A company selling warehouse management software to third-party logistics operators doubled its paid search budget at the start of a financial year. The previous year's report showed search producing leads at a third of the cost of any other channel, so search was the obvious place for the new money. Nine months later, lead volume was up by a fifth, cost per lead had risen by half, and qualified opportunities from search were flat. The search terms report told the story. The new budget had gone to queries such as "warehouse jobs near me" and "free stock count spreadsheet", none of which came from a logistics operator with a buying process under way.

Nothing in that story was a mistake in the campaign settings. The company had already covered the relevant searches with the original budget. The extra money could only go to less relevant searches, and the bidding automation did what it was set up to do. This post explains what paid search is for, how the channel is bought, where search fits beside the other channel families, how to measure search, and the four ways search reporting overstates the channel. The series hub, How B2B Media Actually Works (and What It Cannot Do), sets out the buyer picture that this post relies on.

What Paid Search Is For

Paid search does demand capture. A buyer has recognised a problem, decided on a category, or already knows a company's name, and the buyer types those words into a search platform. Paid search places the company in front of that buyer at that moment. No other channel reaches a buyer at the instant the buyer describes a need in the buyer's own words.

Search reaches only buyers who are already looking, and that group is small. The Ehrenberg-Bass Institute's work on B2B buying puts the share of category buyers who are in market in any given period at around 5 per cent. The remaining 95 per cent have a working supplier and no live buying process, and are not searching for the category. Search volume for a category is therefore a hard ceiling on how much demand search can capture. A company can win a larger share of the searches that exist. A company cannot make more searches exist by spending more in the auction.

Within that narrow buyer state, search does two different jobs, and the two jobs need separate campaigns, budgets and reporting. Non-branded search captures buyers who are describing a problem or a category and have not yet chosen a supplier. Branded search defends the company's own name against competitors bidding on that name. Non-branded search is capture: the buyer might have gone to a competitor. Branded search is defence: the buyer was already looking for the company, and the question is whether the paid listing adds anything the organic listing would not have delivered.

How Paid Search Is Bought

Paid search is bought in an auction that runs every time a query is entered. The advertiser chooses the queries to appear against, sets a bid or a target, and writes the text ad. The platform ranks the competing ads by bid and by a relevance score, and the advertiser pays per click. Cost is driven by how many competitors want the same query and by how relevant the platform judges the ad and landing page. In B2B categories deal values are large, so category clicks can cost many times what a consumer advertiser pays.

Queries fall into four intent tiers. The value of a click to the business differs by tier. Problem queries describe a symptom without naming a category: "picking errors rising in our warehouse". The person searching may be a buyer or a student, and the query is cheap because few advertisers want it. Category queries name the type of solution: "warehouse management software for 3PL" (3PL meaning third-party logistics). The person searching is comparing options, and the query is expensive because every supplier in the category wants it. Brand queries name the company, and the person searching is already on the way to the company's website. Competitor queries name a rival. Competitor queries convert poorly because the buyer was looking for someone else, and the click is priced as though the buyer were looking for the advertiser.

Two mechanics of the auction explain why the warehouse software company's extra budget went to irrelevant queries. The first mechanic is broad matching: the platform extends the ad to queries the platform considers related, and as spend rises the platform has to find more queries to spend the budget on, so the related queries become progressively less related. The second mechanic is automated bidding: the advertiser sets a target cost per conversion, and the bidding system moves budget toward whichever queries hit that target. Cheap, ambiguous queries produce cheap form completions from people who are not buyers, so the bidding system learns to favour those queries. Both mechanics are useful at a modest budget, and both push spend toward cheaper, less relevant traffic as the budget grows. The course Media and Measurement covers how to keep the two mechanics within the relevant queries.

The search results page is also changing. The major search platforms now place generated answers above the results for many queries, and a growing share of searches end without any click. A buyer who asks "what should a WMS for a 3PL include" may read the generated answer and never see an ad. The effect on B2B categories is uneven, and the practical consequence is that the ceiling on search capture is lower than the raw query volume suggests. Our post on zero-click search in B2B marketing covers what that shift means for the wider search plan.

Search behaviour also differs by market. The dominant search platform is not the same in every country, category vocabulary changes between languages and sometimes between English-speaking markets, and in small markets the total volume of relevant category queries can be a few hundred a month. A search plan that copies the home market's query list into a new market through a translation tool captures the wrong queries at the wrong price.

