CRO consultant vs agency: what buyers actually ask for
Every comparison of CRO consultants and agencies is written by one of the two. Here's what 56 real job posts ask for instead, and a test to run on your own funnel.
That’s the short version. The rest is the posts’ own wording, what the published comparisons actually organise around, and a test you can run on your own situation.
Where this comes from
Over September 2026 I read 56 Shopify-related job posts from my Upwork feed, and copied each one’s wording before writing anything about it. By my grouping, 15 of them were for conversion work.
That’s a small sample from one freelancer’s feed, not a survey. It shows how these buyers wrote their posts. It can’t show who they hired or how it went.
What the published comparisons organise around
I read the first-page results in September 2026. All six I opened were published by a company selling something in this market:
Who writes the comparisons
- Convert — an A/B testing platform
- Kirro — an A/B testing platform, €99/month, positioned as the DIY alternative to consultants
- The Good — a CRO services agency
- Omniconvert — three CRO and testing products, plus consulting
- Clicks Geek — a performance marketing agency selling CRO as part of a wider retainer
- Invesp and ConversionRate.store — CRO agencies
Not one is a neutral party, and none of them cites buyer evidence. The closest is a salary figure from Glassdoor.
Three of the six do mention the difference between a hire who implements and one who only recommends. It’s worth being exact about where they put it:
Kirro lists it as red flag number six of seven: “That’s not consulting. That’s a PDF report at consultant prices.” Its actual decision rule is your monthly conversion volume — under 500 conversions DIY, 500 to 2,000 buy an audit and test yourself, over 2,000 hire fully.
Clicks Geek frames it as a feature difference between two things it sells: consultants give “strategic guidance” your internal team executes, agencies give “full execution handled externally.”
Convert treats it as a sub-type of freelancer: “a freelance CRO strategist is very different from a freelance experiment developer. One shapes your program; the other executes within it.”
So they all know the distinction exists. None of them makes it the thing you decide on. The organising axis is size, cost, or traffic volume — and in every case it’s an axis that maps onto what the publisher sells.
What the posts ask for
The buyers put it first. Five of the posts asked for the same hire to find the problem and fix it, and ruled out the alternative in writing:
“personally implement and run A/B tests to fix it — not just hand us a slide deck of recommendations. This is a doer role, not a consulting role.”
“We’re looking for a CRO specialist who can both diagnose and implement. We don’t want a 40 page audit that we then have to find a developer to action.”
“This is not a design-only role. We need someone who can analyze the customer journey, recommend improvements, design when needed, and then implement the changes”
“tell us what’s costing us sales, then help us fix it”
“someone who can own the CRO thinking and commercial strategy as well as execute the required changes”
Notice what the first one does with the word “consulting”. To that buyer, consulting means advice without the work: the slide deck. The line they drew wasn’t between one person and a team. It was between someone who hands over recommendations and someone who carries them through.
A consultant can stop at the deck. So can an agency. That’s why the size question answers less than it appears to.
Two posts that show the other side
Two single posts, not patterns, but they’re worth seeing.
One merchant already had a performance agency, and that agency had done the diagnosis. Their own words: “Our performance agency has identified a current conversion rate of 0.34%… with the largest drop-offs occurring at product discovery, add-to-cart, and checkout.” The post was hiring someone to take those findings and improve conversion within the existing store.
And one post was written by an agency, hiring a freelance builder for a “CRO pass on every new client store off our checklist.”
So the lines between “agency” and “consultant” aren’t as clean as the comparisons suggest. Some agencies diagnose and hand off. Some hire individuals to do the work. The label on the proposal tells you less than what’s actually in the scope.
A test to run on your own situation
Before comparing anyone’s prices, answer three questions about your own store. This is my reading of what the posts above were getting at, not a rule they stated.
Three questions, in order
- Does a diagnosis already exist? Someone has looked at your data and can say, in writing, where visitors drop off and why. Not a list of best practices — your funnel, your numbers.
- If you got a list of recommendations tomorrow, who would build them? If the honest answer is "we'd have to find someone", that's exactly the situation one buyer described: a 40-page audit and then a search for a developer.
- Is the remaining work known, and steady? The fixes are decided, there are a lot of them, and they run across many pages or stores week after week.
What the answers point to:
- No diagnosis yet — hire whoever will own the diagnosis and the fix. One person or a team matters less than one owner. Ask any candidate, consultant or agency, the same question: “Who implements what you recommend, and who checks it worked?”
- Diagnosis exists, and the work is known and steady — this is where the answer flips. You’re no longer buying a diagnosis, you’re buying capacity, and capacity is what a team sells.
- Recommendations with no one to build them — the buyers above had a name for that outcome, and none of them wanted it.
I can’t tell you which of these produces better results. Job posts don’t show outcomes, and I’d be suspicious of anyone who said they could settle it from the outside, including me.
What 56 job posts can’t tell you
Who applied. Who got hired. Whether diagnose-and-implement hires actually delivered more than a separate audit and build. Whether these buyers were right about what they needed.
They’re one kind of evidence: what people wrote when they were about to spend money, which is at least not written to sell you anything.