AI CRO
The CRO Founder's Guide: What to Buy, What to Avoid, What to Pay in 2026
Last updated: [Updated Date]
What to Buy, What to Avoid, What to Pay in 2026
A buyer's guide written by an operator, not a sales team.
Author: Chris McCarron, Founder, GoGoChimp
Date: 14 May 2026
Version: v1.0
Length: ~5,950 words / 12 pages
Abstract
Most CRO buyer's guides are written by agencies trying to qualify you into their own service. This one isn't. It's written by a 13-year Glasgow-based operator who has shipped CRO for ecommerce, SaaS, and non-profits, and whose honest advice in roughly one of every five conversations is "don't hire anyone right now."
This paper covers four things. When you should NOT buy CRO (most founders under £50K/month shouldn't). The three legitimate buying paths and the revenue brackets each one fits. The seven questions that separate operators from cosplay agencies. The pricing benchmarks for the UK, EU, and US in 2026.
It cites Build Grow Scale's 347-store industry research, Baymard Institute's UX benchmarks, and the going rate at the three best-known retainer agencies. It names competitors by name where naming them serves the buyer.
If you finish this paper and conclude the right call is to wait six months and run ads instead, that's the right outcome. Buy when buying is the answer. Not before.
Executive Summary
If you read nothing else, read this.
The three questions every founder must answer before buying CRO
- What's the bottleneck? Not "we want to grow," but the specific page, step, or moment where money leaks. If you can't name it in one sentence, you're not ready to buy. Get a free CRO audit first.
- Do you have enough traffic to test? Below roughly 30,000 monthly sessions on the page you want to optimise, statistical significance becomes impractical. Spend on acquisition first.
- Who owns the implementation? A test that gets won and never shipped is worse than no test at all. If your dev team has a six-month backlog, CRO without implementation capacity is shelfware.
The three biggest CRO procurement mistakes
- Buying retainer when you needed productised. Founders sign £5,000/month for what should have been a £2,500 one-off Speed Sprint. The retainer earns its keep at scale. It does not earn it on a single fixable bottleneck.
- Buying on case studies that don't have receipts. Anyone can claim a 200% lift. Ask for the Loom analytics walkthrough. Ask for the Trustpilot review. Ask for the third-party citation. If they can't show it, walk.
- Buying from someone who won't read your audit personally. Account-manager-led engagements are where the operator hands you to a junior. The post you are about to read is written by an operator. So is the audit you should buy. So is the contact who should answer your first reply.
The three legitimate buying paths
The rest of this paper is the detail behind those three rows.
Section 1: When you should NOT buy CRO
Most CRO content assumes you should buy. This section assumes you maybe shouldn't.
If your reaction to A/B testing is the same as your reaction to flossing (yeah, I should, but never quite get round to it), close this tab. The rest of this paper is for the founders who already floss.
For the rest of you, here are the four scenarios where the honest answer is wait.
1.1 You're pre-£50K/month revenue
Below £50K/month gross merchandise value, your traffic is almost certainly too thin to run statistically valid tests. To detect a 10% conversion lift at 99% significance on a baseline 2% conversion rate, you need roughly 30,000 visitors per variant. That's 60,000 sessions for a single A/B test. A pre-£50K/month store typically does 20-40K sessions a month total.
Run that test and you'll get a result, but the result will be noise dressed as signal. The right move is acquisition. Spend the CRO budget on Google Ads, Meta, organic content, or PR. Come back when you have traffic to test against.
"Below 30,000 monthly sessions on the page you want to optimise, A/B testing returns false positives more often than real wins. Spend on acquisition first. CRO compounds when traffic is already arriving."
Chris McCarron, GoGoChimp
1.2 You don't know where the bottleneck is
If you can't name the page, the step, or the moment where conversion leaks, you don't need a CRO programme. You need a free audit.
Every reputable agency offers one. GoGoChimp's audit is at /audit. Conversion.com runs a "conversion review" (Conversion.com). Speero runs a strategic diagnostic (Speero). Take three audits. Compare what they tell you. Then buy.
Audits are how operators qualify themselves to you. Skip the audit and you're buying blind.
