Paid Ads & Creative
7 Checks: Should I Cut Non-Brand Google Ads in B2B?

Joshua Jackai
16 min

Published 13 November 2026 ยท Last updated and fact-checked November 2026
Short answer: Do not cut non-brand Google Ads on last-click numbers alone. In B2B, a non-brand click often starts a journey that converts weeks later on a brand search or an organic visit. Check your conversion window, the path reports and model comparison first, then run a geo holdout scored on CRM pipeline before you decide.
Here is the situation I see on US B2B accounts every quarter. The non-brand campaign costs four times as much per lead as the brand campaign. The CFO asks why you pay for "generic" clicks when brand search brings in the leads. Then someone in the PPC team notices that a lead clicked a non-brand ad, left, came back three weeks later through branded organic search and filled in the demo form. Google Ads gave the credit to the non-brand click. GA4 gave it to organic. The CRM says "Organic Search".
Every page I found that ranks for this question answers the reverse one: whether you should keep paying for your own brand terms. This guide is the walkthrough that was missing for B2B lead gen teams. It covers the seven checks I run before I would cut a non-brand campaign, in the order I run them, with the Google Ads reports to open and the peer-reviewed research on what paid search actually causes.
Key takeaways
Google Ads applies a 30-day conversion window to Search conversions by default, and you can set it from 1 to 90 days, per Google Ads Help.
Only two attribution models remain in Google Ads: data-driven, the default for most conversion actions, and last click, per Google Ads Help.
Rutz and Bucklin (2011) found that generic search activity raised future branded search, and that the spillover did not run the other way.
Blake, Nosko and Tadelis (2015) found in eBay field experiments that new and infrequent users were positively influenced by non-brand ads.
Gordon and colleagues (2019) found that observational methods often fail to recover the effects measured in randomized ad experiments.
Lewis and Rao (2015) found the median confidence interval on ad return on investment was over 100 percentage points wide across 25 field experiments.
Check | Best for | Report or method | Key number |
|---|---|---|---|
1. Conversion window vs sales cycle | Long B2B cycles | Conversion action settings | 30 days default, 1 to 90 allowed |
2. Path metrics and top paths | Seeing non-brand then brand journeys | Attribution, path reports | Days and interactions to convert |
3. Model comparison | A quick credit check | Attribution, model comparison | 2 models left |
4. GA4 vs Google Ads gap | Reconciling reports | Conversions by conv. time columns | 7 days of GA4 reattribution |
5. Brand search volume | Spotting spillover | Brand query trend | Asymmetric spillover |
6. Geo holdout | A causal answer | Matched regions or Conversion Lift | 100+ point ROI intervals |
7. CRM pipeline and blended cost per lead | The final decision | CRM export by region | Pipeline per dollar |
1. Match the conversion window to your sales cycle: best for long B2B cycles
Before you judge non-brand, check that Google Ads can see its conversions at all. Google applies a 30-day conversion window to Search conversions unless you change it, and the window can run from 1 to 90 days, per Google Ads Help. If first click to form fill takes longer than the window, the credit is simply lost.
Where to look: Goals, then Conversions, then the conversion action's settings
Default for Search: 30 days after the ad click
Allowed range: 1 to 90 days, depending on the conversion source
What to compare it with: CRM days from first touch to demo request
Best for: B2B offers with a research phase of three weeks or more
The window is the period after an ad interaction during which Google Ads records a conversion. A non-brand click on day 1 followed by a brand search and a demo request on day 40 never reaches the non-brand campaign under a 30-day window. It looks like a brand lead. That is not a model problem. It is a setting.
I pull the number from the CRM, not from Google. Take the last 50 closed or qualified opportunities that came from paid search, find the first recorded web visit and the demo request date, and look at the spread. If the middle of that spread sits near or past 30 days, the default window hides part of what non-brand does. Lengthen the window on the primary conversion action, note the date you changed it, and give the account at least one full window before you read the new numbers.
One caution: the window only extends credit for clicks Google already tracks. It does not fix leads that lose their click ID on the way into the CRM. My guide to website lead tracking covers that leak.
"Generic search activity positively affects future branded search activity through awareness of relevance," Oliver Rutz of Yale University and Randolph Bucklin of UCLA wrote in the Journal of Marketing Research.
Data point: Google Ads uses a 30-day default window for Search and Display conversions and allows 1 to 90 days, per About conversion windows, read 13 November 2026.

