: how much of your prospecting converts on your own name?
Pull the search terms report for your biggest prospecting campaign, sort by conversions, and count how many of the converting queries contain your brand name. In most of the audits Echelonn runs it is between 50 and 75%, which means a campaign labelled prospecting is mostly converting people who typed the brand into Google.
Check 2: what are you paying for those brand clicks?
Inside a prospecting campaign contaminated with brand traffic, a branded click routinely costs $6 or more. The same click in a dedicated brand campaign runs at about $0.30 to $0.40, so it is the same person, the same search and the same purchase at 15 times the price or more. It is hard to stop because it flatters the campaign: these people convert well, ROAS climbs, and it looks like something worth scaling.
Check 3: what share of purchases are new customers?
If a campaign brings in mostly new customers, you have something to scale. If it is mostly returning customers, you are re-serving people you already had, however good the ROAS looks. In the attribution report Jackson showed, new-customer share of purchases ranged from 90% on the best channel down to 22% on the worst, with Google at 89%.
Passing this check doesn't prove incrementality, though. A new customer isn't necessarily an incremental one, because plenty of them were warmed up by Meta, YouTube or email, then searched your brand and bought. They are new to your store but not new to your marketing. If you don't use an attribution tool, pull the new versus returning split in Google Ads and compare it with your Shopify orders.
| Check | Where to look | Warning sign |
|---|---|---|
| Brand queries in prospecting | Search terms report, sorted by conversions | 50 to 75% of converting queries contain your brand name |
| Cost of a brand click | CPC on brand queries, by campaign | $6+ in prospecting against $0.30 to $0.40 in a brand campaign |
| New-customer share | Your attribution tool, or Google Ads against Shopify orders | New customers are a small share of the campaign's purchases |
The five levels of an incremental ecommerce Google Ads strategy
Knowing you have the problem doesn't fix it, because there isn't one switch. There are five, they stack, and the order matters: a new creative at level five tells you very little if brand traffic is still leaking into prospecting at level one, since you can't tell whether the ad worked or Google just found more of your own customers. Jackson's view is that most accounts are stuck on level one, and that simply reaching level two usually moves the needle more than any bidding change.

| Level | What you do | What it stops |
|---|---|---|
| 1. Brand separation | Keep brand and non-brand apart in every campaign type | Prospecting budget buying your own name |
| 2. Existing customers | Exclude past purchasers from prospecting | Paying to re-acquire people who already buy |
| 3. Hot audiences | Exclude visitors another channel brought first | Google finishing Meta's job and taking the credit |
| 4. Conversion tracking | New-customer conversion actions and short view windows | Five-second video views counted as sales |
| 5. Creative and funnel | Problem-led ads and pages built for people who don't know you | Ads that only convert people who already do |
Level 1: keep brand and non-brand apart
Almost everyone has a branded Search campaign. Far fewer have the same split on Shopping, and Filippo still sees six-figure accounts without it. On Search, the brand campaign holds your name, misspellings and product names on exact and phrase match with a low bid cap, and those brand terms go in as negatives everywhere else. On Shopping, brand queries go to a low-priority Standard Shopping campaign, never Performance Max, held in place with negative keywords, priority settings and a regular script.
Non-brand then gets split by intent, and the higher up the funnel the keyword sits, the more incremental it tends to be. Someone searching "how to stop gums bleeding" who ends up buying your toothbrush is far more likely to be new than someone searching "bamboo toothbrush", who may well have seen your Meta ad last week. That caveat matters, because non-brand is not automatically incremental: in a niche where you are one of two or three players, your own social ads are probably creating a lot of those product searches.
Level 2: exclude the customers you already have
Do it two ways at once, because neither list catches everyone. The first is a Customer Match list of all purchasers, synced daily from Klaviyo. The second is a tag-based purchasers list from GA4 or Google Ads, which can hold up to 540 days. Emails miss people who checked out with a different address and cookies get lost, so the two together cover more than either alone.
Two details are worth checking this week. In a lot of the accounts Echelonn audits, the Klaviyo sync was set up once and quietly broke months ago, so open it and confirm the list is still growing. And decide when a customer counts as lapsed, roughly three months for supplements and one to two years for furniture, because someone who last bought three years ago is close to a new customer again and can come off the exclusion.
Filippo told a story here that stuck with me. He was auditing an account spending around $500,000 a month, and when he pointed out the gaps the CMO said they didn't care what Google reported because they looked at a third-party attribution tool. His reply was that Google isn't looking at that tool. It optimises on the data you give it, so the lists and conversion settings inside the account decide who it goes after, whatever dashboard you read afterwards.
