Jordan Fry

Dashboards show attribution, often assigning a sale to whichever touchpoint was nearby—most often a brand-name search from someone already going to buy.

• Jordan Fry treats this as a measurement problem, not fraud: nobody is lying; the platform reports what it is credited with, but the two terms describe different things.
• Any ROAS number needs to be checked against data outside the ad platform, such as margin, total revenue, total spend, and a direct test where available.

In one ecommerce account audit, 89% of sales were credited to people searching the brand name across 18 campaigns.

• The account read 3.18x in September and 3.8x in August.
• One product line showed 6x with brand included and 0.83x once brand searches were removed.

Read original · Original 2 · Original 3

Brand search harvests demand built elsewhere in the account, so including it in the same number as prospecting obscures which is working. For anyone reviewing an agency, the audit question is simple: has the account ever been assessed with brand removed?

The same account was carrying roughly $31,000 a month in awareness spend returning 0.4x, while about $3,000 a day of allocated budget went unspent. None of that showed up in the blended ROAS.

Read original

Google's beta Target ROAS calculator takes an average profit margin and returns a number: at 15% margin, that is 6.67x, or 667% in percentage terms.

• They point out that this is breakeven, not a goal: it is simply 1 divided by the margin, the point at which the advertiser made exactly zero on ad dollars.
• The calculator is only as accurate as the margin entered, and may miss shipping, returns, payment fees, and overhead, all of which move breakeven.
• Use it as a floor, and don't change live bids based on a beta tool alone.

If the real margin after returns, fees, and shipping is 12% rather than 15%, breakeven rises from 6.67x to 8.33x on the same account and campaigns. Jordan Fry also describes an ecommerce client with a 3.5x account target whose profitability depends on a margin recorded in store data, not Google Ads.

Read original

Jordan Fry describes Google opening Conversion Lift on Search and Performance Max, allowing eligible advertisers to set up the test without a representative.

• Google holds back a group of users who never see the ads and compares them with users who did; the gap represents sales the ads caused rather than sales they took credit for.
• They flag the limits: at least 1,000 observed conversions, at least $5,000 in campaign budget, and availability only to some accounts.
• For smaller accounts, the fallback is blended measurement across channels and refusing to treat brand-term ROAS as proof.

Jordan Fry writes that the first brand-search lift result an agency runs on its own accounts will be humbling, and would rather find that out deliberately than have a client find it out first.

Read original

They show why a healthy ROAS can still hide wasted spend and give the first audit step to uncover it.

Jordan Fry describes auditing an ecommerce account with a serious budget, where 89% of sales were credited to people searching the brand name.

• September ROAS was 3.18x and August’s was 3.8x, but splitting brand from non-brand showed one product line at 6x with brand included and 0.83x without.
• About $31,000 a month in awareness spend was returning 0.4x, while roughly $3,000 a day in allocated budget went unspent.
• They distinguish sales the platform credits from sales the ads actually created, arguing that brand search harvests demand built elsewhere.
• Their recommended first move: report brand and non-brand as two separate businesses, give non-brand its own target, then decide whether awareness spend is building demand or merely sitting alongside it.

For brands planning a campaign

For an agency pitching measurement cleanup, this post offers a concrete audit step that can change budget decisions: split brand and non-brand reporting before judging campaign performance, then set a separate non-brand target so brand search cannot mask weak prospecting.

Jordan Fry reports that Conversion Lift now works on Search and Performance Max without a Google representative setting it up.

• Eligible advertisers can run the test themselves: Google holds back a group of users who never see the ads, compares them with users who do, and treats the gap as sales the ads caused rather than sales they took credit for.
• They note that the test matters most on brand search, where dashboards cannot show how much revenue would have happened anyway.
• Eligibility requires at least 1,000 observed conversions and at least $5,000 in campaign budget; the feature is not available to every account.
• Ask whether the account meets those thresholds.
• For advertisers who do not qualify, they advise against treating brand-term ROAS as proof, recommend comparing total revenue with total ad spend across every channel, and predict that the first brand-search lift result an agency runs on its own accounts will be humbling.

