Table of contents
Median customer acquisition cost payback is 16 months, and marketers self-rate their ability to demonstrate technology ROI at 4.2 out of 7. That gap is where AI marketing claims live, and where they should be tested.
Key Takeaways
- Median CAC payback is 16 months, improved from 18.
- The strong quartile pays back in 10 months.
- The weak quartile takes 24 months.
- Top performers pay back in six months or less.
- Fastest-growing firms sit at 10 months against 18.
- Demonstrating technology ROI self-rates 4.2 out of 7.
- Generating technology ROI self-rates 4.5 out of 7.
- Median firm growth is 9.9%, the lowest since 2018.
- High-growth firms grow 36.6%; no-growth firms shrink 11.4%.
- Profitability runs 39.5%, 29.6% and 20.3% across those tiers.
- High-growth firms spend 12.0% of revenue on marketing against 5.0%.
- 56.4% of growth spend targets existing markets.
- Marketing budgets are 9.0% of revenue and growing 1.7%.
- 70.6% report a shift toward short-term results.
- Only 9% of martech stacks are fully AI-optimised.
- Retainer-shaped buying costs four to five times more than fixed scope.
Pick a comparator that predates the AI claim
Benchmarkit's CY-2025 benchmarks put median customer acquisition cost payback at 16 months, improved from 18 - an 11% gain - with the strong quartile at 10 months, the weak quartile at 24 and top performers at six months or less. The fastest-growing firms sit at 10 months against 18 for the rest.
These numbers existed before anyone pitched you an AI roadmap, which is exactly why they make a good test. A programme that improves acquisition efficiency will show up here; one that only improves output volume will not.
| Payback cohort | CAC payback | Read |
|---|---|---|
| Top performers | 6 months or less | The ceiling, rarely the target |
| Strong quartile | 10 months | A realistic 12-month goal |
| Fastest-growing firms | 10 months | Efficiency and growth travel together |
| Current median | 16 months | The number to beat |
| Prior-year median | 18 months | An 11% year-on-year gain |
| Weak quartile | 24 months | Fix the funnel before scaling AI |
The measurement base is the real constraint
The CMO Survey 2026 asks marketers to rate martech capability from 1 to 7. Generating technology ROI scores 4.5 and demonstrating that ROI 4.2, against vendor selection 4.9 and systems integration 4.8. Nothing scores above 5, and the ratings are flat against 2024.
Buying is the strongest capability; proving value is among the weakest. Add generative volume to that and you get more activity with the same inability to explain what it earned.
Adoption is not a result
The same survey reports AI covering 24.2% of marketing activities, up from 13.1%, generative AI at 22.4% against 7.0% - a 220% rise - and 55.9% projected within three years. Meanwhile Gartner's 2026 CMO Spend Survey finds only 9% of stacks fully AI-optimised, martech at 19.4% of budget and 56% of that spend consumption-priced.
Adoption percentages describe activity. Payback lives in the efficiency numbers, and consumption pricing means the cost side moves before the benefit side does.
The growth backdrop leaves little slack
Hinge's 2026 High Growth Study reports median growth of 9.9%, the lowest since 2018, with high growth at 36.6%, average at 8.5% and no growth at -11.4%. Profitability follows the same order at 39.5%, 29.6% and 20.3%, and high-growth firms spend 12.0% of revenue on marketing against 5.0%.
Note which way the causality runs in the data: the firms growing fastest also spend more and earn more. An AI programme sold as a way to spend less is arguing against the pattern.
| Growth tier | Revenue growth | Profitability | Marketing spend of revenue |
|---|---|---|---|
| High growth | 36.6% | 39.5% | 12.0% |
| Average growth | 8.5% | 29.6% | Not reported separately |
| No growth | -11.4% | 20.3% | 5.0% |
| Median firm | 9.9% | Not reported separately | Not reported separately |
Referrals still dominate the lead mix
Hinge reports referrals at 39.5% of leads and outreach at 23.5%, with 80% of firms running research (competitive 46.1%, client 45.2%) and 20% running none. Keyword and SEO research fell from 33.5% to 27.0% of firms, while 16.8% reported M&A activity.
If four in ten leads arrive by referral, an AI programme aimed purely at top-of-funnel volume is optimising the smaller half of the mix. The higher-leverage use case is usually speed and quality of response.
Point it at the biggest existing line
The CMO Survey reports growth spend split 56.4% on selling more to existing markets, 22.9% on product development (up from 19.2%) and 14.1% on new markets (down from 17.0%). Marketing budgets sit at 9.0% of revenue and 9.6% of overall budgets, growing 1.7%, with 68% of spend on present-focused activity against 32% future-focused and 70.6% reporting a shift toward short-term results.
Penetration has the widest measurable base, and it is where a short-term bias actually helps. New-market AI experiments are the hardest to attribute and the first to be cut.
Control the cost side, precisely
Fractional Pulse's 2026 comparison finds that buying project-shaped work through a monthly retainer costs four to five times more - USD 180,000 against USD 40,000 - with retainers at USD 5,000 to 25,000 a month on six to twelve month terms and projects running 4 to 24 weeks.
