TL;DR
SplitTesting.com was a productized conversion rate optimization business. Productizing the offer was the lever. Bespoke CRO sells time; productized CRO sells outcomes. We rebuilt the offer around a fixed scope, a fixed price, a defined deliverable cadence, and a defined success metric. The result was 800% revenue growth in under 9 months, 88%+ of clients hitting record revenue, and an exit at 11x EBITDA. The receipts are real. The lessons compound.
- Productizing is an architecture decision, not a marketing decision.
- The offer drives everything: pricing, hiring, ops, and exit multiple.
- Bespoke services exit at 2-4x EBITDA. Productized services exit at 8-12x. Same revenue, different math.
In this article
Why we productized
The CRO industry, when SplitTesting.com was being built, was a custom-services market. Agencies sold hours, retainers, and scopes. Pricing was opaque. Outcomes were debated. The buyer experience was paying $20k a month for an unclear number of A/B tests, run by an unclear number of people, against an unclear definition of success.
We saw a different game. The buyer did not want hours. The buyer wanted a higher conversion rate. If you sold the outcome instead of the inputs, the entire economic model changed.
That insight is the architecture decision. Everything else followed.
What "productized" actually means
A productized service is not a service with a pretty website. It is a service that has been redesigned around four constraints:
- Fixed scope. What is in the box is defined and unchanging. New asks are out of scope unless they buy another box.
- Fixed price. One number on the page. The customer knows it before the sales call.
- Fixed deliverable cadence. Outputs ship on a schedule the customer can rely on. Weekly, monthly, whatever, but unchanging.
- Fixed success metric. Both parties agree, before purchase, on what success will look like.
A custom service violates all four. The scope flexes, the price negotiates, the deliverable timing slips, and success is debated post-hoc.
The interesting thing is that customers prefer the productized model even when it costs more. Removing ambiguity is a value the customer pays for. The premium for productizing is real and large.
Customers do not want hours. They want outcomes. Sell the outcome.
The offer architecture
The SplitTesting.com offer had a specific shape. Three pricing tiers, each defined by:
- Test velocity: how many tests per month. The single most legible metric the buyer understands.
- Implementation included: what we built versus what the client's team built.
- Reporting cadence: weekly reads, monthly strategic, quarterly review.
The pricing was on the page. The scope was on the page. The success metric (incremental revenue from winning tests, measured by an agreed methodology) was on the page. Buying was a decision, not a negotiation.
Within the offer, we standardized the test format. A test brief had a fixed template. A test result had a fixed template. The post-test decision had a fixed template. Customers got the same artifact every time, which made internal decision-making at the client side faster, which made our results compound faster, which made the client renew.
This is the part that most agencies miss. The brief discipline is what makes the productized business scale. Without it, every test is a custom artifact, every result is a custom report, and the labor cost grows linearly with revenue. With it, the labor cost grows sublinearly with revenue, which is the math that produces an 11x exit multiple.
The delivery system
The delivery system was the operating moat. Three components.
1. The test factory
A pod of specialists organized around a recurring test calendar. Strategist, analyst, designer, developer. The pod ran a fixed cadence of tests per month per client. The cadence was sacred. The work that did not fit the cadence got reclassified as out of scope or rescheduled to a future cycle.
The factory model meant that adding a new client did not require hiring a new everything. It required adding capacity to existing pods. Sublinear cost growth.
2. The reporting layer
Every client received the same reporting artifact on the same day of the week. The artifact was templated and partially automated. A junior analyst could produce the report; a senior strategist reviewed it. The client could compare reports across months without translating between formats.
Standardized reporting is a competitive moat at the client side, because the client's internal team gets faster at acting on the report. The faster they act, the more the program works. The more it works, the longer they stay.
3. The hand-off protocol
Tests do not produce value unless results get implemented. We built a hand-off protocol with the client's engineering team, with a fixed format, fixed cadence, and fixed escalation path for stuck implementations. The protocol was as much a deliverable as the tests themselves.