For brands planning a campaign

If the client qualifies, propose a Conversion Lift test on brand search before renewing spend assumptions. If not, shift the measurement plan to blended revenue and total ad spend rather than relying on unverified brand ROAS.

Jordan Fry describes a Google test of a Target ROAS calculator based on the profit margin a business enters.

• In Google's example, a 15% margin produces a 667% Target ROAS, or 6.67x breakeven—not a target.
• At that number, the advertiser makes exactly $0 on ad dollars because the calculation is simply 1 divided by margin.
• The calculator is only as good as the margin entered, and the article notes it may not account for shipping, returns, payment fees, or overhead.
• If the real margin after those costs is 12% rather than 15%, breakeven rises from 6.67x to 8.33x, changing the bid on the same account and campaigns.
• They recommend using the calculator as a floor, not a target, and point to an ecommerce client holding a 3.5x account target whose profitability depends on a margin recorded in store data, not Google Ads.
• The suggested steps: get margin after returns, fees, and shipping; compare the calculator's breakeven with the current target to see the gap; and remember it is a beta, so do not change live bids based on the calculator alone.

For brands planning a campaign

Before setting or defending a ROAS target, audit margin, including returns, fees, shipping, and overhead. Treat the calculator's output as a breakeven floor, and keep live bid changes separate from a beta tool's recommendation.

September ROAS was 3.18x, compared with 3.8x in August.

• One product line showed 6x ROAS with brand included but 0.83x without it.
• Around $31K a month in awareness spend returned 0.4x, while roughly $3K a day in allocated budget went unspent.
• Fry's point: platform credit and sales caused are different things.
• Brand search harvests demand built elsewhere.
• The recommended first step is to report brand and non-brand as separate businesses, give non-brand its own target, then decide whether awareness spend is building demand or merely sitting near it.

Report brand and non-brand separately, set a distinct target for non-brand, and evaluate awareness spend against demand-building rather than blended ROAS.

A strong blended ROAS can still mask below-breakeven non-brand performance, as the product-line example shows.

For brands planning a campaign

For an agency pitching an ecommerce measurement review, ask for a brand/non-brand split by campaign and product line, a separate non-brand target, and an explanation of awareness spend and unspent budget. Use the split to show where blended ROAS masks weak non-brand performance.

Examples: Source 1

The test holds back a group of users who never see the ads, then compares them with users who did; the difference represents sales the ads caused rather than sales they took credit for.

• Fry flags qualification thresholds of at least 1,000 observed conversions and at least $5,000 in campaign budget, and notes that the test is not available to every account.
• For smaller accounts, the advice is not to treat brand-term ROAS as proof, but to compare total revenue with total ad spend across every channel and run the test only if the account is large enough.
• Fry expects the first brand-search lift result an agency runs on its own accounts to be humbling.

Where eligible, run a holdout-based Conversion Lift test; first check observed conversions and budget thresholds.

Most small and mid-size accounts will not meet the thresholds, so use blended measurement instead of a direct lift read.

For brands planning a campaign

For qualified ecommerce clients, plan a Conversion Lift test on brand search or Performance Max before accepting brand ROAS as proof of incremental sales. For smaller accounts, measure blended revenue against total ad spend and explain that a lift test is not yet available.

Examples: Source 1

Google's example gives a 15% margin a 667% Target ROAS, which Fry explains is breakeven: 1 divided by the margin.

• The calculator is only as good as the margin entered; Fry notes that the article says it may not account for shipping, returns, payment fees, or overhead.
• If the real margin is 12% instead of 15%, breakeven moves from 6.67x to 8.33x in the same account.
• Fry recommends using the calculator as a floor, not a target.
• Fry also cites an ecommerce client whose account target is 3.5x and whose profitable margin lives in store data rather than Google Ads.

Get the margin after returns, fees, and shipping; compare the calculator's breakeven with the current target; treat the gap as room to maneuver, and avoid changing live bids based on the calculator alone.

A target that looks acceptable in Google Ads can still be unprofitable if the true margin is lower than the figure entered in the calculator.