Level 3: exclude the audiences your other channels warmed up
Once existing customers are excluded, the next easiest conversions are people who have already visited your site, and for most DTC brands that mostly means people Meta sent. If you spend six or seven figures a month on Meta there is a large pool of them, and they will convert on almost any Google campaign you put in front of them.
The fix is a GA4 audience of users whose first visit came from social, built on "first user source" with a pattern that matches facebook, instagram, snapchat and meta, and a membership of 30 or 60 days. GA4 shows the size as you build it. In one account spending more than $1m a month on Meta, that audience was 967,000 people, 49.4% of all users.
Use it as an exclusion on the campaigns you are counting on for new customers, above all Demand Gen and YouTube, where there is no search term to tell you how cold the person was. On Search and Shopping, Echelonn usually stops at excluding existing customers, and retargeting campaigns obviously keep the audience in.
Level 4: tell Google what counts as a conversion
Start by splitting the purchase conversion into new-customer and returning-customer actions, and make the new-customer one the primary goal on prospecting campaigns. That way Google can only score a result on paper when the buyer is new, which matters most for supplements, skincare and anything else with heavy repeat purchase.
Then look at the conversion windows, which is the part most people miss. Google counts conversions from clicks, and also from "engaged views": someone watches 10 seconds of a skippable ad, or five seconds of a short one, skips it, and buys later on their own. That counts as a full conversion, exactly as a click would, for three days by default. Filippo regularly finds accounts where someone has stretched that window to 30 days, which makes the campaign look far better and tells you far less.
For an incremental setup, set the engaged-view window to one day, the minimum, and leave view-through conversions off. To see where you stand, segment a Demand Gen or Performance Max campaign by conversions and then by ad event type. The healthy account on screen showed 327 conversions from clicks and 47 from engaged views. In the $500,000-a-month account it was the other way round, with about 90% of conversions coming from views with no click at all.
Level 5: let the creative and the landing page pick the audience
This is the idea most people haven't met on Google: your creative and your landing page steer who the algorithm goes after. Google scales towards whichever audience converts on your funnel, so if the funnel only works on people who already know you, that is the only audience it will find, and you hit a ceiling at $5,000 or $10,000 a day that looks like limited search volume and isn't.
| Campaign type | The lever | Finds warm people | Finds new people |
|---|---|---|---|
| Search | Keyword and landing page | "Bamboo toothbrush" to a product page with an offer | "How to stop gums bleeding" to an advertorial |
| Shopping | The listing | One generic listing per product | A listing per use case: sleep, anxiety, kids, cramps |
| YouTube | The creative type | Promo that opens on the offer | Problem first, product introduced late |
| Discover | The angle | "Buy one, get one free" image | Native, problem-led headline with no offer |
Collection pages and standard product pages tend to be the least incremental destinations, because they say almost nothing to someone who has never heard of the brand. Advertorials, comparison pages and pages built around one long-tail search do much better with cold traffic.

On Shopping the same idea becomes duplication. One magnesium supplement can run as five listings, each with its own title, image and landing page, built to win "magnesium for sleep", "magnesium for anxiety", "magnesium for kids" and so on, searches a single generic listing would never show for. The same advertiser can also hold several slots in the carousel for one search, which matters a lot for brands with one or two hero products. Asked in the Q&A whether Google penalises this, Filippo said no, with one rule: don't duplicate GTINs.

The account where it worked: Discover from zero to $10,000 a day
Discover is the scrolling feed under the Google search box, and inside Demand Gen campaigns you can target it on its own. The closing example was an account that went from no Discover spend to about $10,000 a day: $460,000 spent between August and October for $1.06m of conversion value, a 2.3x return, with $12,500 spent on the last day of the period.
The funnel had no offer in it. A native-looking ad led with a problem ("Waking up at 3am every night? It might not be stress"), clicked through to an advertorial of roughly 5,000 words with almost no call to action, and only then reached a sales page. It works because every other level was already in place, so the campaign was non-brand, excluded customers and hot audiences, and could only score a conversion on a click.

From incremental ROAS to profit per new customer
A clean setup tells you Google really did win the customer. It doesn't tell you whether that customer was worth winning. Karl's half of the session was about that second question, and at StoreHero we look at it as three numbers side by side: what it cost to acquire the customer, the profit on their first order, and the profit they return over time.
Take a $100 order. Shopify shows $100, but $30 of that is product cost, then shipping, packing and tax come out, then the cost of acquiring the customer, and what's left is the contribution margin on that new customer. In Karl's example, letting acquisition cost drift from $25 to $35 shrinks that slice so much that the brand needs three times the order volume to make the same profit.