Where Paid Search Fits in the Plan

Search harvests demand that other activity created. The buyer who types a category query heard about the category from a peer, or saw the company's advertising on a professional network months earlier. Gartner's research on the B2B buying journey shows buyers spending most of the journey away from suppliers, in independent research and in conversations inside the buying committee, and the search query is often the first moment that journey becomes visible to the company. Search belongs in the plan as the capture channel funded to cover the relevant queries, and no further.

In my experience, the right search budget is the amount needed to hold a competitive position on the category and problem queries that convert to opportunities, plus a tested branded defence budget. Once that amount is reached, any further budget should go to demand creation on professional networks, video or display, because demand creation grows the number of category searches available next year. A plan in which search takes more than about half of the media budget has usually stopped growing the pool search draws from.

Search also feeds the other channels. The search terms report is the only place a company can read buyers' problems in buyers' own words, and the words buyers use in those searches should be used in the messaging for social, video and display campaigns. A rise in branded search volume is one of the few leading indicators that demand creation is working, so branded query volume should be tracked as a measure of the creation channels, not claimed as an outcome of the search channel.

How to Measure Paid Search

Search is a capture channel, so search can fairly be held to conversion and pipeline metrics within the quarter. The measure for non-branded search is cost per qualified opportunity by query group, taken from the CRM. The table below is an illustrative worked example for the warehouse software company, with invented figures.

Illustrative figures, invented for the worked example. Lead to opportunity rates come from the CRM, matched back to the query group through the click identifier. The blended row is what a single "paid search" figure in a monthly report would show, and the blended row is dominated by brand queries that would mostly have converted through the organic listing.

The table shows why cost per lead misleads on search. Problem queries have the cheapest non-branded leads and an expensive cost per opportunity, because most people searching for a symptom are not buyers. Category queries have leads three times as expensive and opportunities at two thirds of the cost. Competitor queries are the worst source of pipeline by a wide margin. An optimisation that lowers the blended cost per lead will move budget from category queries to problem queries and make the pipeline result worse.

Branded search needs a different measure, because attributed cost per opportunity for branded search answers the wrong question. The question for branded search is how many of those conversions would have disappeared if the branded campaign had been paused. The most cited evidence on that question is a large-scale field experiment published in Econometrica in 2015 by Thomas Blake, Chris Nosko and Steven Tadelis. Working with a large online marketplace, the authors switched off paid search on the marketplace's own brand terms in a controlled way and measured the change in traffic and sales. Almost all of the traffic the branded ads had been credited with arrived anyway through the organic listing. The authors also tested non-branded search across regions and found a small positive effect, concentrated among people who had rarely or never used the marketplace, and no measurable effect on frequent users. The study is one company in one category, and a B2B company whose name competitors bid on may get a different result. The method transfers: pause branded search in some regions or on alternating weeks, hold the organic listing constant, and compare total branded conversions with the periods when the campaign ran.

Where the Numbers Lie

Search reporting is cleaner than the reporting for any other channel. A click and a form completion sit in a single tracked sequence, so the numbers look like proof of cause, and the business trusts search numbers more than the numbers deserve. Four patterns overstate the channel.

1. Branded and non-branded search are reported as one figure

A single "paid search" cost per lead blends a cheap, high-converting branded campaign with an expensive, lower-converting non-branded campaign, and the branded campaign drags the blended figure down. In the worked example above, the blended cost per opportunity is $1,500, and the non-branded cost per opportunity that budget decisions should rest on is above $3,000 for the best query group. A company that adds budget on the strength of the blended figure gets the non-branded result. Report branded and non-branded search as two channels.

2. Branded search claims conversions that would have happened anyway

A buyer who types the company's name into a search platform is already on the way to the website. If the paid listing sits above the organic listing, the buyer clicks the paid listing, the platform records a conversion, and last-click attribution gives the whole conversion to branded search. The organic listing would have taken the same buyer to the same page for nothing. A software company reporting 200 branded search opportunities a quarter may be adding twenty of those opportunities, or none, and the attributed figure cannot tell the difference. Run a pause test, use the incremental figure as the branded search result, and treat the rest of the branded conversions as demand the creation channels built.