2. Read path metrics and top paths: best for seeing non-brand then brand journeys
The path reports show whether non-brand clicks sit early in journeys that end on brand. Google Ads attribution reports include path metrics, which show how many days and interactions it takes users to convert, and top paths, which show the most common sequences of ad interactions before a conversion, per Google Ads Help.
Where to look: Goals, then Measurement, then Attribution
Top paths: the most common ad sequences before a conversion
Path metrics: days and interactions to conversion
What to filter for: paths that start on a non-brand campaign and end on brand
Best for: answering "does non-brand start the journeys brand finishes?"
Open top paths, pick the demo request conversion action, and set the date range to at least one full sales cycle. Then read the paths, not the totals. On the accounts I have audited, the useful pattern is short: a non-brand campaign first, a brand campaign last, and two to five interactions in between. If that pattern is rare, non-brand is not feeding brand inside Google Ads, and the case for keeping it rests on direct leads only.
Two limits matter. The reports only see Google ad interactions, so a non-brand click followed by an organic visit shows up as a one-step path or not at all. And a path is a sequence, not a cause. A buyer who was always going to search your name may also have clicked a generic ad on the way. That is the selection problem Blake, Nosko and Tadelis found at eBay, and it is why check 6 exists.
"Because search clicks and purchase intent are correlated, we show that returns from paid search are a fraction of non-experimental estimates," Steven Tadelis of the University of California, Berkeley and co-authors wrote in Econometrica.
Data point: In the eBay experiments, brand keyword ads showed no measurable short-term benefit, while new and infrequent users were positively influenced by non-brand ads, per Blake, Nosko and Tadelis (2015).

3. Run model comparison on last click vs data-driven: best for a quick credit check
Model comparison shows how much credit non-brand gains or loses when you change the attribution model. Google Ads now offers two models, data-driven and last click; first click, linear, time decay and position-based are no longer supported, per Google Ads Help. Data-driven is the default for most conversion actions.
Where to look: Attribution, then Model comparison
Models left: data-driven (default) and last click
Data-driven: credit based on your past data for that conversion action
Last click: all credit to the last-clicked ad and keyword
Best for: seeing whether non-brand looks better when early clicks count
Compare the non-brand campaign's conversions and cost per conversion under both models. If non-brand gains a lot of conversions under data-driven against last click, it is doing early work that last click hides. If the two numbers are close, non-brand is either converting on its own or not feeding brand clicks inside Google Ads.
I treat this as a pointer, not a verdict. Data-driven attribution still spreads credit across clicks that happened. It does not ask what would have happened without the ad. Li and Kannan (2014) built a model that attributes the incremental value of each channel from individual touch data, and found channel contributions that differed significantly from the usual metrics, then checked the model with a field study. That is the gap between credit and cause, and model comparison stays on the credit side.
"These channels' relative contributions are significantly different from those found by other currently used metrics," H. Alice Li and P.K. Kannan of the University of Maryland wrote in the Journal of Marketing Research.
Data point: Google moved conversion actions that used the retired rules-based models to data-driven attribution, and last click remains selectable, per About attribution models, read 13 November 2026.

4. Explain why GA4 and Google Ads disagree: best for reconciling reports
GA4 and Google Ads give the same lead to different sources because they count different things. Google Ads only credits its own ad interactions, so a non-brand click followed by an organic return can still count for the ad. GA4 attributes across all channels and can reattribute conversions for up to 7 days, per Analytics Help.
Google Ads credits: Google ad interactions only
GA4 credits: every channel, including organic and direct (direct only when the path is all direct)
Google Ads default date: the date of the ad interaction
Conversions (by conv. time): the date the conversion happened
Best for: explaining to finance why the totals differ
There is also a date gap. The standard Conversions column in Google Ads files a conversion under the date of the click, while the Conversions (by conv. time) columns file it under the date of the conversion, per Google's help page on conversions by time. A non-brand click in October that converts in November shows up in October in one view and November in the other. When someone compares a Google Ads export with a GA4 report for the same month, both can be right and still differ.
For a cut decision, I stop trying to make the two match. I pick one source of truth for the decision, usually the CRM, and use GA4 and Google Ads only to explain where a lead came from. The leads in the Reddit question above, non-brand click then branded organic return, are normal behavior for a considered purchase. They are not a landing page problem by themselves.
"I stop trying to make the two match and pick one source of truth for the decision," said Joshua Jackai, founder of Demand Designer.
Data point: GA4 attribution models exclude direct visits from credit unless the path is made up only of direct visits, per Analytics Help, read 13 November 2026.

