When a CRO programme is the wrong thing to buy
Conversion optimization is the wrong purchase if you do not have enough traffic to read a test, if your product has no demand, if nobody internally can approve a change, or if the site is so broken that the fixes are obvious and do not need testing. In all four cases the money buys you a slower version of an answer you already have.
You do not have the traffic yet
Sample size is driven by conversions, not visitors. Below roughly a thousand conversions a month, classic testing struggles to detect the effects that actually occur.
The effects are small. Winners in DRIP’s database produced a median conversion uplift of 1.88%. Detecting a change that size takes tens of thousands of conversions per variant. Detecting a 20% change takes a few hundred.
That does not mean nothing can be done. It means the method changes: research, structural changes shipped whole, and before-and-after measurement with the confounds acknowledged. What it should not mean is paying for a testing programme that will spend six months producing inconclusive results while everyone gradually loses confidence in the discipline.
Nobody wants the product
Conversion optimization multiplies existing demand. It does not create it.
If your funnel is leaking because visitors arrive, look, and decide the product is not for them, that is a positioning or product problem and no amount of button testing reaches it. The tell is qualitative rather than quantitative: session recordings show people reading carefully and leaving, and post-purchase surveys return “I was not sure this was for me” rather than a specific objection.
Fix the offer first. Then optimise the funnel that carries it.
Nobody internally can approve a change
This one kills more engagements than the first two combined, and it never appears in the proposal.
If every change needs sign-off from three stakeholders, and one of them designed the current page, the roadmap will stall by week six. The agency will produce excellent briefs that never ship, and both sides will spend the engagement being frustrated with each other.
The fix is not a better agency. It is naming one internal person with the authority to approve a test without a meeting, before the engagement starts. If nobody in the organisation can be given that authority, the engagement will not work regardless of who you hire.
The problems are obvious and do not need testing
Some sites do not need a testing programme. They need three fixes.
Baymard’s meta-analysis of fifty studies attributes 48% of cart abandonments to unexpected extra costs at checkout, 19% to mandatory account creation and 18% to a checkout that is too long. If all three are present on your site, you do not need to test whether they are problems. They are documented, they are common, and testing them is spending weeks confirming what the research already established.
The same goes for speed. Only around 42% of mobile sites pass all three Core Web Vitals. If you are in the failing majority, fixing it does not need a test to justify.
Ship the known fixes. Then start testing, from a baseline that is not embarrassing.
What to do instead, in order
| Situation | Buy this instead |
|---|---|
| Under 1,000 monthly conversions | A one-off audit plus research, not a testing retainer |
| No demand for the product | Positioning and offer work |
| No internal decision-maker | Nothing. Fix the authority problem first. |
| Obvious documented problems | Development hours to ship the known fixes |
| All four resolved | A conversion programme |
The order these get fixed in
Traffic and demand first, because they sit upstream of everything else. Authority second, because it is free and it is the fastest fix on the list. Known problems third, since they need development hours rather than a retainer. Only then a programme.
Skipping to step four because budget happens to be available this quarter is the most common sequencing error in this category, and it produces an engagement everyone experiences as disappointing without anyone being able to say precisely why.
Why an agency telling you this is worth listening to
An agency that has never turned down a client has no methodology, only capacity. The question to ask any firm you are evaluating is which situations they decline, and to listen carefully to whether the answer is specific or evasive.
The scope, and the cases where it is premature, are set out at https://parahgroup.com/cro-program/
