Google Ads for B2B startups: is search worth it before buyers search for you?

1. Blog
2. Demand & Channels

By Saroj Jha · November 2, 2026 · 15 min read

Google search is worth testing for a B2B startup only when enough buyers already search for the problem you solve. Search captures demand; it does not create it. So before spending, I check monthly searches for the problem in Google’s own Keyword Planner, work out how many clicks the test needs from cost per click and conversion rate, and ask whether those searches can supply that many clicks in about ten weeks. If they can, I run exact and phrase match on non-brand problem terms, feed qualified opportunities from the CRM back into Google, and stop on a rule written before launch. If they cannot, the budget belongs in creating demand first.

Here is how I would decide, using one fictional company from start to finish.

Running case: Example Co. Fictional

Example Co. is a fictional seed-stage B2B SaaS company that tracks vendor-contract renewals for procurement teams at companies with 200 to 2,000 employees. Its annual contract value is $12,000, gross margin is 80%, a qualified opportunity becomes a customer about one time in four, and the sales cycle runs about 60 days. The founder has a test budget of around $10,000 and one question: should Google search be a channel this year? Every Example Co. number in this post is fictional and its arithmetic is shown.

Is anyone searching for what you sell yet?

Search ads answer a query someone has already typed. That makes them demand capture, and I covered the difference between capturing and creating demand in Demand Creation vs. Demand Capture. The practical consequence for this post is narrow: the size of a search test is capped by how many people search for your problem each month, not by your budget.

New categories hit this cap first. If your buyers do not yet know a product like yours exists, they search for the symptom (“contract auto-renewed without notice”) or the old workaround (“renewal tracker spreadsheet”), not for your category name. Those symptom and workaround terms are where an early B2B search test lives or dies.

Google’s Keyword Planner is the place to check, because it is Google’s own estimate. Google says the tool lets you view estimates of the number of searches a keyword gets each month and the average cost for your ad to show on those searches, and that keywords with very low search volume are not discoverable or forecastable^[4]. I read the Keyword Planner page on 1 November 2026. I treat its numbers as a ceiling to plan against, not a promise.

To turn searches into a click ceiling, I multiply monthly searches by the share of those auctions I expect to appear in and by the click-through rate I expect when I do. Both assumptions are my rule of thumb for a first test, not a benchmark, and the test itself will replace them.

Example Co., step 1 Fictional

Keyword Planner shows about 1,900 searches a month for Example Co.’s category terms and 12,100 for adjacent problem and workaround terms, 14,000 in all. Assuming the ads appear in half of those auctions and 5% of viewers click: category terms alone give 1,900 × 0.50 × 0.05 = 47.5 clicks a month; all terms give 14,000 × 0.50 × 0.05 = 350 clicks a month.

Do this

In Keyword Planner, list the problem, symptom and workaround phrases your customers used in sales calls, not only your category name, and record monthly searches and the average cost for each.

How much budget does a search test need?

A test needs enough conversions to tell you something. Google’s own guidance gives a useful floor: for Target CPA, Google recommends measuring performance over the last 30 days, including at least 30 conversions^[3]. I borrow that 30 as the size of a first test, even though at this stage I bid manually or with Maximize clicks rather than Target CPA.

The arithmetic has three inputs. Clicks needed equal the conversions you want divided by the landing-page conversion rate, spend equals clicks times cost per click, and duration equals clicks needed divided by the click ceiling from the previous section. If duration runs past about three months, my rule of thumb, not a benchmark, is that the test is too slow to make a decision this year.

The trap is to set the budget first and the conversion count second. A $3,000 budget at an $11 cost per click buys about 272 clicks, which at a 3.5% conversion rate is fewer than ten demo requests. Ten conversions cannot separate a working channel from a bad week.

Example Co., step 2 Fictional

Target: 30 demo requests. Expected landing-page conversion rate 3.5% and average cost per click $11. Clicks needed: 30 ÷ 0.035 = 857.1, rounded up to 858. Spend: 858 × $11 = $9,438. At 47.5 clicks a month from category terms alone, 858 clicks take 858 ÷ 47.5 = 18.1 months; at 350 clicks a month from all terms they take 858 ÷ 350 = 2.45 months, about 75 days or 10.6 weeks. Daily budget: $9,438 ÷ 75 = $125.84.