Most CRO agencies wash their hands at the test result. We did not. The 88%+ of clients hitting record revenue was downstream of the implementation discipline, not the testing discipline. Most agencies do not get credit for outcomes they help cause because they do not own the hand-off.
Why it exited at 11x
The 11x EBITDA exit was not luck. It was a function of the offer architecture. Buyers paid 11x because the business had:
- Predictable revenue. Monthly recurring with low churn, because the offer was clear and the outcomes were real.
- Sublinear cost growth. The factory model meant adding revenue did not require proportional headcount.
- A defensible position. "We are the productized CRO" was an ownable narrative in a market full of generalist agencies.
- An obvious roadmap. The next adjacent product (productized analytics, productized retention) was visible to the buyer.
Custom services exit at 2-4x EBITDA because the cost grows with revenue and the buyer cannot see how to scale it. Productized services exit at 8-12x because the cost grows slower than revenue and the buyer can see the next product.
Same revenue line. Different exit multiple. The architecture decision is worth multiples.
What this means for AI services
In 2026, the AI services market looks exactly like the CRO market looked then. Custom engagements. Hours billed. Opaque outcomes. Buyer confusion. There is a productizing opportunity that mirrors the SplitTesting.com move almost perfectly.
The productized AI services that will exit at 10x+ have:
- Fixed scope (for example, "we ship one production AI workflow per quarter")
- Fixed price (one number, on the page)
- Fixed cadence (defined deliverables on a known schedule)
- Fixed success metric (a P&L number the buyer can see)
This is what I am building inside Automatic. It is also what most AI consultancies will fail to do, because the easy money in year one is custom work and the hard money in year three is productized work. The agencies that do the hard work win.
The pattern is the same. The leverage is different.
The bottom line
SplitTesting.com grew 800% in under nine months, 88%+ of clients hit record revenue, and the business exited at 11x EBITDA. The architecture decision was productizing. The execution was operating discipline. The exit multiple was math.
If you are running a services business, the question worth asking this quarter is whether you are selling hours or selling outcomes. The first model exits at 3x. The second exits at 10x. The work to get from one to the other is real, but it is bounded, and the math at the end is worth it. For the AI program version of this discipline, see the AI transformation playbook for consumer brands.
FAQ
What is SplitTesting.com?
SplitTesting.com was a productized conversion rate optimization business that sold outcome-based CRO services to consumer brands. Under Nicholas Harris's leadership, the company grew 800% in under 9 months, served clients to record revenue performance for 88%+ of accounts, and exited at approximately 11x EBITDA.
How is a productized service different from an agency?
A productized service has fixed scope, fixed price, fixed deliverable cadence, and a fixed success metric. A traditional agency has flexible scope, negotiated pricing, variable deliverables, and disputed success metrics. The productized model commands higher prices, retains customers longer, and exits at substantially higher multiples.
Why do productized services exit at higher multiples?
Productized services have sublinear cost growth (revenue scales faster than headcount), predictable recurring revenue, defensible positioning, and a visible product roadmap. These four properties drive exit multiples of 8-12x EBITDA, compared to 2-4x for custom services businesses.
What was the most important architecture decision at SplitTesting.com?
The most important architecture decision was the offer itself: fixed scope, fixed price, fixed cadence, fixed success metric. Every other decision (hiring, pricing, ops, exit strategy) followed from the offer. Productizing the offer is the lever; everything downstream is execution.
Can this model be applied to AI services?
Yes, and the opportunity is large. The AI services market in 2026 looks like the CRO market did before SplitTesting.com. Custom engagements dominate, outcomes are opaque, buyers are confused. A productized AI services business with fixed scope, fixed price, and a fixed success metric is the play.
How long did the productization take?
The offer redesign was the work of one quarter. The delivery system took two more quarters to stabilize. Revenue compounded once the model was in place; the 800% growth in under 9 months happened after the architecture was stable, not during the redesign.