1.3 You've migrated platform in the last 60 days
Replatformed from Magento to Shopify? Just moved off WooCommerce? Wait. Traffic, analytics, and conversion patterns take roughly 60 days to stabilise after a migration. Testing during that window means you're testing against a moving baseline. The variant might "win" because the migration noise broke the control, not because your hypothesis was right.
Sit out the 60 days. Make sure GA4 is clean. Confirm Microsoft Clarity is logging correctly. Then start.
1.4 You don't have conversion analytics installed
If you don't have GA4 properly configured with ecommerce events, plus Microsoft Clarity for session replay and heatmaps, you have no measurement substrate. CRO on a site without analytics is acupuncture in the dark.
Install GA4 first. It's free. Install Microsoft Clarity second. Also free. Run them for 30 days. Then buy CRO.
This is the cheapest part of the entire CRO stack and the most often skipped.
Section 2: The three legitimate buying paths
There are three legitimate ways to buy CRO in 2026. They map to revenue brackets and to the specific problem you're solving. Pick the one that fits.
Path A: Self-serve AI tools (£200-500/month)
What it is. Software you buy and run yourself. The tool generates test variants, the tool runs the test, the tool calls the winner. You set the strategy. You implement the result.
Named examples. VWO Insights AI (VWO). Mutiny (Mutiny). AB Tasty's AI features (AB Tasty). Convert's AI module (Convert). Optimizely's AI-assisted experimentation (Optimizely).
Right when: you have at least 100,000 monthly sessions on the pages you want to test, an in-house product or growth function with bandwidth to manage testing weekly, and the discipline to ship the implementation when a variant wins. SaaS founders with a product team and ecommerce founders with a tech-led in-house culture are the natural fit.
Wrong when: you're a solo founder, or your team has no testing experience, or you don't have implementation capacity. The tool will generate the test. You won't ship the winner. The licence becomes shelfware.
Expected lift. Build Grow Scale's 2026 research across 347 ecommerce stores (Stafford, 2026) found that self-serve AI tools delivered conversion lifts of 4-7%. That's real but small. It's the price of running CRO without an operator setting the hypotheses.
Build Grow Scale's 2026 review of 347 ecommerce stores (Stafford, 2026) found self-serve AI tools delivered 4-7% lift. Expert-guided AI on the same software delivered 28-34%. The software didn't change. The operator did.
Honest trade-off. Self-serve is cheap. It's also the lowest ceiling. If you can run 4-7% lift in-house at £300/month, that's a fine outcome for a pre-£500K/month operation. Past that revenue, the missed lift from not having operator-set hypotheses costs more than the retainer would.
Path B: Productised engagements (£500-£2,500 one-off)
What it is. A scoped fixed-price intervention. The vendor identifies a specific bottleneck, runs the work, hands you the result, walks away. No retainer. No commitment.
Named examples. GoGoChimp runs three productised products: the AI Headline Lab (£500), the Speed Sprint (£1,500 one-off page-speed engineering, modelled on the BeeFriendly result documented at youtu.be/z2bjGvAkqn0), and the Pricing Experimentation Audit (£2,500). Frictionless Commerce runs productised teardowns (Frictionless Commerce). Conversion.com offers fixed-scope audits. Several individual operators on Twitter sell productised Loom-walkthrough audits at £200-£800.
Right when: you have a specific named bottleneck. You ran the audit, the audit said "your homepage LCP is 12 seconds and you're losing 40% of mobile traffic to it," and you want that one thing fixed. Productised is also the right starting point when you're trialing a vendor before committing to a retainer. £1,500 one-off is a survivable test of whether they're operators or cosplayers.
Wrong when: you have multiple compounding bottlenecks, or you want sustained testing programme rhythm. Productised solves the discrete problem. It does not build the long-term experimentation engine.
Expected lift.
- Page speed engineering: 10-25% conversion lift typical. BeeFriendly Skincare went from $48K/year to $1,447,225/year after a 2.24-second page-speed reduction (Ezra Firestone case, 2017, full video at youtu.be/z2bjGvAkqn0; full annotated teardown in our page-speed Shopify case study). Affordable Golf's homepage LCP went from 21.3s to 6.1s in March 2026, a 71% improvement on the single biggest mobile-conversion drag (see the Affordable Golf annotated teardown).
