The State of B2B Paid Advertising

What 379 B2B advertisers can teach us about where spend is going, how paid media is actually being used, and what "good" really looks like.

Most B2B marketers have a benchmarking problem. We know our own cost per lead. We know what LinkedIn says our campaigns generated. We know how much of the budget Google consumed last quarter. What we usually don't know is whether any of it is normal.

Should an ABM company put 30% of its paid budget into LinkedIn? Is Meta becoming a serious B2B acquisition channel, or is its growth concentrated among a few advertisers? If your website CPL is $500, is that weak — or efficient once you account for how many cheaper leads fall outside your ICP? And perhaps most importantly: are the numbers in the ad platforms even close to reality?

We grouped advertisers two ways — buyer archetype (who they sell to) and go-to-market motion (how they sell).

GTM motion is assigned from each company's website.

That distinction turns out to matter a lot. One of the clearest findings in the data is that if you're trying to figure out what your paid strategy should look like, HOW you sell may be more useful than WHO you sell to.

1. Your GTM motion is one of the strongest predictors of your media mix

Marketers naturally benchmark themselves against their competition — companies going after similar buyers. That makes intuitive sense, but it isn't the strongest pattern in the data. Go-to-market motion explained much more of the difference in channel allocation than buyer archetype did.

ABM advertisers put a typical 30% of paid budget into LinkedIn, with a median Meta allocation of effectively zero. PLG companies look very different — a typical 12% on LinkedIn and 6.5% on Meta. Demand-generation companies are more Google-concentrated, at a typical 62%.

The difference remains within the same buyer category: in nine of the eleven archetypes with enough data to compare, ABM advertisers devote a larger LinkedIn share than non-ABM advertisers selling to the same buyer — an average of 13.5 percentage points more LinkedIn when you're ABM.

But don't turn the rule into dogma

There are meaningful exceptions — Finance & Accounting and Sales & RevOps. In those buyer groups the ABM mix is less LinkedIn-heavy and closer to non-ABM peers. The better rule: start with your GTM motion, then check whether your specific buyer behaves differently.

2. As companies grow, paid investment doesn’t just rise — it jumps

It makes sense that companies invest more in paid media as their revenue increases. What is more interesting is how sharply that investment steps up.

Among advertisers matched to estimated company revenue, companies in the $1–10 million revenue band spent a median of approximately $5,400 per month. That increased to $14,000 among companies with $10–50 million in revenue — and then jumped to approximately $63,000 among companies above $50 million. In other words, companies with more than $50 million in revenue invested nearly 12 times as much per month as companies in the $1–10 million band.

Companies do not simply add more budget to the same channel mix as they grow — they expand the mix itself. Among companies with $1–10 million in revenue, 59% advertised on only one of the three platforms we track. Above $50 million, that fell to 22%. Interestingly, Meta is most often the second platform advertisers add.

As the company matures, investment increases, more platforms come online, and the paid operation becomes more complex. That progression makes intuitive sense: larger companies have more budget to test new platforms and absorb the learning costs that come with building a new paid channel.

3. AI-native companies allocate less to demand capture — but the pattern is concentrated

AI-native advertisers appear to use paid media differently. Across the three major platforms they allocated just 38% of pooled spend to Google, versus 69% among other advertisers. LinkedIn instead received 43% of AI-native spend — their largest platform by pooled dollars — versus just 13% for everyone else.

Note: AI-native = companies whose core product is generative AI (28 of 379), classified by us from company identity — not companies that merely have AI features.

On Meta, AI-native advertisers directed 34% of pooled spend to traffic, awareness and engagement objectives, versus 8% among other advertisers. A similar upper-funnel tilt appears on LinkedIn, where engagement accounts for 43% of pooled AI-native spend while only 1% goes to website-conversion campaigns.

Many AI-native companies are still establishing new categories. Their LinkedIn and Meta spend looks less like demand capture and more like market education — building a category narrative before buyers are ready to convert.

4. Meta is gaining B2B budget — and marketers fund it as a direct-response channel

Among advertisers we could observe consistently over time, the average share of paid budget on Meta increased from 20.8% to 24.9%, and adoption from 54% to 63%. The movement was lopsided: 42 advertisers shifted more than ten points toward Meta, 19 away. Google moved the other way — average share fell from 43.6% to 40.9%. LinkedIn was roughly flat.

This isn't a quirk of composition or budget size: in a balanced panel, Meta gained ~5.4 points among ABM advertisers and ~5.5 among PLG, while Google declined in both. Meta is taking a larger share of the existing wallet, primarily at Google's expense.