For brands planning a campaign

When reviewing ROAS targets for an ecommerce client, ask for margin after returns, fees, and shipping; compare it with any Google calculator breakeven; and check whether the current target ties back to the P&L. Do not change live bids based on the beta calculator alone.

Examples: Source 1

In one audit, 89% of sales were credited to brand searches, and blended ROAS hid weak non-brand performance.

• In another, brand search showed 10.1x ROAS and $242K of revenue in September, but Fry says a dashboard cannot show how much would have happened anyway.
• In the Target ROAS post, Fry warns that many accounts run on a ROAS number nobody can tie back to the P&L.
• The through-line is to bring in business data—margin, total revenue, total spend, and holdout tests—before accepting a platform metric as proof.

Pair platform reporting with margin data, total revenue and spend, and holdout testing where available.

For brands planning a campaign

Use this principle in a client audit: ask what each platform metric is credited for, what business outcome sits behind it, and what direct test or P&L input can confirm it.

Examples: Source 1 · Source 2 · Source 3

The posts discuss Google as an ad platform, not as a sponsor or client.

View example

Both treat brand traffic and account structure as things to isolate before trusting platform numbers. Davies' impression-share diagnosis and Performance Max brand-exclusion script address the same measurement problem they raise with brand versus non-brand splits.

Davies' examples focus on diagnosing slower sales through impression share and deciding whether to exclude brand from Performance Max; Jordan Fry's posts focus on ROAS targets, margin-based breakeven, and Conversion Lift holdout testing.

Read original · Original 2

Both argue ROAS can look healthy while margin, new-customer acquisition, and brand-driven conversions hide weak incremental performance.

Tesselaar gives a step-by-step Performance Max operating structure—branded search protection, margin-split asset groups, CRM signals, value rules, and search themes—and introduces nCAC and POAS. Jordan Fry concentrates on auditing reported ROAS, separating brand from non-brand, and testing causation with Conversion Lift.

Read original · Original 2

Daniv's numbered diagnostic lists use a similar teaching format to Jordan Fry's breakdowns of measurement choices.

Daniv focuses on SEO delivery and client decision-making, including agency-client bottlenecks; Jordan Fry focuses on paid-search measurement and causation. The overlap is the ecommerce and agency audience and practical list format, not the channel being discussed.

Read original · Original 2

The posts explain why ROAS can conceal margin losses, returning-customer bias, and diminishing returns; another gives a practical Performance Max structure based on brand separation, product margins, customer data, and conversion value. Robin is the stronger match when the audit must go beyond a brand/non-brand split to examine account economics and campaign structure.

The reviewed examples focus on profitability and Performance Max setup, not self-serve Conversion Lift eligibility or a margin-based Target ROAS calculator. Jordan Fry is the closer match if the agency’s immediate deliverable is to test whether branded Search or Performance Max sales were incremental, then set a margin-informed breakeven floor.

Examples: Source 1 · Source 2

One post recommends checking daily impression share against the period before sales dipped to distinguish weaker market demand from competitor pressure. Another describes using a script to compare brand and non-brand traffic in Performance Max before deciding whether to route brand demand through Search instead.

These examples offer useful checks for demand shifts and Performance Max brand traffic, but Jordan Fry’s posts do not show a broader sequence for separating credited from caused sales, calculating breakeven ROAS from actual margin, and testing incrementality through Conversion Lift. Martin is a better fit for those specific account diagnostics; Jordan is better evidenced for the complete measurement need.

Examples: Source 1 · Source 2

The posts turn SEO into four concrete workstreams and identify practical causes of campaign delay, such as missing access, fragmented feedback, and unclear decision ownership. Andrew is more relevant if the campaign needs to improve the client-agency operating process alongside channel work.

The posts concern SEO delivery and agency collaboration, not ecommerce Google Ads measurement, ROAS targets, or incrementality testing. Jordan Fry is the more direct choice for this need; Andrew is a better alternative only if client decision-making or a plain-language SEO program is the actual bottleneck.

Examples: Source 1 · Source 2