- Headline rewrites: 5-15% conversion lift typical. Super Area Rugs ran a 216.29% revenue increase in 37 days primarily on above-the-fold copy.
- Pricing tests: 10-30% conversion lift typical, with high variance. Pricing tests have the highest upside and the highest risk in the productised category.
Productised CRO is the right starting point when you have one named bottleneck and want it solved without a retainer commitment. It is also the cheapest way to test whether a vendor is an operator or a cosplayer. £1,500 down is a survivable trial.
Honest trade-off. Productised cannot compound. You get the one win. After that you either run another productised engagement, or you graduate to a retainer. The Speed Sprint is a one-shot intervention, not an ongoing programme.
Path C: Retainer programme (£2,500-£10,000/month)
What it is. A monthly engagement with an agency that runs continuous experimentation. Operator-set hypotheses, ongoing testing across the site, monthly reporting, quarterly strategy reviews.
Named examples. GoGoChimp's Sprint (£2,500 one-off), Growth (£2,500/month, 3-month minimum), and Scale (£5,000/month) tiers (gogochimp.com/#pricing). Conversion.com is the recognised retainer agency in the enterprise bracket, working with established ecommerce and SaaS brands at retainer fees that typically start in the high four figures and scale into five figures monthly (Conversion.com). Speero sits in the US/EU mid-market with retainer pricing that generally lands in the same bracket (Speero). Neither publishes exact rates publicly; both qualify by RFP.
Right when: you have at least £200K/month revenue, meaningful ad spend (typically £20K+/month in paid acquisition), and the goal is sustained compounding lift. Retainer CRO is how you keep finding revenue at the same site for 18-36 months.
Wrong when: you have a single fixable bottleneck. Wrong when your traffic is too thin to support continuous testing. Wrong when your dev team has no capacity to ship winners.
Expected lift. Build Grow Scale's 2026 research found expert-guided AI CRO (the retainer category) delivered 28-34% conversion lift (Stafford, 2026). Enzymedica UK went from 3.4% baseline to 16.9% on Black Friday 2021 (a 4.97× lift on the same promo day versus the prior year's 7% Black Friday) and held 11% through December, one of the worst months for health supplement sales (Loom analytics review). EM360 went from 0.12% to 7% conversion in 30 days. Donate For Charity ran 494.64% more donations in 30 days. The full pillar context for expert-guided AI CRO is in our Complete Guide to AI-Powered Conversion Rate Optimisation.
The 347 Method, Build Grow Scale's industry research, found expert-guided AI CRO delivers 28-34% lift versus 4-7% from DIY tools. The software is identical. The operator is what compounds.
Honest trade-off. Retainer CRO costs at least £30,000/year in the entry tier and £60,000+/year at Scale. It only pays back at revenue brackets where 5-10% incremental conversion lift covers the fee with comfortable margin. Below £200K/month, the maths gets tight.
The operator-vs-tool gap shows up at enterprise scale too. JPMorgan Chase ran a five-year pilot with Persado (2016 through 2019) that produced up to a 450% lift in click-through rates on AI-generated marketing copy versus human-written controls (Persado press release, 2019; WARC analysis). The pilot was operator-led from the JPMorgan side. The tool was the same one anyone could license. The differential was the briefing discipline that produced the inputs.
Section 3: The seven questions to ask every CRO vendor
If you only do one thing before signing a CRO contract, do this. Send the seven questions. Read the answers. The answers will tell you whether you're talking to an operator or a salesperson.
Question 1: "What's your statistical significance threshold?"
Why it's diagnostic. Most agencies report at 95% significance. That sounds rigorous. It isn't. At 95%, one in twenty winning tests is a false positive. Run twenty tests, ship the "winners," and you've shipped one regression you didn't catch.
What good looks like. GoGoChimp tests at 99% on high-volume engagements. That's stricter than the industry default and means roughly one false positive in a hundred. Conversion.com publicly references 95%+ (Conversion.com). Speero runs Bayesian methodology, which sidesteps the binary cutoff but should still be explained clearly. (Baymard Institute on testing rigour).
Answers that disqualify. "We don't really track significance." "We call winners on lift, not on statistical significance." "We use 90%." Any of those, walk.
