Why Customer Acquisition Cost Keeps Climbing: 5 Overlooked Structural Causes

Summary: Rising acquisition cost is not all about pricier traffic. Shrinking bid pools, lost attribution signals, longer decision chains, weak follow-up capacity, and unaccounted tooling cost are five more common and more controllable structural leaks—each with a completely different signal and first move.

"Traffic keeps getting more expensive" has become the conclusion of nearly every growth meeting, but blaming the whole cost increase on traffic price hides the part you can actually control. According to iResearch's "2026 China Digital Marketing Trends Report," the average enterprise customer acquisition cost reached about 486 RMB per person in 2026, up about 23.5% YoY and about 60% cumulatively over the past five years; QuestMobile data shows the combined acquisition cost across major e-commerce platforms and content communities rose from about 380 RMB per person in 2023 to about 620 RMB per person in Q1 2026. This article breaks down five easily overlooked structural causes and gives each one's identifying signal and first move—their cost mechanisms are entirely different, and handling them together is often the root of "the more you optimize, the more expensive it gets."

1. The takeaway first: what's expensive is not traffic itself, but five structural leaks

Split acquisition cost into two parts: the cost of getting traffic, and the conversion cost once traffic arrives. Most teams watch only the first, yet the part actually rising fast is often the second—or is not in any real business step at all, but a measurement artifact.

Of the five structural causes, only the first relates directly to traffic price; the other four belong respectively to measurement distortion, process change, follow-up capacity, and cost-accounting scope. This means that even if traffic price stops rising, most enterprises' acquisition cost can still keep climbing.

2. Cause 1: the bid pool is shrinking, forcing paid spend into a narrower band

Generative answers absorb a large share of top-funnel queries that would otherwise flow to websites. Informational questions get resolved right on the results page, so paid ads can only compete in a narrower band of high-intent keywords, while the number of bidders has not fallen—pushing up cost per click.

iResearch shows the average Baidu bid lead cost across industries rose about 15% to 20% versus 2025, one named reason being that after generative search went mainstream, about 36.2% of users only look at the AI answer box and no longer scroll down to ads. Separate statistics show the cost per thousand impressions across short-video and search platforms rose about 122% versus the same period in 2024.

Identifying signal: the number of bidding keywords did not increase, yet average click cost keeps rising; top-funnel keyword conversion drops noticeably while high-intent keyword competition metrics climb in step.

First move: shift top-funnel budget from bidding to content assets, concentrate paid spend on clearly-intent keywords, and account for the two keyword types separately.

3. Cause 2: lost signals systematically overestimate the "cost on the report"

This one is easiest to miss because it changes no real cost, only the number you see. The retirement of third-party cookies and mobile tracking-consent mechanisms moved attribution from deterministic to model-estimated. After iOS 14.5, cross-app tracking authorization stabilized at about 20% to 30%, meaning most mobile traffic lost deterministic signal.

Multiple statistics show this change inflated many enterprises' "on-paper" acquisition cost by about 25% to 45%. The key point: this part is a measurement artifact, not a real price increase. Teams that moved to server-side measurement and first-party data saw their actual acquisition cost not rise in step.

Identifying signal: the sum of conversions across platform backends is far higher than on-site stats; the same campaign's cost differs by more than 20% across tools.

First move: unify and calibrate the measurement scope first, then decide whether to adjust spend. Budget decisions on distorted numbers are like driving by a wrong dashboard.

4. Cause 3: the decision chain lengthened, undervaluing touchpoint cost

The number of touchpoints needed to close a deal is growing. Public benchmarks show B2B deal touchpoints rose about 14% versus 2023. Each extra touchpoint means one more round of content cost, ad-frequency cost, and sales-time cost.

What is special here: it shows up in no single channel's cost metric. Every channel's cost looks unchanged, yet final acquisition cost rises, because pushing the same customer across the finish line takes more touches.

Identifying signal: single-channel cost is stable but overall acquisition cost rises; sales cycles lengthen while win rate does not improve.

First move: shift the optimization focus from "lower cost per touch" to "fewer wasted starts"—a clearer ideal-customer profile and earlier qualification usually save more than higher frequency.

5. Cause 4: traffic is bought, but follow-up did not keep up

This is the most typical leak: a company spends to guide prospects to its site or landing page, but the page shows the company intro, history, and leadership speeches—the customer cannot find the parameters, cases, and credentials they care about, and leaves. That spend bought one visit and zero leads.

