Yes, we put ourselves on our own list. So does every agency listicle you have ever read; they are just quieter about it. So here are the criteria in the open, nine genuinely good alternatives described fairly from their own live sites, and the checks that let you verify anybody’s claims, ours included.
The criteria: does the shop actually build tests rather than slide decks; is QA documented rather than asserted; is flicker treated as an engineering problem; how broad is the platform coverage; and can you find out what anything costs without three discovery calls. Where a firm publishes numbers, we quote their site; where they do not, we say so. Every entry was checked against its live site this week, because agency positioning drifts fast enough that last year’s listicle is already describing companies that no longer exist in that form.
One structural note before the list: the market splits into three shapes, and comparing across them is how buyers end up disappointed. Dedicated build shops sell execution and assume you own strategy. Programme agencies sell outcomes and bundle the build inside. Strategy consultancies sell the thinking and often subcontract the code. Price a shape against its own kind, and be suspicious of any pitch that claims to be all three at once.
The dedicated build shops
These firms sell development as the product. If you already own strategy, in-house or through a CRO agency, this is the category you are hiring from.
- Optyv (that is us). Ecommerce-only test development and CRO engineering: client-side and server-side builds, QA documented per build, winner implementation included. 1640+ tests shipped. Weaknesses, honestly: not the cheapest option in this list, and deliberately narrow, so if your programme is B2B SaaS or your stack is exotic, several firms below fit better.
- BrillMark. A dedicated A/B test development agency, and the biggest pure-play name in it: their site claims 15,000+ tests and 200+ agencies and brands served across Optimizely, Convert, AB Tasty, Kameleoon, VWO and Adobe Target. Best for high-velocity build capacity on any platform; consider that strategy is deliberately not on the menu.
- EchoLogyx. A global development partner with QA as a first-class service and offices across four time zones, tool-agnostic to an unusual degree. Their site publishes per-client results rather than aggregate totals. Best for enterprise teams on unusual stacks wanting follow-the-sun coverage; consider that neither pricing nor headline numbers are published.
- ABTestDeveloper. Per-hour test programming and QA, advertised from 25 dollars an hour, with 8+ years and 1000+ campaigns claimed. Best for smaller agencies buying overflow capacity by the hour; consider that the process depth is lighter than the enterprise shops above.
- OptiPhoenix. A CRO and experimentation agency with engineering roots, selling QA-led test development at offshore economics, with 150+ brands claimed and Kameleoon and Convert certifications. Best for brands wanting strategy and build under one roof at a sharp price; consider that white-label agency work is not their stated focus.
Build included, programme attached
- Spiralyze. A predictive CRO agency that builds everything in-house with a team their site puts at 170, selling outcomes with a guarantee rather than capacity. Best for B2B SaaS teams wanting zero internal lift; consider that you cannot hire the build bench without the programme.
- Cro Metrics. An experiment-led growth agency claiming 35,000+ experiments over nearly two decades, with engineering capable of complex builds inside a broader marketing remit. Best for enterprises consolidating experimentation and growth in one partner; consider that a quiet white-label bench is not what they sell.
- Convertica. A white-label CRO service built specifically for marketing agencies, with development, design and the testing software bundled, and 1,100+ websites claimed over seven years. Best for agencies reselling done-for-you CRO under their own brand; consider that the packages centre on one platform, VWO.
A note on the offshore question, since several of these firms compete on it. Time-zone spread and lower rates are real advantages, and the good offshore-heavy shops have QA processes precisely because distance punishes sloppiness. The honest trade is overlap hours and context: a build partner nine hours away needs briefs twice as precise, and the cost of imprecise briefs lands on whoever writes them, which is you. Price the total, not the rate.
The strategy-first benchmarks
- Speero. The CXL spinout, selling experimentation programme design, research and culture rather than a coding bench. Best for maturing how your organisation tests; consider if what you actually need this quarter is builds.
- Conversion. The merged Conversion and Widerfunnel entity, describing itself as the world’s leading dedicated CRO agency, with 100+ specialists and enterprise clients. Best as the full-programme benchmark at enterprise budgets; consider if you are budget-constrained or need execution only.
How to verify anybody on this list
Every claim above comes from a company’s own website, which is exactly as reliable as that sentence implies. The good news is that build quality is checkable before you sign anything, in about a week, for the cost of one small brief. Aggregate numbers deserve particular scepticism in both directions: a firm claiming fifteen thousand tests is telling you about volume, not about what happened to yours, and a firm publishing no numbers at all may simply be honest about how hard they are to count.
Ask for a real QA document from a real build, with the client name redacted; a shop with a process has one in two minutes. Ask who exactly writes the code, because benches vary more than brochures. Ask them to show flicker handling on your own site, not a demo store. Ask how they prove a purchase event fires once. And ask who owns the code at the end, in writing. The wider decision of whether to outsource at all is its own question, and the honest arithmetic on prices sits in our agency cost guide.
One more honesty note: win rates and lift claims are the least comparable numbers in this industry, because the statistics behind them are so easy to flatter. Judge shops on process you can inspect, not outcomes you cannot audit. If ecommerce is your world and you want to run the five checks on us, our test development practice is where to start.