Question 2: "How many tests do you run per quarter?"
Why it's diagnostic. Test cadence is the single best proxy for whether a CRO programme is actually running or just being talked about. An engaged programme on a meaningful-traffic site runs 8-15 tests per quarter at minimum. A serious programme runs 20-35.
What good looks like. GoGoChimp's Growth tier runs 30+ AI experiments per quarter per client. Conversion.com and Speero both publicly reference 20-40 tests per quarter on retainer accounts. The number depends on traffic, but the answer should be a number, not a description.
Answers that disqualify. "It depends." "We focus on quality not quantity." That last one is fine as additional context. It is not fine as the entire answer. Quality without quantity is one good idea per year, which is not a CRO programme. It's a hunch.
Question 3: "Show me a named loss, not a named win"
Why it's diagnostic. Every CRO agency has wins on the case-studies page. Wins are easy to dress up. Losses are not. A vendor who can describe a recent test that failed, why it failed, and what they learned, is a vendor running a real testing programme. A vendor who has no losses to share is either lying or not testing.
What good looks like. The honest answer is a specific story. "We ran a free-shipping-threshold test on Client X in Q3. Our hypothesis was that lowering the threshold from £75 to £50 would lift conversion 8%. It dropped it 3% because cart value compressed faster than conversion lifted. We rolled it back in week three." That's an operator answer.
Answers that disqualify. "We don't really lose." "Most of our tests win." "We're very selective about what we test, so failures are rare." All three are tells that the testing programme is theatre, not science.
Question 4: "Who sets the test hypotheses, the operator or the tool?"
Why it's diagnostic. This is the single biggest determinant of outcome. The Build Grow Scale 347-store research found that the SAME AI tooling produced 4-7% lift when the tool set the hypotheses and 28-34% when the operator set the hypotheses (Stafford, 2026). The software is identical. The hypothesis layer is everything.
What good looks like. "A senior strategist reviews every audit and writes the hypothesis list. The AI generates variants and runs the tests. The operator calls the winner." That's the operator-led model. GoGoChimp calls it OperatorAI (the agency's CRO methodology, distinct from OpenAI's Operator agent product released January 2025).
Answers that disqualify. "The AI handles it." "Our platform automatically prioritises tests for you." "It's all AI-driven." If the operator isn't in the loop on hypothesis design, you're buying a £5,000/month VWO licence with a markup.
Question 5: "What's your sample-size methodology?"
Why it's diagnostic. Sample-size calculations are the boring discipline that separates real CRO from theatre. A serious agency will tell you their minimum sample size per variant, their pre-test power calculation method, and how they handle interim results.
What good looks like. "We use a power calculation based on baseline conversion rate, minimum detectable effect, and 80% power at 99% significance. Minimum sample size per variant runs 5,000-30,000 sessions depending on the test." That's a competent answer. It's also a boring answer. Boring is fine.
Answers that disqualify. "We let the platform decide when to call it." "Two weeks is usually enough." "We end the test when significance hits 95%." That last one is the most dangerous because it sounds rigorous. Peeking (calling a winner the moment significance is hit) inflates false positive rates substantially (Baymard Institute on common testing errors).
Question 6: "How do you decide which test to run first?"
Why it's diagnostic. Prioritisation is where good CRO programmes find their compounding edge. The best agencies use a prioritisation framework. ICE (Impact / Confidence / Ease) is the simplest. PIE (Potential / Importance / Ease) is Conversion.com's published framework (Conversion.com on PIE methodology). PXL (Bayesian-weighted prioritisation) is more sophisticated.
What good looks like. A named framework, applied to your specific audit, with reasoning. "Your homepage hero gets 47% of inbound sessions and converts at 1.8%. That's the highest-impact test by 4×. We'd run it first."
Answers that disqualify. "We test what feels important." "We start with the easy wins." "Our strategist has good instincts." None of those are frameworks. They're vibes.
Question 7: "Who's the named contact who will personally read my audit?"
Why it's diagnostic. This is the agency-versus-operator test. At a real operator-led agency, a senior person reads your audit personally and signs the document. At a large agency, you're typically handed to an account manager who didn't conduct the audit and doesn't know your site.