5. Watch brand search volume as the spillover signal: best for spotting awareness effects
If non-brand ads build awareness, brand searches should move with them. That is what Rutz and Bucklin (2011) found for a major lodging chain: generic search activity raised later branded search, and branded search did not raise generic search. The effect runs one way, from non-brand to brand.
Signal to track: weekly impressions on your brand campaign, plus brand queries in Search Console
Compare against: weekly non-brand spend, lagged by your sales cycle
What spillover looks like: brand volume rises a few weeks after non-brand spend rises
What no spillover looks like: brand volume flat whatever non-brand does
Best for: a cheap first read before a formal test
I chart three lines on one weekly axis: non-brand spend, brand campaign impressions and brand queries from Search Console. Then I look for brand moving a few weeks after non-brand. Spend changes that happened for other reasons, like a budget cut in August, are useful natural tests.
This check has the same weakness as the path reports. A trade show, a podcast mention or a sales push can raise brand searches in the same weeks. So I use it to decide whether a holdout is worth running, not to decide the cut. If brand volume never moves with non-brand spend over a year of data, I would expect a small holdout result and size the test to match.
"Branded search does not affect generic search, demonstrating that the spillover is asymmetric," Rutz and Bucklin wrote in the Journal of Marketing Research.
Data point: Rutz and Bucklin used a Bayesian dynamic linear model on a paid search campaign for a major lodging chain, a B2C setting, so B2B teams should treat it as direction, not size, per the paper.

6. Run a geo holdout: best for a causal answer
A geo holdout is the only check on this list that measures what non-brand causes. You pause non-brand in a set of matched regions, keep it running in the rest, and compare total leads and pipeline. Google's Conversion Lift has a geography-based option that supports offline data, but it is not open to every account, per Google Ads Help.
Design: matched US states or metro areas, split into test (paused) and control (running)
Length: at least one full sales cycle plus the conversion window
Google route: Conversion Lift, geography-based, through your account rep
Do-it-yourself route: location exclusions on the non-brand campaigns in test regions only
Best for: the go or no-go on a cut
I match regions on past leads, pipeline and brand search volume, not on population. For a US B2B account, that usually means pairing states or metro areas with similar lead history and keeping big outliers like the head office state out of the test. Keep brand campaigns, budgets and bids the same in both groups, and freeze other changes during the test.
Expect noise. Lewis and Rao (2015) ran 25 large field experiments for US retailers and brokerages and found the median confidence interval on return on investment was over 100 percentage points wide. A B2B account with 40 demo requests a month will not get a tight answer from a two-week test. Run it longer, use pipeline rather than leads, and accept a directional result if that is what the volume allows.
"Observational methods often fail to accurately recover the treatment effects generated from randomized advertising experiments on Facebook," Brett Gordon of Northwestern University and co-authors wrote in Marketing Science.
Data point: Google's geography-based Conversion Lift reports incremental conversions, incremental conversion value, incremental ROAS and incremental cost, per About Conversion Lift, read 13 November 2026.

7. Score the test on CRM pipeline and blended cost per lead: best for the final decision
Score the holdout on what the business buys: qualified pipeline and blended cost per lead across brand, non-brand and organic, by region. If pipeline in the paused regions falls by more than the non-brand spend saved, keep non-brand. If pipeline holds while spend drops, cut or trim it.
Primary metric: qualified pipeline per region, from the CRM
Secondary metric: blended cost per qualified lead, all search sources combined
Also watch: brand campaign leads and organic leads in the paused regions
Decision rule: pipeline lost against spend saved
Best for: a decision finance will sign off
Blended is the point. A non-brand cut that lifts brand cost per lead looks great in the brand campaign report and terrible in the bank account. So I add every search-sourced lead in each region, brand paid, non-brand paid and organic, and divide total paid search spend by that number. Then I do the same with pipeline.
Here is an illustration, not client data. Say the control regions keep spending and the test regions save $12,000 over the test. If test-region pipeline drops by $30,000 against control after adjusting for region size, non-brand was earning more than it cost at your close rate. If test-region pipeline drops by $3,000, the cut is cheap. The formula below runs the same arithmetic with your numbers.
"A non-brand cut that lifts brand cost per lead looks great in the brand report and terrible in the bank account," said Joshua Jackai, founder of Demand Designer.
Data point: On Bing brand terms, competitors took 18% to 42% of clicks from brands that faced competition but did not advertise, per Simonov, Nosko and Rao (2018), which is why I keep brand campaigns running during a non-brand test.