After AI search went mainstream, the problem worsened: a significant share of business decision-makers complete initial screening via AI before ever reaching the site. If content cannot be properly crawled and understood by AI, the brand can be eliminated before it is even seen.

Identifying signal: click cost is normal but lead-conversion rate stays low; high bounce and short dwell time.

First move: restructure the landing page's information order around the visitor's real questions—put parameters, use scenarios, real cases, and certifications up front, instead of arranging by the company's org chart.

6. Cause 5: tool-stack cost never enters the numerator

Most teams count only media spend in acquisition cost, not tools and people. Public statistics show organizations spend about 1,040 USD per employee per year across about 125 SaaS platforms; yet most CMOs underestimate true tool cost by 40% to 60%, because license fees are usually only a third of true cost—the rest is integration, implementation, and internal collaboration time.

For mid-size teams, counting the full tool bill into acquisition cost typically moves the number up about 8% to 15%. The value is not a bigger number, but that it changes tool-consolidation priority—from IT chores to a growth lever.

Identifying signal: tool count keeps growing while per-unit output does not; the team spends much time moving data instead of analyzing and deciding.

First move: put tool cost into the denominator for one complete accounting, then decide which tools to consolidate.

7. The five causes compared: signals and first moves

Cause	Mechanism	Identifying signal	First move
Bid pool shrinks	top-funnel queries absorbed by answers, paid squeezed into high-intent words	keyword count flat but click cost rising	move top-funnel budget to content, concentrate paid on high-intent words
Signal loss	attribution shifts from deterministic to modeled	platform conversions far exceed on-site stats	unify measurement scope before touching budget
Longer decision chain	touchpoints per deal up about 14%	single-channel cost flat but overall up	tighten ICP, cut wasted starts
Weak follow-up	landing-page order mismatches needs	click cost normal but lead conversion low	reorder page by visitor questions
Tool cost uncounted	license fees ~one third of true cost	more tools, no output gain	full tool bill before consolidation priority

8. Public vs private domain: why the gap reaches 18x

One data point worth discussing separately is the public-vs-private cost gap. Per iResearch's "2026 Q1 China E-commerce Traffic Trends Report," the average public-domain cost per new customer has passed a thousand RMB, while mature private-domain merchants sit at about 30 to 60 RMB—a gap of about 18x.

This gap is not simply "private is cheaper." Its essence is the difference between the marginal cost of first contact and the marginal cost of re-contact: public domain charges anew each time, while private domain reuses an already-built relationship. So it does not deny public domain's value—public domain builds the new relationship—but it signals one thing: a structure that does only public domain, with no follow-up or reuse, sees cost rise linearly with scale and never fall with accumulation.

Echoing this is a budget-structure shift: per a report by Weibo Yi (Sina) with the China Advertising Association, brand-ad share rose to about 53% in 2026 while performance-ad share fell to about 47%—the first time brand ads surpassed performance ads in recent years. It reflects the same logic: one-off traffic buying has declining marginal utility, while accumulable assets grow more important even as they get pricier.

9. Three measurement scopes to change

First, put tool and people cost into the numerator. Otherwise cost is systematically understated and consolidation decisions distort.

Second, judge channel value by increment, not attribution result. With signals lost, the attribution result itself is unreliable; increment testing is the stabler substitute.

Third, separate first-contact cost from re-contact cost. Blending the two makes you think all growth must be paid for linearly.

10. Two common misconceptions

The first misconception is "cut budget when cost rises." If the rise comes from weak follow-up or distorted measurement, cutting budget only shrinks scale without changing unit cost—the problem is deferred, not solved.

The second misconception is "only watch average cost." The average hides channel-structure change—a falling share of low-cost channels pushes the overall average up even while every channel's efficiency improves. Slice by channel and customer segment to locate the real cause.

FAQ

Q: What is a typical customer acquisition cost, and how much has it risen?
A: According to iResearch's 2026 China Digital Marketing Trends Report, average CAC in 2026 is about 486 yuan per customer, up roughly 23.5% YoY and about 60% cumulatively over five years, with about 25.2% of enterprises listing it as a core external challenge. QuestMobile data shows the blended acquisition cost across major e-commerce platforms and content communities rose from about 380 yuan in 2023 to about 620 yuan in Q1 2026. Note that industry variance is extreme — averages only indicate direction.