What good looks like. A first name. A LinkedIn URL. A specific role. "Chris McCarron, Founder, reads every audit on engagements under £10K/month. On larger engagements he co-reads with the senior strategist." That's an operator answer. The full hiring framework for vetting an agency is in our companion guide on what to look for in a conversion rate optimisation agency.
Answers that disqualify. "You'll be assigned an account manager once you sign." "Our team will handle it." "Our process is collaborative across the agency." All three mean the person who sells you the engagement is not the person who delivers it. That's not always disqualifying, but it should drop your willingness-to-pay substantially.
Account-manager-led engagements are where the operator hands you to the junior. The free audit at /audit is read by the founder. The retainers are run by the founder plus one senior strategist. That's the entire delivery surface.
One more diligence question worth asking: ask the vendor for the benchmark they hold themselves to. The honest answer for landing-page conversion is Unbounce's Conversion Benchmark Report, which puts the cross-industry median at 6.6% across 41,000 landing pages, 464 million pageviews, and 57 million conversions. If a vendor quotes higher industry medians without naming the dataset, they are either cherry-picking a vertical or using aggregator numbers that inflate the baseline. The follow-up is to ask what conversion lift they would expect against the 6.6% floor for your specific vertical.
Section 4: Red flags to walk away from
Five patterns appear often enough in the UK and US CRO market that they deserve naming.
Red flag 1: "Performance-guaranteed" language without an outcome definition
Several agencies advertise "performance-based pricing" or "results-guaranteed" pricing. The pitch sounds founder-friendly: "we only earn if we deliver results."
Read the fine print. The "results" are almost always defined in a way that makes the guarantee meaningless. "We guarantee at least one winning test per quarter." A winning test at 95% significance is not the same as revenue lift. "We guarantee 5% conversion lift." But conversion is measured against a baseline you don't get to verify. "We guarantee positive ROI." Measured how? On what attribution window?
A real performance guarantee names the outcome (incremental revenue), names the measurement window (12 months post-engagement), names the attribution method (controlled cohort, holdout group, or matched-market), and names the consequence of missing it (fee refund, free month, or specific compensation). Anything looser is marketing.
Red flag 2: £15,000+ engagements for page-speed-only work
Page speed is engineering. It's high-leverage engineering. BeeFriendly's page-speed work was the single largest revenue intervention in GoGoChimp's history at roughly 30× revenue growth from a 2.24-second speed reduction (youtu.be/z2bjGvAkqn0). But the work itself is bounded. An experienced operator can ship page-speed transformation on a Shopify store in 1-3 weeks. The right price for that is £1,500-£3,000 productised. The Speed Sprint exists at exactly that price point.
Agencies charging £15,000+ for page-speed-only engagements are pricing on revenue impact (which is legitimate value-based pricing) but in a market where productised alternatives at one-tenth the cost exist. Pay the £15K only if the engagement bundles ongoing optimisation or extensive custom engineering. Otherwise, run the productised version first.
Red flag 3: No published case studies with verifiable receipts
A case study without verification is fiction. Verifiable means at least one of:
- A Loom or video walkthrough of the actual analytics dashboard showing the result
- A third-party citation (Trustpilot review, podcast appearance, named press piece)
- A named client who has consented to be cited publicly
GoGoChimp's case-studies page lists eight named clients with public-naming permission. The BeeFriendly result has a public video. Enzymedica has a Loom analytics walkthrough. VectorCloud's 29.57% landing page conversion has the original Unbounce A/B Test Centre screenshots from February 2018.
If a vendor's case studies are all anonymous ("Client A in the SaaS sector"), and they have no podcast appearances, no Trustpilot reviews, no named press, walk. The case studies might be real. They might also be aspirational.
Red flag 4: Significance threshold not disclosed
If the vendor's website does not name a statistical significance threshold, and the proposal does not name one, ask. If the answer is vague, walk. This is the cheapest tell of an unserious testing programme.
Red flag 5: Account-manager-led from day one
If the first conversation is with a sales rep, the second conversation is with an account manager, and you have never spoken to the person who will actually do the work, the engagement is structurally misaligned. The operator who runs the audit, sets the hypotheses, calls the winners is the person whose attention you're buying. If you never speak to that person directly, you're buying a process, not an operator.