Do this

Before launch, write down the conversions you want, your expected conversion rate and cost per click, and compute clicks, spend and weeks. If the weeks exceed about 13, widen the terms or put the money into demand creation.

Should you bid on your brand, generic terms or competitors’ names?

Search terms fall into three groups with different economics. Non-brand terms (the problem, category and workaround) are where new buyers are found. Brand terms are people already looking for you. Competitor terms are people looking for someone else.

Brand terms deserve more doubt than they get, because they look like the best campaign in the account. In a large field experiment at eBay, published in Econometrica in 2015, brand-keyword ads had no measurable short-term benefit: when eBay stopped them on Yahoo and Microsoft, only 0.529% of click traffic was lost and 99.5% was kept through organic links^[1]. The same paper’s experiment on non-brand ads, run by switching them off in a random 30% of US metro areas, found that new and infrequent users responded to ads, while frequent users who would have bought anyway accounted for most of the spending^[1].

Competition changes the brand answer. A study of thousands of brands in randomized experiments on Bing in January 2014, published in Marketing Science in 2018, found that brand ads added 1% to 4% of clicks when no competitor bid, but competitors took 18% to 42% of clicks from brands that faced competition and chose not to advertise^[2]. In my view the rule follows directly: bid on your own name when a competitor shows up on it, and not before. A startup with little brand search usually has nothing to defend yet.

Google gives you two controls here. Brand inclusions limit a Search campaign to queries associated with the brands you select, and brand exclusions stop a campaign serving on queries associated with a brand you want to avoid^[5]. I use an exclusion of my own brand in non-brand campaigns, so they cannot take credit for people who already knew my name.

Competitor terms are allowed as keywords. Google’s trademark policy says it will not restrict trademarks used as keywords, while restrictions apply to trademarks used in the ad text itself^[6]. That is Google’s policy, not a statement of trademark law, so I keep competitors’ names out of ad copy and check with counsel before any comparative claim. In practice competitor clicks tend to convert worse, because the searcher wanted someone else.

Example Co., step 3 Fictional

Brand searches are about 90 a month and no competitor bids on the name, so Example Co. runs no brand campaign. Competitor-name terms cost about $14 a click and convert at 1.5%: $14 ÷ 0.015 = $933.33 a demo request, against $11 ÷ 0.035 = $314.29 for non-brand terms, 2.97 times as much. Competitor terms wait until non-brand terms work.

Do this

Search your own brand name on Google once a week. If competitors appear, open a small brand campaign; if not, add your brand as an exclusion to the non-brand campaigns.

Which match types and settings should a small B2B account use?

Google defines three match types. Broad match, the default, may show ads on searches related to your keyword; phrase match on searches that include the meaning of your keyword; exact match on searches with the same meaning or intent as the keyword^[7]. Google also says it is critical to use Smart Bidding with broad match^[7], and that is my reason to hold broad match back at the start. Smart Bidding learns from conversions, and a new B2B account has almost none.

So I start with exact and phrase match on the problem terms from step 1, and I read the search terms report every week. Google describes it as showing the actual terms a significant number of people searched for that triggered your ad, and negative keywords added from it default to exact match^[8]. In B2B the report is where you find job seekers, students and people searching for a free template, and you remove them as negatives.

Two newer features need a decision rather than a default.

AI Max for Search campaigns is Google’s suite of targeting and creative enhancements: search term matching, text customization and final URL expansion, which Google says reaches more queries through broad match, asset-based targeting and keywordless technology^[9]. Search partners are included by default when you create a Search campaign, and you can opt out in the campaign’s network settings^[10].

Both widen reach, and both make a small test harder to read, so I leave them off until the core terms convert and then test each one separately. I read these Google Ads Help pages on 1 November 2026, and the features change often, so check them again before you build.

Example Co., step 4 Fictional

Example Co. launches one Search campaign with two ad groups (category terms and workaround terms), exact and phrase match only, search partners off, AI Max off. The first week’s search terms report shows “contract manager jobs” and “free renewal tracker template”; both go in as negatives.

Do this

Put a 15-minute weekly slot in the calendar to read the search terms report and add negatives, and keep a log of each change in your test log.

How do you get CRM pipeline back into Google Ads?

A demo request is not revenue. If Google only sees form fills, it will learn to find people who fill forms, including students and competitors. The fix is to send later stages from the CRM back to Google Ads.