How advertisers use that growing Meta budget matters. The old stereotype casts Meta as cheap impressions and retargeting — but 89% of Meta spend goes toward lead and sales objectives. Traffic and awareness are common experiments, but attract little money. The gap between what marketers test and what they fund is the story.

The shift is not universal — Demand Gen advertisers and companies selling to Security buyers did not show the same movement toward Meta.

5. Your feeling is right – media costs are mostly rising

Everyone feels like serving ads is getting more expensive, and they're not wrong. We hypothesize it's because outbound has decreased in effectiveness and AI answers are eating into search. (More on that in this article.) How much costs grow varies by channel and GTM motion.

ABM advertisers saw broad but more moderate increases: Google CPM rose 20% and LinkedIn CPC 18%, while their cost per Meta click was roughly flat (−3%).

PLG advertisers saw some of the steepest increases. Their LinkedIn CPM rose 43% and their Meta CPM 35%, while Google CPM barely moved (+2%). In other words, PLG advertisers are paying noticeably more on Meta — the platform where they have been putting more of their budget.

Demand Gen advertisers saw a split on LinkedIn: CPM dipped slightly (−6%), but the cost of a click rose 42%.

These differences could reflect changes in audiences, campaign objectives, or ad formats — not simply higher platform prices. But the practical takeaway is clear: there is no single cost trend across B2B advertising. What is becoming more expensive depends partly on how a company goes to market.

Change over time is only half the cost story. The platforms also begin from very different price levels.

LinkedIn costs much more to reach the same number of people

LinkedIn costs remain much higher than the alternatives. Within the same advertiser, LinkedIn CPM is typically 5.3× Meta’s and 8.3× the cost of Google’s impression-based formats (aka display). Advertisers pay a substantial premium for its professional audience and targeting.

To be clear, we’re not advocating going and buying Google Display ads. (If you want to know why Display so often fails to drive good results, read more here.)

6. Google has a playbook. Meta has a playbook. LinkedIn has several.

Across Google the prevailing strategy is remarkably consistent — Search dominates. Of 259 Google advertisers, 246 run Search and the typical advertiser puts 93.5% of Google spend there. Meta has one dominant spending pattern too: many objectives are used, but when real money is committed, leads and sales dominate.

LinkedIn is where it gets interesting — there is no single LinkedIn playbook. The right strategy changes materially with both how you sell and who you sell to. For PLG advertisers, 57% of LinkedIn spend goes to direct-response (Lead Gen + website conversion). For ABM, it flips: 51% goes to upper-funnel — engagement, brand, video.

7. Website analytics dramatically understate LinkedIn's lead output

Judge paid channels entirely through website activity and LinkedIn looks surprisingly small. As a referring source it shows ~243,000 revealed visitors, versus nearly 1 million for Meta and 8.6 million for Google. But LinkedIn has a structural difference: a significant share of its conversion happens without the person ever reaching your website, through native Lead Gen Forms.

8. Ad platform-reported conversions and site analytics confirmed conversions are way off from each other

There's distortion in the opposite direction too. Comparing platform-reported conversions with those we can independently confirm (website fills or LinkedIn native forms), the numbers diverge sharply:

Based on the advertiser median.

It's tempting to call all of that "inflation." That's too simplistic. Platforms and independent web measurement operate under different attribution regimes — view-through conversions, longer windows, modeled outcomes. The lesson isn't that one system is true and the other fraudulent; it's that marketers routinely place unlike numbers next to each other and call them the same thing.

9. Attribution methodology seriously changes how channels could be judged

Attribution rules can change the channel story even before platform-reported and confirmed numbers are compared. Across confirmed website fills (April–August), first- and last-touch were nearly identical for Google but diverged for Meta and LinkedIn.

Meta receives ~8% less credit under last touch than first; LinkedIn ~11% less — both appear earlier in converting journeys more often than Google. Any-touch widens the gap: Google rises ~1.1× vs last touch, Meta ~1.6×, LinkedIn ~7×. Those are overlapping path counts, not additive credit. The implication isn't that LinkedIn caused seven times more conversions — it's that last-touch reports miss much of LinkedIn's presence, and a meaningful share of Meta's, earlier in the path.

10. A channel’s apparent CPL depends on which conversions count

The cheapest channel for leads changes depending on whose conversions you count

In the platform dashboards, Meta looks dramatically cheaper than Google: its reported CPL is roughly one-third of Google’s. But when the comparison uses independently confirmed conversions, Meta becomes at least 1.6× more expensive than Google, depending on the attribution model.

The exact difference changes under first-, last- and any-touch attribution, but the ranking does not. LinkedIn remains the most expensive of the three under both measurement approaches.