This isn't disqualifying for very large engagements (£20K+/month) where the agency has a depth of senior bench. It is disqualifying at every smaller bracket.
Section 5: What good CRO costs in 2026
Pricing in the CRO market is opaque by design. Most agencies don't publish rates. This section publishes them.
The numbers below come from GoGoChimp's published pricing, the productised market visible on agency landing pages, and qualitative bracket framing for retainer agencies that don't publish exact rates (Conversion.com, Speero, and the comparable UK / EU mid-market bench).
5.1 Productised entry: £500-£2,500 one-off
The lowest legitimate price for actual CRO work in 2026.
At this tier, you're buying one operator's time for a discrete project. There's no ongoing relationship. The work ships, you implement, the engagement ends. Frictionless Commerce, GoGoChimp, and a handful of solo operators on Twitter run this tier well.
5.2 Hybrid (productised + light retainer): £2,500-£5,000/month
The middle bracket where most £200K-£1M/month businesses land.
What's included: 8-15 tests per quarter, monthly written report, quarterly strategy call, ongoing page-speed and analytics monitoring, copy testing across landing pages and product pages.
GoGoChimp's Growth tier sits here at £2,500/month with a 3-month minimum. Comparable UK and EU CRO-focused agencies generally run in the £3,000-£5,000/month range with variable inclusions; ask for the inclusions list in writing before signing, because what one agency calls "8-15 tests per quarter" another counts as "8-15 hypotheses generated."
At this tier, you're buying programme rhythm and an operator-led hypothesis layer. The expectation is sustained 15-25% conversion lift over 6-12 months.
5.3 Full operator-led retainer: £5,000-£10,000/month
The bracket where the 28-34% lift figure becomes realistic.
What's included: 20-35 tests per quarter, dedicated senior strategist, AI personalisation, customer journey optimisation, autonomous testing agents, monthly written report plus monthly review call, 90-day performance guarantee on the top tier.
GoGoChimp's Scale tier is £5,000/month. Conversion.com and Speero both operate above that floor for ecommerce and SaaS at scale, with mid-bracket retainers generally starting in the high four figures monthly and scaling into five figures depending on test volume and seniority of bench (Conversion.com, Speero). At five-figure monthly retainers and above you're in enterprise CRO territory, where a handful of UK and US specialist agencies compete on senior bench depth rather than tooling.
"There's few agencies that can do what GoGoChimp achieve. I really appreciate everything you've done to grow my business."
Neil Patel, co-founder, CrazyEgg (engagement: Patel's YouTube channel, not the CrazyEgg platform)
5.4 Regional pricing variation
UK pricing generally runs lower than US equivalents for comparable scope. EU pricing tends to sit at or just below UK rates on average, with the Nordics running higher. The differential is partly currency, partly cost-of-living, and partly the deeper US CRO agency market driving competitive premium pricing.
If you're a US founder buying from a UK or EU operator, you're often getting the same operator quality at meaningfully less than the equivalent US bench rate. The remote-delivery norm post-2020 makes geographic arbitrage straightforward.
5.4 The form-field arithmetic that justifies the retainer
The Baymard Institute's checkout-usability research found the average ecommerce checkout carries 14.88 form fields, roughly twice the 6-8 actually required for the vast majority of orders (Baymard checkout usability research). Most checkouts can lose 20-60% of their form elements without losing required data. A retainer that ships one well-tested checkout simplification per quarter typically pays back at the £200K/month revenue bracket on that single test alone. The maths is not about cleverness; it is about removing structurally unnecessary fields that compound abandonment.
5.5 What pricing should not look like
- Hourly billing. CRO is value-based work. Hourly billing structurally misaligns the agency from outcome.
- Setup fees above 20% of monthly retainer. Large setup fees are sales-trap pricing.
- "Pricing on application" with no published anchor. Reputable agencies publish at least a starting range.
- Performance-only pricing. As covered in Section 4, this is almost always structurally unfavourable to the buyer.
Section 6: A worked example for a £200K/month Shopify store
This is the buying decision for a real-shape business. The numbers below are typical for the bracket; adjust to your specifics.
