Market and competitive research.
August 2026. Competitive status, what the alternatives cost, the sample size the loop actually needs, and the point where the numbers stop being defensible.
Three findings that shape the pitch.
The category thinned out
Of the three companies usually named in AI website personalisation, one was acquired into a site builder and one deprecated the product line and left. What remains is the older client-side generation.
The substrate is funded, the application is not proven
Sandboxed agent execution has attracted institutional capital. That validates what Vivere runs on. It says nothing about whether autonomous CRO is a durable business.
The traffic requirement is narrower than convenient
At the low end of the floor, a fixed-horizon test confirms large wins in a two-week loop and small ones in months. That argues for an agent allowed to say no.
Two of three named rivals are gone.
Eighteen months emptied out the nearest category.
Intellimize
Acquired by Webflow for a reported eight figures. The team and the technology went into Webflow's site builder. Webflow is the competitor to name now, and it is a different one.
Mutiny
Raised about $72M and reached eight-figure ARR, then deprecated website personalisation entirely and moved to go-to-market content. A funded company with revenue chose to leave this market.
VWO, Optimizely, AB Tasty and peers
Still scripts and visual editors. Render flicker, a third-party runtime on the page, and nothing that accrues to the customer's codebase once the subscription ends.
v0, Bolt.new, Lovable
Prompt-driven code generation with no analytics ingestion and no live conversion data. They compete for attention, not for the same job.
Mutiny is the consequential one. Two readings are available, that the client-side model is less durable than it looked, or that the market moved toward approaches which touch the codebase directly. The founder wrote about the decision publicly, and reading that account is a prerequisite to framing competitive risk credibly.
Capital is backing the execution layer.
$21M
E2B Series A
Led by Insight Partners, July 2025.
$33M
Raised in total
Firecracker microVM sandboxes for agent code execution.
88%
Of the Fortune 100
Signed up. Perplexity, Hugging Face, Groq and Vercel named.
Evidence about the substrate, not about the application built on it. The two are worth keeping separate in a pitch, because conflating them is the kind of thing an investor notices.
Seven reports. A sevenfold spread.
No single figure survives scrutiny, so the sizing argument should not rest on one.
Verified Market Research
$10.4B
BusinessResearchInsights
$4.26B
DataIntelo
$3.8B
MarketGrowthReports
$2.86B
WiseGuyReports
$2.67B
GlobalGrowthInsights
$2.0B
Virtue Market Research
$1.45B
The usable version is a range: roughly $2B to $10B today, growing at a reported 9% to 12%. The argument itself should rest on bottom-up replacement of a specific, checkable cost, which is the next section.
The strongest number on the page.
Independently checkable, unlike anything top-down.
In-house growth pod
$260K to $325K
CRO agency, mid-tier
$60K to $180K
CRO agency, entry
$24K to $60K
Median site
2.35%
Top 10% of sites
11.45%
On page speed, Portent found that a page loading in one second converts 32% higher than one loading in three, and that each extra second between zero and five costs about 4.42% of conversions. Use that rather than the familiar one-second-costs-7% line.
What the loop can actually confirm.
How much traffic a page needs before the loop can act is the most load-bearing claim in the pitch.
50% lift, 3.0% to 4.5%
6 days~5,000 visitors
30% lift, 3.0% to 3.9%
14 days~12,900 visitors
20% lift, 3.0% to 3.6%
31 days~27,800 visitors
10% lift, 3.0% to 3.3%
118 days~106,300 visitors
| Uplift | 900/day | 1,500/day | 2,500/day |
|---|---|---|---|
| 50% lift | ~6 days | ~3 days | ~2 days |
| 30% lift | ~14 days | ~9 days | ~5 days |
| 20% lift | ~31 days | ~19 days | ~11 days |
| 10% lift | ~118 days | ~71 days | ~43 days |
Sequential testing with always-valid p-values changes the shape of this but not its magnitude. It permits stopping early when an effect is large and unambiguous. It does not reduce the evidence needed to detect a small one.
So the correct behaviour at the bottom of the traffic band is to ship quickly on large wins and report insufficient signal on subtle ones. That makes the Analyst’s authority to halt a run the most important safety property in the architecture, and a stronger position with a technical buyer than claiming the statistics are solved.
Figures with no traceable source.
Listed rather than dropped, because they are still in circulation and will come back otherwise.
- $15B / $3.5B / $250M
- Total, serviceable and obtainable market. No source, no methodology.
- Company counts by region
- No defensible figure exists without primary dataset work.
- 1 second costs 7% of conversions
- Attributed to Akamai everywhere, traceable to Akamai nowhere.
- Lovable's early ARR figures
- Secondary reporting only. The growth is real; the numbers are not filed.
Resolve before pitching.
- 01
Why Mutiny left website personalisation
The founder's public account determines whether the nearest precedent is a warning or a tailwind.
- 02
Traffic and stack data for the first market
Domains running PostHog alongside live ad pixels, from BuiltWith or Tracxn. Before the first outbound push, not after.
- 03
Which sequential method the Analyst uses
The speed-up from mSPRT, Bayesian or group-sequential testing is specific to the method and the effect size.
- 04
A traceable page-speed citation
Or drop the Akamai figure entirely and use Portent, which is already clean.
Lead with what can be checked.
The architecture is sound on its own terms: a closed loop native to the customer’s codebase, executed in a sandbox, with trust established through observability and a one-command revert rather than an approval queue. The nearest category has emptied out, and the execution substrate is attracting capital.
The weaknesses are in the numbers rather than the architecture. Third-party market sizing does not survive scrutiny, and the traffic floor is defensible for large wins and optimistic for subtle ones. Neither undermines the business. Both argue for leading with competitor status, salary and agency costs, infrastructure funding, and statistics stated honestly.
Go touch grass.
Vivere is not open yet. Join the waitlist and we will come to you first.