We should note that while the median cost of a Google click rose by roughly 11%, Google’s platform-reported cost per conversion increased by approximately 27%. Advertisers are paying more for each visit, and those visits are converting less often.

This is based on Google’s own conversion reporting, so it should be treated as a directional measure of changing efficiency — not a confirmed cost per lead.

Top-level CPL is only the beginning of the economics

CPL quietly assumes every lead is equally valuable. A lead from a 20-person company when your product starts at $100,000 a year isn't equivalent to a lead inside your ICP. So we pair CPL with a more useful benchmark:

Website Known Qualified CPL

Website ICP fit by buyer archetype

Every archetype fit figure is highly concentrated in one advertiser, so the pooled rate can mislead in either direction. This table pulls the outlier out: it shows the pooled rate, the top advertiser's share of the archetype's evaluable leads, and the rate with that advertiser removed.

11. The closer you get to the question marketers actually care about, the harder measurement becomes

Eventually someone asks the only question that matters: did the advertising cause any of this to happen? That's incrementality. Where form-fill lift was measurable, exposed populations converted at roughly 4–6× the rate of comparison populations, with essentially no statistically significant negative lift.

The harder finding is how rarely B2B programs generate enough conversions to estimate lift reliably. Across the incrementality cohort, comparison groups produced fewer than one form submission per 100,000 people. Of the 140 advertisers with incrementality evidence, 92 lacked enough comparison-group submissions at 90 days to produce a defined form-fill lift result. That doesn't mean their advertising didn't work — the measurement is underpowered.

12. ABM & Demand Gen teams should use website engagement as an earlier incrementality signal

If you sell a product with a nine-month sales cycle, waiting for enough demo requests to prove causality isn't useful. Website visits happen far more often than form submissions. Among audiences where submission lift wasn't yet measurable, many already had enough activity to calculate revealed-visit lift — and where both could be measured, their direction agreed 92% of the time. That makes visit lift an earlier indicator: not proof of revenue, but evidence that advertising is changing the behavior of the population you're targeting.

So what does "normal" actually look like?

The most useful benchmark is rarely "the average B2B advertiser." A PLG company selling to developers shouldn't benchmark against an enterprise ABM company selling to compliance teams. So the rest of this report is designed as a lookup table: start with how you sell, then who you sell to, then compare channel economics where the data supports it.

Benchmark explorer

Filter on two dimensions — GTM motion and buyer archetype — to assemble a paid-media benchmark for companies like yours. Motion × archetype cohorts below the 8-advertiser floor are suppressed. CPL figures (†) are within-cohort medians in mixed account currencies — read as context, never ranked across cohorts.

The five questions we'd use to benchmark a B2B paid program

There's no single number that tells you whether your paid strategy is good. But the data suggests a useful sequence.

• 1. Are we comparing ourselves to companies with the same GTM motion? Start there before comparing channels or CPL.
• 2. Is our channel mix unusual for companies like us? Being an outlier isn't inherently bad — but you should know that you're one.
• 3. Are we using each platform for the same job our peers are? Especially on LinkedIn, where objective mix changes substantially by motion and buyer.
• 4. What are we paying for a lead we'd actually want? Track ordinary CPL, but pair it with Website Known Qualified CPL where identity and ICP coverage support it.
• 5. Are we measuring attribution — or causality? Platform reporting, confirmed conversions and incrementality answer different questions. Use each for what it can actually answer.

Methodology and important caveats

• This benchmark includes 379 Primer customers that spent on Google, LinkedIn or Meta during the 12 months through August 2026. It describes this observed panel — not a statistically representative census of every B2B advertiser.
• Spend exists in advertisers' account currencies, so cost comparisons are constructed only where currency differences do not invalidate the comparison.
• GTM motion and buyer archetype are separate classifications and are never substituted for each other.
• ICP fit reflects only the testable firmographic portion of each advertiser's own ICP definition — company size, geography, industry and revenue. It is not a universal Primer definition of a "good lead," and does not evaluate every persona, technology, keyword or intent requirement.
• Website Known Qualified CPL is explicitly website-only. Native platform conversions that can't be evaluated on the same basis are excluded — so LinkedIn deserves special caution, since a substantial share of its confirmed conversions occur through native Lead Gen Forms.
• Platform-reported and independently confirmed conversions operate under different attribution methodologies; differences shouldn't automatically be read as false reporting.
• Incrementality results are observational rather than randomized experiments, and are available for only 140 of 379 advertisers — generally the largest, most instrumented accounts.

The purpose isn't to tell every marketer to copy the median. It's to give you something better than a vague industry average: a way to understand what companies that actually look like yours are doing — and enough context to know when following them would be a mistake.