The setup
- Revenue: £200,000/month gross
- Sessions: 180,000/month
- Conversion rate: 1.8%
- AOV: £62
- Ad spend: £35,000/month across Meta and Google
- Tech stack: Shopify, Klaviyo for email, Hotjar for heatmaps, GA4 for analytics
- Team: Founder, one in-house marketer, one part-time designer, dev work outsourced
The bottleneck
The free audit identifies three problems, ranked by impact:
1. Homepage LCP at 5.8 seconds on mobile. Bounce rate on mobile homepage is 68%.
2. Product page above-the-fold copy is generic ("Premium quality, handcrafted") rather than benefit-led.
3. Checkout has a guest-checkout option but it's hidden three clicks deep.
The buying decision
This founder is in the bracket where Path B (productised) makes sense as the first move, with Path C (retainer) as the natural follow-on if the productised work pays back.
Step 1. Speed Sprint (£1,500 one-off, week 1-3). Fix the homepage LCP. Affordable Golf's parallel example: 21.3s to 6.1s on the same intervention type in March 2026. Expected conversion lift on this specific business: 12-18%.
Step 2. Headline Lab (£500, week 4). Run 8 AI-generated headline variants on the product page hero. Implement winner. Expected lift: 5-10%.
Step 3. Decision point (week 8). If the two productised interventions have delivered combined 15-25% conversion lift (the typical outcome), the maths for retainer becomes obvious. Conversion at 1.8% to ~2.1% means revenue at £200K/month to ~£235K/month. £35K/month incremental revenue covers a £5K/month Scale retainer at 7× ROAS in the first month.
Step 4. Scale retainer (£5,000/month, month 3 onwards). Continuous testing programme. 20-35 tests per quarter. Expected sustained lift over 12 months: 28-34% per Build Grow Scale's 2026 research.
The 12-month maths
This is the buying path the paper recommends. Start productised. Earn the retainer with results. Compound from there.
Section 7: Conclusion and next step
If you've read this far, you have what you need to make the call.
The right buying path for most founders is to start with an audit. Take three audits if you can. GoGoChimp's free audit at /audit is read personally by the founder, not handed to an account manager. It will name the bottleneck, name the buying path that fits, and tell you which of the three legitimate paths matches your bracket.
Even if the answer is "don't hire anyone, fix this one Liquid file and run more ads," the audit will tell you that. The audit's job is to tell you the truth about your funnel. Not to qualify you into a retainer.
If the audit names a discrete bottleneck, run the productised intervention. £500-£2,500 one-off. Implement. Measure. Decide whether to graduate.
If the audit names compounding bottlenecks at £200K+/month revenue, the retainer path becomes the right call. The Growth tier (£2,500/month) is the floor. The Scale tier (£5,000/month) is where the 28-34% lift figure becomes realistic with the 90-day performance guarantee attached.
Buy when buying is the answer. Not before.
Next step: book the free audit at gogochimp.com/audit. It runs 48 hours from submission to delivered Loom walkthrough. If you'd rather see the productised options first, the catalogue is at gogochimp.com/services.
Frequently Asked Questions
When should I NOT buy CRO in 2026?
Don't buy CRO if you're under £50K/month revenue, if your traffic is below roughly 30,000 monthly sessions on the page you'd want to test, if you've replatformed in the last 60 days, or if you don't have GA4 and Microsoft Clarity properly installed. Below those thresholds the maths doesn't work, and the right move is to spend the CRO budget on acquisition or analytics infrastructure first.
How much should I expect to pay for CRO in 2026?
Three brackets. Self-serve AI tools run £200-£500/month and suit founders with 100,000+ monthly sessions and in-house testing bandwidth. Productised one-off engagements run £500-£2,500 for a single fixable bottleneck. Operator-led retainers run £2,500-£10,000/month at the mid-market and into five figures monthly at enterprise. GoGoChimp's Sprint is £2,500 one-off, Growth £2,500/month, Scale £5,000/month.
What's the difference between productised and retainer CRO?
Productised CRO is a fixed-scope, fixed-price one-off intervention. The vendor diagnoses a specific bottleneck, ships the work, walks away. Retainer CRO is a monthly engagement that runs continuous experimentation across the site, with operator-set hypotheses, 20-35 tests per quarter, and compounding lift over 12-36 months. Start productised if you have one named bottleneck. Graduate to retainer once the productised work pays back.
