How to test whether your branded search ads are incremental

Branded search ads usually have the best cost per acquisition in a B2B SaaS account. That is also why they are hard to judge. Someone who types your company name into Google already knows you, and many of them would click your organic listing if the ad were not there. The real question is how many branded conversions would happen anyway.

You answer it with a controlled test: switch paid brand off for set periods, keep everything else stable, and compare total conversions from all channels in that market against periods when brand is on. If total conversions barely move, most of your brand spend is buying clicks you would have had for free. If they drop, paid brand is doing real work, often because competitors or partners would take the top of the page without it.

Why brand CPA looks so good

Google Ads credits a conversion to the ad that was clicked. On a brand campaign, the person clicking is often a returning visitor, an existing trialist or a buyer who first heard of you through a colleague, a review site or a LinkedIn ad. The ad was the last step on a path that other work built.

So brand CPA is low by construction. It measures how cheaply you collect demand that already exists. How much of that demand you would lose without the ad is a separate question, and only a test answers it.

The number also flatters blended results. A healthy brand CPA can hide a weak non-brand programme and makes brand look like the obvious place for extra budget.

Why brand costs rise

Brand cost per click is normally low because your own site is the most relevant result for your own name. It rises when someone else enters the auction. Two groups do this.

Competitors. Google's trademark policy does not restrict the use of trademarks as keywords, so a competitor can bid on your brand name. Trademarks in ad text can be restricted after the owner files a complaint, which limits what their ad can say but not whether it shows.

Channel partners and resellers. Google's policy allows resellers to use a trademark when their landing page clearly facilitates buying that product. For software sold through partners, this means your own resellers can compete with you on your name. Branded costs can rise for this reason alone, with no change in competitor activity.

Rising brand cost is a good trigger for a test. If the pressure comes from partners, a clause in the partner agreement about bidding on your brand may solve it more cheaply than a bigger brand budget. If it comes from competitors, the test tells you what it costs to leave the space open.

Test designs compared

There are four practical ways to run the test. The right one depends on your volume, how many markets you sell in and whether you can split a market into regions.

For most mid-sized B2B SaaS accounts, a switchback or matched-market test is the one you can actually run.

The switchback design in detail

We use a pulse switchback when a market cannot be split cleanly. The market becomes its own control. Here is the design we use:

1. Blocks of 14 days off and 14 days on, repeated for three cycles. Two weeks gives B2B buyers time to convert after searching; three cycles guard against one unusual fortnight deciding the result.
2. Every block runs Monday to Sunday. B2B conversions follow a strong weekday pattern, so each block must contain the same mix of days.
3. Avoid seasonal months. Schedule around summer holidays and plan the final read before the December dip. A block that falls in a quiet month will look like a loss of conversions whatever brand is doing.
4. Primary metric: total primary conversions from all channels in the market, in event time. For most SaaS companies that means trial or demo requests, counted on the day they happened, from GA4 or the CRM. Google Ads conversions cannot be the metric because they drop to zero when brand is off and are reported against the click date.
5. Read each off block against the on blocks either side. Averaging the neighbouring on blocks controls for gradual trends such as growth or a slow season starting.
6. Treat the read as directional at modest volumes. At a few hundred conversions a month, no single two-week block will be statistically significant. The repeats show whether the direction is consistent, and three off blocks pointing the same way are a usable answer.

Freeze non-brand budgets, bid strategies, landing pages and large paid social pushes during the test. If a change cannot wait, log the date.

What to watch during the test

Three signals tell you what is happening while brand is off.

• Organic brand clicks in Google Search Console. If paid brand clicks mostly move to your organic listing, organic brand clicks should rise during off blocks. This is the clearest sign that paid brand was collecting existing demand.
• Competitor impression share on your brand terms. Auction insights for Search shows impression share, position above rate and absolute top of page rate for each competitor. Check your brand keywords weekly and compare competitors' presence across on and off blocks.
• Total conversions by channel in event time. Watch organic and direct in GA4 alongside the total. A fall in paid search that is matched by a rise in organic suggests a transfer. A fall in the total suggests real loss.

If a competitor or reseller takes the top position during an off block and total conversions fall, paid brand is defending that position.

How to act on the result

Use a simple decision rule once the three cycles are complete.

Agree the decision rule with finance or leadership before the test starts. Whatever you choose, keep reporting brand and non-brand separately, so the brand CPA never masks the rest of the account. For how the two measurement systems differ during a test like this, see why GA4 and Google Ads never match.

Running the test across several European markets

Brand incrementality is local. A company that is well known in the Netherlands and new in Poland will see very different results, because organic brand rankings, competitor activity and partner networks differ by country.

• Test one market at a time and start where volume is high enough to read.
• Do not carry a result from a mature market into a launch market. In a new market, brand volume is small, and a competitor bidding on your name can take a large share of it. We keep brand running from day one when launching paid search in a new European market for this reason.
• Check local-language brand variants and misspellings. Paused brand should mean all brand keywords in that market, including local variants.
• If a matched-market test is possible, pick markets that match on brand maturity as well as size.

Before you start: a checklist

1. Confirm the primary conversion and where you will count it (GA4 or CRM, event time).
2. Pull at least three months of baseline daily conversions for the market.
3. Map the calendar: avoid holiday months, trade shows and product launches.
4. Record current competitor and reseller impression share on brand terms.
5. Write the decision rule down and get sign-off.
6. Schedule the pauses in advance so nobody switches brand back on early.

Brand testing is one of the first measurement projects we run on a new account, because the answer changes how much budget is left for non-brand and competitor campaigns. If you want help designing one, see our measurement services and paid search services.