ROI has a denominator. Halving the cost of the engagement improves the return as reliably as improving the result, and it is the half you control on signature day.
| ROI input | 2026 figure | How to move it |
|---|---|---|
| CAC payback | 16 months median | Fix conversion before adding volume |
| Demonstrating technology ROI | 4.2 out of 7 | Freeze definitions, name an owner |
| Martech share of budget | 19.4% of budget | Retire a tool per tool added |
| Consumption exposure | 56% of spend usage-priced | Write a usage cap into the contract |
| Engagement shape | Retainers cost 4-5x fixed scope | Buy phases, not years |
| Growth spend focus | 56.4% existing markets | Aim AI at penetration first |
Operating discipline shows up in the numbers
The CMO Survey reports 71% of leaders calling agility key, training down to 3.8% of marketing spend from 5.8% and headcount growth down 50% year on year. Vendor research from EOS Worldwide reports that companies running EOS with a professional implementer grew 2.8 times faster in a validated study of 305 companies - worth reading as directional, since the publisher sells the system.
The transferable point is unglamorous: a cadence with owners and numbers beats a tool. That holds whether the operating system is EOS or a spreadsheet.
It is also the cheapest intervention available in 2026. With spend growth at 1.7% and training at 3.8% of spend, a weekly review with owners attached costs nothing and is the one change that makes every other number legible.
A payback plan you can actually audit
Set the baseline at 16 months CAC payback, target the strong quartile at 10, accept that measurement capability starts at 4.2 out of 7, cap consumption while 56% of martech spend is usage-priced, and buy the work as phases rather than a retainer at four to five times the cost.
Our data intelligence team builds that baseline before any automation is scaled, and our growth team owns the number afterwards. We apply the same test to channel spend in our paid social ROI analysis.
Which AI use cases can actually be measured
Payback claims are only testable where a clean before-and-after exists. Content generation at 73.9% adoption produces output volume that is easy to count and hard to value; audience targeting at 45.2% can be tested against a holdout; automation at 48.9% shows up in cycle time. Sorting on measurability, not adoption, keeps the business case honest.
The rule we apply is simple: if you cannot name the number, the cadence and the comparator before the pilot starts, the pilot is research - fund it as research and stop calling it ROI.
| Use case | 2026 adoption | Measurable as | Payback horizon |
|---|---|---|---|
| Content generation | 73.9% | Cost and time per asset | One to two quarters |
| Personalisation | 65.4% | Lift against a holdout | Two quarters |
| Marketing automation | 48.9% | Cycle time and touch coverage | Two to three quarters |
| Data analysis | 46.3% | Decisions made per reporting cycle | Hard to isolate |
| Audience targeting | 45.2% | CAC and payback months | Against the 16-month median |
| Generative engine optimisation | 41.5% working on it | Share of AI answers cited | Two to four quarters |
What to ask before you sign
Three questions expose most AI ROI claims. Which number moves, against which published comparator? What cadence is it reviewed on? What decision happens if it does not move? With median growth at 9.9% and spend growth at 1.7%, there is no budget for a fourth answer.
If you want your own baseline scored against these figures, send us your reporting pack and we will mark it against the 2026 benchmarks rather than a vendor deck.
Frequently Asked Questions
How long should an AI marketing programme take to pay back?
Judge it against acquisition efficiency rather than adoption. Benchmarkit's CY-2025 benchmarks put median customer acquisition cost payback at 16 months, improved from 18, with the strong quartile at 10 months, the weak quartile at 24 and top performers at six months or less. If an AI programme is working, that number should move within a year - not the number of licences in use.
Why is AI ROI so hard to demonstrate?
Because the underlying measurement capability is weak before AI is added. The CMO Survey 2026 asks marketers to self-rate martech capability from 1 to 7 and gets 4.5 for generating technology ROI and just 4.2 for demonstrating it - the second-lowest score in the set, and flat against 2024. AI does not fix an attribution problem; it increases the volume of things you cannot attribute.
What does the growth backdrop look like?
Tight. Hinge's 2026 High Growth Study reports median professional services growth of 9.9%, the lowest since 2018, with high-growth firms at 36.6%, average-growth at 8.5% and no-growth firms at -11.4%. Profitability follows the same order at 39.5%, 29.6% and 20.3%. There is little slack in 2026 budgets for unproven spend.
Where should AI-driven gains show up first?
In the biggest existing line. The CMO Survey 2026 reports 56.4% of growth spend going to selling more to existing markets, 22.9% to product development (up from 19.2%) and 14.1% to new markets (down from 17.0%). An AI programme aimed at penetration has the widest measurable base; one aimed at new markets is the hardest to attribute.
How do we stop AI ROI claims from being unfalsifiable?
Attach every claim to a number with a cadence and a decision. Benchmarkit's payback quartiles give the efficiency comparator, the CMO Survey's 4.2 out of 7 gives the honest measurement baseline, and Fractional Pulse's finding that retainer-shaped buying costs four to five times more gives the cost control. A claim without a comparator is a demo, not a result.
Sources
Benchmarkit - CY-2025 B2B SaaS Performance Metrics Benchmarks
The CMO Survey - Highlights and Insights Report 2026
Hinge Research Institute - 2026 High Growth Study
Gartner 2026 CMO Spend Survey via Chief Marketer
Fractional Pulse - Fractional executive engagement comparison 2026
EOS Worldwide - Companies running EOS grew 2.8x faster (vendor research)