What conversion lift should I realistically expect?
Build Grow Scale's 2026 research across 347 ecommerce stores found self-serve AI tools deliver 4-7% lift on average. Expert-guided AI on the same software delivers 28-34%. Productised page-speed work typically delivers 10-25%. Headline rewrites typically deliver 5-15%. Pricing tests run 10-30% with high variance. Anyone promising 100%+ lift without a named client and a verifiable receipt is selling marketing, not CRO.
How do I tell an operator-led agency from a salesperson-led one?
Send the seven questions from Section 3. The signal is in the answers, not the website. Ask for a named loss, not a named win. Ask who reads your audit personally. Ask what the statistical significance threshold is, who sets the hypotheses, and what the sample-size methodology is. Operators answer in specifics with frameworks named. Salespeople answer in adjectives.
What's the role of AI in 2026 CRO?
AI generates variants and runs tests; the operator sets hypotheses, prioritises tests, and calls winners. The Build Grow Scale 347-store research showed the same AI tooling produces 4-7% lift when the tool drives strategy and 28-34% when the operator does. The software is identical. The hypothesis layer is the differentiator. GoGoChimp's OperatorAI methodology (distinct from OpenAI's Operator agent product released January 2025) is the canonical example of the operator-led model.
How long until CRO pays back?
Productised page-speed engagements typically pay back within 30-60 days at the £200K+/month revenue bracket. Productised headline work pays back within 30 days when traffic is sufficient. Retainer programmes typically show first compounding signal in months 2-3 and full ROI by month 6. The worked example in Section 6 (£200K/month Shopify store) hits 14× annualised ROI by month 12 on a £52K total investment. Below £100K/month, payback windows extend; below £50K/month, the maths often doesn't close at all.
References
- Stafford, Matthew. (2026). 2026 CRO Year in Review: What Worked, What Failed, What's Next. Build Grow Scale, 9 April 2026. https://buildgrowscale.com/cro-trends-2026-recap
- Baymard Institute. (2024). E-commerce UX Research and Cart Abandonment Benchmarks. https://baymard.com
- Conversion.com. (2026). PIE Prioritisation Methodology and Service Pricing. https://conversion.com/blog/
- Frictionless Commerce. (2026). Productised Teardown Service Listings. https://www.frictionlesscommerce.com/
- Speero. (2026). CRO Retainer Service Methodology. https://speero.com/
- GoGoChimp. (2026). BeeFriendly Skincare Case Study Video (anonymised). https://youtu.be/z2bjGvAkqn0
- GoGoChimp. (2021). Enzymedica UK Loom Analytics Walkthrough. https://www.loom.com/share/d20fd92f4d5e49a88a92c9c0d5e28570
- McCarron, Chris. (2026). Website Lead Capture Choices That Lift Quality and Conversion. CMO Times, Q&A. https://cmotimes.com/qa/website-lead-capture-choices-that-lift-quality-and-conversion/
- GoGoChimp. (2026). Pricing Tiers (Sprint / Growth / Scale). https://www.gogochimp.com/#pricing
- GoGoChimp. (2026). OperatorAI Methodology. https://www.gogochimp.com/methodology
About the author
Chris McCarron founded GoGoChimp in June 2013. Thirteen years of operator (not consultant) experience across ecommerce, SaaS, and non-profit conversion rate optimisation. Creator of OperatorAI (GoGoChimp's CRO methodology, distinct from OpenAI's Operator agent product released January 2025). Nominated for Digital Doughnut Digital Marketing Agency of the Year 2021. Based in Glasgow, Scotland; works with clients across the UK, EU, and US.
Client outcomes include BeeFriendly Skincare ($48K/year → $1,447,225/year), Enzymedica UK (3.4% → 16.9% Black Friday conversion), Super Area Rugs (216% revenue increase in 37 days), EM360 (0.12% → 7% B2B conversion in 30 days), and Donate For Charity (494.64% donation lift in 30 days). Full case studies at gogochimp.com/case-studies.
Want us to do this for your site?
Book a free AI audit. 15 minutes. We’ll show you three things your site is missing and what we’d test first.
Book my free AI audit →