Comparison matrix: which check to run first
Start with the free checks. Checks 1 to 4 take about an hour and a half in total and often settle the question before any spend is at risk. Check 5 tells you whether a holdout is worth running. Checks 6 and 7 cost leads and time, so run them only when the free checks leave the cut decision open.
Check | Time to run | Cost | Tells you | Does not tell you |
|---|---|---|---|---|
1. Conversion window | 15 minutes | Free | Whether Google can see late conversions | Whether the ads caused them |
2. Path reports | 30 minutes | Free | Whether non-brand starts brand journeys | Organic returns |
3. Model comparison | 15 minutes | Free | How credit shifts between models | Incremental effect |
4. GA4 vs Google Ads | 30 minutes | Free | Why totals differ | Which source is right |
5. Brand search trend | 1 hour | Free | Whether brand moves with non-brand | Cause, if other campaigns overlap |
6. Geo holdout | One sales cycle plus window | Lost leads in test regions | What non-brand causes | Long-term brand effects |
7. CRM scoring | 2 hours after the test | Free | Pipeline lost against spend saved | Anything without clean CRM sources |
Holdout formula: pipeline lost against spend saved
Use one formula with your own CRM numbers. Scale control pipeline to the test group's size: control pipeline times the test regions' share of past pipeline, divided by the control share. Subtract actual test pipeline to get pipeline lost, then compare it with the non-brand spend saved.
Expected test pipeline: control pipeline x test share / control share
Pipeline lost: expected test pipeline minus actual test pipeline
Keep non-brand when: pipeline lost, at your close rate, is worth more than spend saved
Methodology
I read the Google Ads Help and Analytics Help pages cited here on 13 November 2026 and checked each paper's DOI, journal and year through Crossref and OpenAlex. The worked numbers in check 7 are an illustration, not client data. The research covers eBay, Bing, Facebook, a lodging chain and US retailers, so B2B effects may be larger or smaller.
Frequently asked questions
Should I cut non-brand Google Ads if brand search gets the conversion?
Not on last-click reports alone. Non-brand clicks can start journeys that end on brand or organic search weeks later, and research found generic search raises later branded search. Check the conversion window, path reports and model comparison, then run a geo holdout scored on CRM pipeline before cutting.
How do I credit a non-brand click that converts later on brand search?
Google Ads credits its own ad interactions, so under data-driven attribution the non-brand click can share credit with the later brand click if both fall inside the conversion window. GA4 may credit organic search instead. For the cut decision, use one source of truth, usually the CRM.
Is non-brand search incremental?
Sometimes. In eBay field experiments, non-brand ads positively influenced new and infrequent users, while frequent users who would buy anyway took most of the spend, so average returns there were negative. Your result depends on how many new buyers your non-brand terms reach, which only a holdout measures.
How long should a non-brand holdout test run?
At least one full sales cycle plus your conversion window, measured from the CRM. For a B2B account with a 45-day cycle and a 60-day window, that means about three to four months. Shorter tests on small lead volumes give results too noisy to act on.
How I run this check on client accounts
On a paid search audit I run checks 1 to 4 in the first session, because they are free and they often change the question. Often the cut debate ends at check 1, when the conversion window turns out to be shorter than the sales cycle.
When it does not, I design the holdout with the client's sales lead, so the pipeline numbers come from the CRM they already trust. If you want a quick read on which ads to kill and which to keep, the Kill Sheet below gives you the thresholds I use every Monday.
Sources
Peer-reviewed research
1. Blake, T., Nosko, C. and Tadelis, S. (2015). Consumer Heterogeneity and Paid Search Effectiveness: A Large-Scale Field Experiment. Econometrica. https://doi.org/10.3982/ECTA12423 2. Rutz, O. J. and Bucklin, R. E. (2011). From Generic to Branded: A Model of Spillover in Paid Search Advertising. Journal of Marketing Research. https://doi.org/10.1509/jmkr.48.1.87 3. Gordon, B. R., Zettelmeyer, F., Bhargava, N. and Chapsky, D. (2019). A Comparison of Approaches to Advertising Measurement: Evidence from Big Field Experiments at Facebook. Marketing Science. https://doi.org/10.1287/mksc.2018.1135 4. Lewis, R. A. and Rao, J. M. (2015). The Unfavorable Economics of Measuring the Returns to Advertising. The Quarterly Journal of Economics. https://doi.org/10.1093/qje/qjv023 5. Li, H. A. and Kannan, P. K. (2014). Attributing Conversions in a Multichannel Online Marketing Environment: An Empirical Model and a Field Experiment. Journal of Marketing Research. https://doi.org/10.1509/jmr.13.0050 6. Simonov, A., Nosko, C. and Rao, J. M. (2018). Competition and Crowd-Out for Brand Keywords in Sponsored Search. Marketing Science. https://doi.org/10.1287/mksc.2017.1065
Google documentation (not peer reviewed)
Google Ads Help. About conversion windows. Read 13 November 2026. https://support.google.com/google-ads/answer/3123169
Google Ads Help. About attribution models. Read 13 November 2026. https://support.google.com/google-ads/answer/6259715
Google Ads Help. About attribution reports. Read 13 November 2026. https://support.google.com/google-ads/answer/1722023
Google Ads Help. Understand conversions by time with new columns. Read 13 November 2026. https://support.google.com/google-ads/answer/9549009
Google Ads Help. About Conversion Lift. Read 13 November 2026. https://support.google.com/google-ads/answer/12003020
Analytics Help. Attribution in Google Analytics. Read 13 November 2026. https://support.google.com/analytics/answer/10596866
Free workbook
The Kill Sheet
Know when to kill an ad and when to let it run, with thresholds for the five signals that matter.
Five signals, three verdicts, ten minutes every Monday. Free.