This is also why ROAS on its own misleads. To an ad platform, two brands with the same revenue and the same spend are the same business, and they aren't, which is why your target should start from your own breakeven ROAS and not from a benchmark.
| Brand A | Brand B | |
|---|---|---|
| Revenue | $500,000 | $500,000 |
| Ad spend | $200,000 | $200,000 |
| ROAS | 2.5 | 2.5 |
| Gross margin | 60% | 40% |
| Left after ad spend | $100,000 | $0 |
Profit now, profit soon or profit later?
How much you can afford to pay for a new customer depends on when they pay you back, and a furniture brand shouldn't run acquisition like a subscription supplements brand. Karl splits businesses into three types, and knowing which one you are sets the target you give Google.
| Type | Typical brand | First order has to | Karl's example |
|---|---|---|---|
| Profit now | Furniture, one-off purchases, CPG without subscription | Pay for itself, since you can't count on a second order | $6 to acquire, $13 of profit on order one |
| Profit soon | Beauty, skincare, apparel | Roughly break even, with margin following within months | A routine product that gets reordered |
| Profit later | Subscription brands | Nothing yet, if 3 to 12 month value is proven | A $15 loss on order one, earned back over time |
It also depends on what people buy first, because some products bring in customers who stay and others bring in customers you never see again. That's worth knowing before you decide which product leads your new-customer offer and which listings get the Shopping budget.
What to check before Black Friday
We finished by asking everyone for one tip with BFCM just over a week away.
- Check how your promotion will actually show on Shopping. Price drop annotations, the sale badge and Merchant Center promotions are three different things. If your store runs a struck-through "sale" price all year, Google stops believing the anchor price and the sale badge is off the table when you finally cut prices for real.
- Give cold acquisition its own targets. Split brand from non-brand and new from returning before the week starts, or you end up congratulating Google for sales you already had.
- Reset your CPA and ROAS targets for the discount. The targets you ran for ten months assume full price. Watch profit per order through the sale, because a discount only pays when it gets people to spend more than they otherwise would.
- Pick two goals, not three. Record sales, record profit and clearing inventory rarely happen in the same week, and the brands that do best choose two and plan their offers around them.
- Don't switch off after Cyber Monday. From mid-December to mid-January, the stretch some buyers call Q5, ad costs fall and people are still shopping, mostly for themselves.
Where to start
Run the three checks first, since they take about twenty minutes and tell you which level your account is really on. Then work up in order, and don't touch creative until brand and existing customers are out of your prospecting campaigns. Echelonn's 2026 Google Ads playbook goes deeper on the setup itself. As you scale, keep one number from outside the ad account in view, such as blended MER or contribution margin per new customer, so you can see whether the growth is real.
An incremental setup tells you Google won the customer. StoreHero shows what they cost, the profit on their first order and what they return over time, by channel and by product. Book a demo and bring your numbers.
Key takeaways
- In Echelonn's audits, 50 to 75% of the conversions in a typical "prospecting" campaign come from searches containing the brand's own name.
- A brand click inside a prospecting campaign routinely costs $6 or more, against $0.30 to $0.40 in a dedicated brand campaign.
- A new customer is not automatically an incremental one: many were warmed up by Meta, YouTube or email before they searched.
- The five levels go in order: brand separation, existing customers, hot audiences, conversion tracking, then creative and funnel.
- Set the engaged-view conversion window to one day, so a five-second video view can't claim a sale three days later.
- Judge the result on profit per new customer, because the same ROAS means very different things at a 60% and a 40% margin.
Frequently asked questions
Is Google Ads worth it for an ecommerce store?
It is when it reaches people who would not have bought anyway. Check how many of your prospecting conversions come from searches for your own brand name and what share of purchases are new customers. If most of the spend is re-buying existing demand, fix the setup before you judge the channel.
Is $500 a month enough for Google Ads?
It can be enough to learn from, as long as it goes on non-brand searches or Shopping listings and you know your breakeven ROAS first. The setup matters more than the budget: Echelonn's point in the session was that the same five levels apply whatever you spend.
Is $20 a day good for Google Ads?
$20 a day is about $600 a month, which is workable for a small catalogue if the budget is kept away from brand searches and existing customers. Compare the cost of each new customer with the profit on their first order, not with ROAS alone.
What is incremental ROAS?
Incremental ROAS is the revenue that would not have happened without the ad, divided by the ad spend. It is lower than platform-reported ROAS because it leaves out brand searchers, existing customers and people another channel had already convinced.




