Are Your Brand Search Ads Generating Sales or Just Replacing Organic Clicks?

Your brand-search campaign can report excellent returns while buying traffic your website would have received anyway. In a 2014 NBER paper examining eBay’s 2012 advertising tests, researchers reported that organic search replaced 99.5% of the traffic lost when branded paid-search ads were removed in one analysis.

That historical result concerns one highly recognized business. It does not establish that your brand ads are unnecessary. It identifies a practical question: how much additional business does your advertising create, rather than receive credit for?

Why Strong Brand ROAS Can Hide Spending on Existing Demand

Someone searching for your company name already knows something about your business. They may have seen an earlier campaign, received a recommendation, or purchased from you before. A paid result can capture that visit without being the reason the person wanted to buy.

Suppose your brand campaign spends $2,000 and receives credit for 400 orders worth $100 each. Its reported return on ad spend, or ROAS, is 20: $40,000 in attributed revenue divided by $2,000 in advertising costs. Its attributed cost per order is only $5.

Those figures look attractive, but they do not answer what would happen without the campaign. Some customers might click your organic result instead. Others might use a saved link or return directly.

Attribution assigns credit to a marketing interaction. Incrementality measures the additional outcome caused by that interaction. A campaign can receive substantial credit while creating relatively little additional demand.

You can turn the disagreement between paid-search and SEO teams into a shared measurement question: “What happens to total business outcomes when we remove a defined portion of brand advertising?”

This question protects both teams from misleading conclusions. A drop in paid conversions does not automatically mean lost customers, and an increase in organic clicks does not automatically mean the ads had no value.

Build a Clear Organic Baseline for Brand Queries

Testing brand advertising without understanding organic coverage creates avoidable risk. Your homepage might rank well for the exact company name while performing poorly for a common misspelling or a brand-plus-product query. Mobile results may also present a different choice from desktop results.

For example, customers searching your name alone may find your homepage immediately. Customers searching your name plus “returns” may need a support page. Treating both searches as interchangeable can hide a useful reason for a particular advertisement.

SparkTraffic, a website traffic platform, provides a ctr calculation tool for calculating observed organic CTR and comparing it with position-based planning scenarios. It can help you make the search baseline explicit before you assess advertising overlap.

Organic click-through rate is search clicks divided by search impressions, multiplied by 100. Your inputs should use the same dates, query definitions, country, and device filters.

You can establish a brand-query dictionary that includes the company name, relevant variations, and common misspellings. Product names that have wider meanings need review rather than automatic inclusion.

Keeping those definitions fixed makes the comparison repeatable. Support queries, competitor comparisons, and brand-plus-category searches can remain separate where they represent different customer tasks.

The baseline should also record which page appears and what else occupies the search results. Strong organic visibility makes substitution plausible, but does not prove it.

A position-based CTR scenario cannot measure advertising incrementality. Its role here is to support the organic diagnosis, while the business test measures whether customers and contribution change.

Run a Limited Holdout Instead of Switching Off Brand Ads Everywhere

A company-wide pause can expose revenue before you know how customers will respond. A simple comparison with the previous week can also confuse advertising effects with seasonal demand, promotions, or ordinary variation.

Imagine that sales fall after you pause brand ads during a quieter trading period. Without a comparison group, your team may blame the pause for a decline that would have happened anyway. The opposite error occurs when a promotion hides genuinely lost sales.

A geographic holdout creates a group where the defined advertising is withheld while comparable regions retain it. The comparison gives you a better estimate of what would have happened under normal advertising conditions.

Match Regions and Keep the Test Scope Stable

Regions can differ in brand awareness, customer mix, and purchasing patterns. Comparing a large established market with a small new market would therefore produce a weak test, even if both use the same advertising settings.

You can select regions with comparable historical patterns and randomly assign matched regions to test and control groups where feasible. Checking their pre-test trends helps identify differences that would make the comparison unreliable.

The intervention should remove only the agreed brand advertising. Other campaigns, prices, landing pages, and promotions should remain stable enough to preserve the comparison.

Your team also needs to check whether other campaigns or accounts continue serving ads on the same brand queries. Otherwise, the supposed holdout may still receive the advertising you intended to remove.

Geographic spillover matters too. Customers may travel, search in one location, and purchase in another. Consistent targeting and an agreed method for assigning business outcomes to regions reduce this measurement problem.

Define Business Outcomes and Risk Limits Before Launch

A small pilot limits exposure, but it does not automatically produce a precise answer. If normal regional sales variation is larger than the effect you need to detect, a short test may leave you unable to distinguish useful advertising from unnecessary spending.