Google’s offline conversion import does this. Google assigns a unique ID, the Google Click ID (GCLID), to each ad click, and offline conversion import lets you record what happens offline after that click^[11].

For advertisers new to it, Google recommends starting with enhanced conversions for leads instead^[11], which Google describes as an upgraded version of offline conversion import that uses user-provided data such as email addresses, hashed and matched to signed-in Google accounts^[12]. Google advises including GCLIDs with uploaded events whenever possible^[12]. Google also says that from 15 June 2026, API uploads of both moved to its Data Manager API^[11], so check which path your CRM connector uses.

The conversion window matters as much as the import. Google’s default click-through conversion window is 30 days and can be set from 1 to 90 days depending on the conversion source; conversions after the window are not recorded^[13]. A B2B deal that qualifies on day 40 is invisible to a 30-day window. I set the window to cover the time from click to qualified opportunity, and I keep the setup of GA4 and UTMs in Tracking Setup for Startups.

Example Co., step 5 Fictional

Example Co. stores the GCLID and email with each demo request in its CRM and uploads two conversion actions: “qualified opportunity” (usually about two weeks after the click) and “closed won” (about 60 days). It sets the conversion window for both to 90 days, because 60 days of sales cycle would overrun the 30-day default.

Do this

Add a hidden GCLID field to your demo form, map it to the CRM contact, and confirm one test lead arrives in Google Ads as a qualified-opportunity conversion before you spend the budget.

When should you stop, and what counts as success?

Write the stop rule before the first click, in money, not clicks. My rule of thumb, not a benchmark, starts from what a customer can be worth: the maximum acquisition cost I accept is one year of gross profit. Divide that by the number of qualified opportunities it takes to win one customer and you get the most you can pay per qualified opportunity. Divide the test spend by that number and you get the minimum count of qualified opportunities the test must produce.

Then judge only at the end. Search tests are noisy in their first weeks, and early pauses mostly measure luck. At the end there are three outcomes: pass (enough qualified opportunities under the ceiling), change (enough demo requests but too few qualify, which points at the terms or the offer), or stop (too few demo requests, which usually means the demand is not there yet).

A pass does not prove the ads caused those deals. Some buyers would have found you anyway, as the eBay brand result shows^[1]. When search becomes a material share of spend, test incrementality with a holdout, which I cover in Incrementality Testing for Startups.

Example Co., step 6 Fictional

Ceiling: one year of gross profit is $12,000 × 0.80 = $9,600; at one win in four qualified opportunities, the most Example Co. can pay per qualified opportunity is $9,600 × 0.25 = $2,400. Minimum to pass: $9,438 ÷ $2,400 = 3.93, so 4 qualified opportunities. Result after 858 clicks: 26 demo requests (3.0%) and 9 qualified opportunities, $9,438 ÷ 9 = $1,048.67 each. It passes, and the expected 9 × 0.25 = 2.25 customers imply a cost of about $4,194.67 each, to be confirmed when the deals close.

Do this

Put the ceiling, the minimum count and the end date in your test log before launch, and agree with your cofounder that nobody changes them mid-test.

What changes by stage?

• Pre-seed — Run the Keyword Planner check and the budget arithmetic, and usually stop there. Problem searches are often too thin to fill a test, and the money does more in founder-led outreach and content that creates demand.
• Seed — Run one non-brand Search test with exact and phrase match, CRM import working before launch, and a written stop rule. Add a brand campaign only when a competitor appears on your name.
• Series A–B — Once qualified-opportunity conversions arrive steadily, test broad match with Smart Bidding and AI Max as separate experiments, add competitor terms, and measure search with a holdout before raising its share of budget.

Free tool

Size the search test before you spend

The Channel Test Budget Calculator takes your target conversions, conversion rate and cost per click and returns the clicks, spend and weeks a test needs, the same arithmetic as step 2 with your own numbers.

If search passes, the next questions are how much of your paid budget it deserves and how it sits beside other channels; Which marketing channels should a startup start with? and How Early-Stage Startups Should Use Digital Ads cover those. If your buyers are better reached by job title than by query, LinkedIn Ads for B2B Startups is the companion to this post.

If you want me to look at your own funnel, positioning and channels, my Growth Audit ends with three prioritized fixes and a 30-day action plan.