3. Impression share and quality score are reported as outcomes

Impression share is the share of eligible auctions in which the ad appeared. Quality score is the platform's estimate of the ad's relevance. Both numbers are diagnostics for the person running the account, and both appear in monthly reports as results. A campaign can hold 90 per cent impression share on a query group that produces no opportunities. A report that celebrates a rise in impression share from 60 to 85 per cent is reporting that more money was spent in the auction. Keep impression share and quality score in the account manager's working file, and report opportunities by query group to the business.

4. Platform-modelled conversions are treated as CRM opportunities

The search platform's conversion column counts form completions inside the platform's attribution window, includes conversions the platform has modelled statistically where consent settings block tracking, and counts the same person twice if that person completed two forms. The CRM records one lead once and records whether the lead became an opportunity. When a report puts "conversions: 340" beside "cost per conversion: $210", the business reads 340 leads worth $210 each. The CRM may show 260 leads and 36 qualified opportunities. Use the platform's conversion column to compare campaigns within the search account, and use the CRM for any comparison between search and another channel.

What Paid Search Cannot Do

Paid search cannot reach the buyers who are not searching, and around 95 per cent of the market is not searching in any given period. Paid search cannot create the familiarity that makes a buyer type the company's name; that familiarity comes from creation channels and from reputation. Paid search cannot grow past the volume of relevant queries in the market. Paid search cannot reach the whole buying committee, because the finance lead and the operations director rarely search for the category, so a search-only plan speaks to one person in a group of six to ten. Paid search cannot fix a landing page that loses the click or a sales team that replies four days later, because search controls nothing after the click.

Paid Search FAQs

Should a B2B company bid on its own brand name?

Bid on the brand name when competitors are bidding on that name and a pause test shows the organic listing loses clicks to them. Do not bid on the brand name because the branded campaign reports a low cost per lead; that low cost per lead mostly counts buyers who were already on the way to the website. Revisit the pause test whenever a new competitor appears in the branded auction.

Why does paid search stop producing more leads when the budget goes up?

The volume of relevant searches in a category is fixed by how many buyers are in market. Once the campaign covers the relevant queries at a competitive position, extra budget goes to broader match queries and to higher bids on the same queries. Lead volume may rise because ambiguous queries produce cheap form completions, and qualified opportunities stay flat.

Is bidding on competitor names worth the cost in B2B?

Rarely, as a practitioner judgement. The buyer searching for a competitor's name wanted the competitor, the click is priced at category rates, and the conversion rate is low. Competitor bidding can make sense during a known event, such as a rival's price rise or product withdrawal, when that rival's customers are looking for alternatives. Outside an event of that kind, competitor queries are usually the most expensive source of pipeline in the account.

How do generated answers and zero-click results change B2B paid search?

Generated answers sit above the results for many informational queries, and a growing share of searches end without any click. Problem and early category queries are the most affected. The number of clicks available to capture falls, the ceiling on search capture falls with it, and the case for demand creation channels that build familiarity before the search grows stronger.

What is the right measure for a B2B paid search campaign?

For non-branded search, cost per qualified opportunity by query group, with the opportunity taken from the CRM. For branded search, the incremental conversions shown by a pause test. Cost per click, cost per lead, impression share and quality score are working metrics for the person running the account, and none of those four metrics shows whether the channel produced pipeline.

Go Deeper

This post carries the judgement: what search is for, what each query tier is worth, and where search reporting overstates the channel. The method sits in the course Media and Measurement, which covers how to structure a search account by intent tier, how to join query groups to CRM opportunities, and how to design a branded search pause test. The free module in B2B Marketing Fundamentals covers the ideal customer profile and the offer, both of which search inherits. Why a last-click record of a branded search conversion says nothing about cause is set out in why marketing attribution misleads.

Sources

• Ehrenberg-Bass Institute for Marketing Science, 2021, How B2B Brands Grow: the finding that around 95 per cent of category buyers are out of market in any given period, with the in-market share varying by purchase cycle length.
• Gartner, The B2B Buying Journey: typical buying groups of 6 to 10 decision makers, a non-linear journey, and most buying time spent in independent research rather than with suppliers.
• Thomas Blake, Chris Nosko and Steven Tadelis, Consumer Heterogeneity and Paid Search Effectiveness: A Large-Scale Field Experiment, Econometrica, 2015: controlled tests at a large online marketplace showing near-zero incremental effect from branded search ads and a small non-branded effect concentrated among infrequent users.