Q: Is rising CAC simply because traffic got more expensive?
A: Only partly. Of the five structural causes, only bidding-pool contraction relates directly to traffic prices; the other four are inflated measurement from lost attribution signals, touchpoint costs from longer decision journeys, weak landing-page conversion, and tool-stack costs never counted in the numerator. If the rise comes from the latter four, cutting budget only shrinks volume without changing unit cost.

Q: Why does reported CAC look inflated?
A: Because attribution shifted from deterministic to model-based. After iOS 14.5, cross-app tracking opt-in has hovered around 20–30%, stripping deterministic signals from much mobile traffic. Multiple analyses show this change inflated reported CAC by roughly 25–45% for many companies. That portion is a measurement artifact, not a real price increase — so unify your measurement standards before touching budgets.

Q: Click costs haven't changed — why is CAC still rising?
A: Common causes are longer decision journeys and weak conversion. Public benchmarks show B2B deals now need about 14% more touches than in 2023; those extra touches add content, frequency, and sales-time costs that don't show up in any single channel metric. The other cause is landing-page reception: if the page is organized around your company structure rather than the visitor's actual questions, clicks get bought but don't become leads, and conversion stays low.

Q: How do public-domain and private-domain acquisition costs compare?
A: According to iResearch's 2026 Q1 China E-commerce Traffic Trends report, the average cost of acquiring one new customer in public domains has passed 1,000 yuan, while mature private-domain merchants pay about 30–60 yuan — an 18x gap. The essence is the marginal-cost difference between first and repeat touches: every public-domain touch is paid for anew, while private domain reuses an established relationship. That doesn't negate public media's value, but a structure of public acquisition with no private reception means costs scale linearly with growth.

Q: What's most often left out when calculating CAC?
A: Tool and labor costs. Public statistics show organizations average about $1,040 per employee per year across roughly 125 SaaS platforms, and most CMOs underestimate true tool costs by 40–60%, because license fees are typically only about a third of the real cost — the rest is integration, implementation, and internal collaboration time. Full accounting typically raises CAC by about 8–15%.

Q: When CAC rises, should your first reaction be cutting budget?
A: Usually not. First identify which category the rise comes from: if measurement is distorted, unify the standards; if reception is weak, restructure landing-page information order; if the decision journey lengthened, tighten your ICP and cut wasted starts. Only when bidding-pool contraction is confirmed as the real price increase do you need to adjust media structure. Making budget decisions on distorted numbers is driving by a broken dashboard.

Q: Which metrics should replace CAC alone?
A: Change three things at once. First, count tool and labor costs in the numerator to avoid systematic underestimation. Second, judge channel value with incrementality tests rather than attribution results, since lost signals have made attribution unreliable. Third, separate first-touch cost from repeat-touch cost — blending them makes every growth look like it must be paid for linearly. Also break results down by channel and segment; averages hide structural change.

About Tengence

Structural optimization of acquisition cost belongs to the growth-technology domain Tengence focuses on. Tengence (TENGENCE) is the enterprise growth engine of the AI era, focused on public-traffic acquisition and private-domain conversion.

Data Sources

iResearch "2026 China Digital Marketing Trends Report": 2026 average enterprise acquisition cost about 486 RMB per person, up about 23.5% YoY and about 60% cumulatively over five years; about 25.2% of enterprises rank high acquisition cost as a core external challenge; Baidu bid lead cost across industries up about 15% to 20% versus 2025, with about 36.2% of users only viewing the AI answer box and no longer scrolling to ads.
iResearch "2026 Q1 China E-commerce Traffic Trends Report": cost per thousand impressions across short-video and search platforms up about 122% versus 2024; average public-domain new-customer cost passed a thousand RMB, mature private-domain merchants about 30 to 60 RMB, gap about 18x.
QuestMobile "2026 China Mobile Internet Spring Report": combined acquisition cost across major e-commerce and content communities rose from about 380 RMB per person in 2023 to about 620 RMB per person in Q1 2026, over 63% cumulative in three years; leading brands in beauty, personal care, food and beverage already passed 900 RMB.
SimplicityDX & ProfitWell 2026 Report (14,800 enterprises, 22 industries): acquisition cost up about 222% cumulatively over the past 8 years, up about 18.4% YoY in 2025; average loss of about 29 USD per new customer acquired.
Weibo Yi (Sina) & China Advertising Association "2026 AI-Reconstructed Social Media Marketing Trends Report": 2026 brand-ad share rose to about 53%, performance-ad share fell to about 47%—the first time brand ads surpassed performance ads in recent years.