Your planning should assess whether the available regions, historical variation, and test duration can detect a commercially meaningful difference. Conversion delays also matter, especially when enquiries take weeks to become customers.

The measurement plan can include:

• Total orders or qualified customer outcomes across all acquisition channels.
• Contribution after relevant variable costs and advertising spending.
• Brand paid clicks, organic clicks, and direct visits as supporting diagnostics.
• Predefined limits for material business harm or tracking failure.

Paid-platform conversions should not be the primary outcome. Removing the ad necessarily removes many interactions that the platform could credit, even when the customer still purchases through another route.

Risk limits should protect the business, not provide a reason to stop whenever the result looks favorable. If the available data cannot support a useful test, that limitation should remain explicit.

Separate Click Replacement From the Budget Decision

Organic traffic recovery can make an advertising pause look successful before you examine the commercial effect. Some paid clicks may disappear because they represented low-value activity, while others may have belonged to profitable customers.

Suppose paid brand clicks fall by 1,000 and organic clicks rise by 850. The apparent click replacement is 85%. That is a useful diagnostic, but changing demand and other influences mean it is not, by itself, a causal estimate of preserved sales.

The following fictional example shows how the business calculation differs. It uses pooled test and control groups for simple arithmetic, not a recommendation to rely on a two-region test.

Adjust the Comparison Before Valuing Lost Orders

Before the test, both groups record 1,000 total orders in comparable periods. During the test, the control group falls to 980 orders, while the group without the defined brand ads records 970.

The control group’s 2% decline suggests a common demand change. Applying that change to the test group gives an illustrative expected result of 980 orders under continued advertising.

The test group records 10 fewer orders than that adjusted expectation:

Estimated additional orders from advertising = 980 expected orders − 970 observed orders = 10

This adjustment is a simplified estimate. A properly designed analysis also needs to account for regional variation, uncertainty, and whether the groups remained comparable.

Assume each genuinely additional order contributes $55 after product, payment, fulfillment, and other relevant variable costs. The estimated contribution lost through the pause is $550.

If the pause saves $2,000 in advertising costs, the estimated net contribution improvement is:

$2,000 saved advertising cost − $550 lost contribution = $1,450

That is a different conclusion from the campaign’s original $5 attributed cost per order. Dividing the same $2,000 spend by 10 estimated additional orders gives $200 per incremental order.

Compare Uncertainty With the Break-Even Threshold

A favorable point estimate does not settle the decision if the test remains imprecise. The data might still be compatible with a larger loss than the estimated 10 orders.

At $55 contribution per order, approximately 37 additional orders would cover $2,000 in advertising spending. If losing 40 orders remains plausible, the lost contribution would be $2,200, exceeding the saving by $200.

Your decision therefore depends on whether the evidence can distinguish the estimated loss from a loss large enough to make the pause unattractive. A statistically inconclusive result is not proof that the campaign has no effect.

You can define this commercial threshold before the test. It gives the measurement team a concrete question to resolve instead of asking for the smallest possible difference or the most reassuring result.

Retain Brand Advertising Where the Evidence Supports It

Brand advertising does not need one company-wide verdict. Different query groups can have different roles, and competitive conditions can change. A campaign that adds little value for the exact company name may still help customers find a relevant product or service page.

Competitor ads deserve attention, but their presence alone does not establish that every brand click requires protection. You still need evidence about customer outcomes and the cost of preserving them.

You can use test results to make narrower decisions: reduce coverage where organic results absorb demand without a meaningful commercial loss, retain advertising where additional contribution justifies the cost, and leave uncertain groups unchanged until better evidence is available.

Changes should also preserve a clear customer experience. Ads and organic results should accurately describe prices, availability, and destinations. Misleading urgency or exaggerated claims may increase clicks while attracting unsuitable customers and creating complaints.

Financial and health-related businesses need particular care with disclosures and claims. Privacy requirements also apply to the measurement process; aggregate regional outcomes do not justify collecting unnecessary personal data.

After a budget change, you can schedule a review around meaningful shifts such as competitor entry, weaker organic coverage, or a new product launch. An old result should not become a permanent rule after the conditions change.

The practical outcome is a joint SEO and paid-search decision based on additional business value. Your team no longer has to defend organic clicks against paid clicks. It can identify which advertising helps customers reach the business, which spending mainly changes the route they take, and where the next test would reduce uncertainty.