Roughly ten weeks before Black Friday is the last point at which hiring conversion help can change the outcome of that weekend. After about six weeks, a testing programme cannot produce a readable result in time, and what you are buying is execution of known fixes rather than optimisation.
Both are legitimate purchases. They are different purchases and should be priced differently.
The arithmetic that sets the date
Test results are governed by the calendar, not by effort or budget.
The median test in DRIP’s database ran 42 days, and benchmarks collected by roast.page put the median nearer 23 days. Take the optimistic figure. A programme starting in mid-October has an audit and roadmap in the first fortnight, a first test launched in week three, and a result somewhere around week six, which is the second week of November.
That leaves time to ship one winner. It does not leave time to ship a second, and most tests do not win: Optimizely’s analysis of more than 127,000 experiments puts the average win rate near 12%.
| Weeks before Black Friday | What you can realistically buy |
|---|---|
| 12+ | A full programme. Research, roadmap, several tests read before the freeze. |
| 8 to 12 | Audit plus one or two tests. One winner is a reasonable expectation. |
| 6 to 8 | Audit plus execution of documented fixes. Testing is unlikely to read. |
| Under 6 | Execution only. Fixes, speed, offer setup, post-purchase. |
| Under 2 | Nothing new. Freeze and protect what works. |
What is still worth doing at six weeks
Quite a lot, and none of it requires a test to justify, because the research already exists.
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, against an average abandonment rate of 70.22%. If those are present on your store, fixing them is not optimisation, it is repair.
Speed is the same. Only around 42% of mobile sites pass all three Core Web Vitals, and Google and Deloitte’s analysis of over 30 million mobile sessions associated a 0.1 second improvement in mobile load time with an 8.4% rise in retail conversion and a 9.2% rise in average order value. Mobile carried 76% of BFCM traffic and 69% of orders in 2025, so this is where the volume is.
Accelerated checkout is a toggle. Shop Pay’s conversion improvement runs up to 50% against guest checkout and 91% on mobile.
And the post-purchase offer, which cannot cost you the order you just took. Across our accounts these add around 10% to average order value.
What an honest agency should tell you in October
That a testing programme started now will mostly pay off after Black Friday, not during it.
A firm that sells you a full testing retainer six weeks out, knowing the first result lands in December, is selling you something you cannot use in the window you are buying it for. Ask directly: what will be live and proven before 20 November, and what is realistically a January result.
A firm that has never turned down a client on timing grounds has capacity rather than a method.
The better trade if you are late
Buy execution now and start the programme in December.
December is a genuinely good month to begin. Traffic is still elevated, you have just generated the richest dataset of your year, and the freeze that protected peak has no reason to extend past it. Most brands restart in January and give up the best remaining sample of the year.
The peak weekend also hands you research for free: conversion by device, average order value against your threshold, post-purchase take rate, top products by revenue, and every support ticket theme. That is a month of customer research collected in four days, and it is the best possible input to a roadmap.
What to have ready if you do hire now
Tool and analytics access in week one. Your ninety-day baseline numbers written down. Margin rank order, supply constraints and anything legal has to approve. And one named person who can approve a change without convening a meeting.
That last one decides more than the agency choice does. On a compressed timeline it is the difference between shipping two fixes and shipping six.
Frequently asked questions
Is it too late to hire a CRO agency before Black Friday?
For a testing programme, roughly ten to twelve weeks out is the practical cutoff. Under six weeks you are buying execution of documented fixes, which is still worthwhile and should be priced differently.
What can be fixed in the last month before peak?
Checkout cost visibility, guest checkout, mobile speed, accelerated payment and a post-purchase offer. None require a test to justify.
When should testing stop before Black Friday?
Around two to three weeks out. A test running through peak splits your most valuable traffic with a variant you were not confident enough to ship.
Is December a bad time to start a conversion programme?
It is one of the best. Traffic is still elevated and you have just collected the richest customer dataset of your year.
The execution list is in our BFCM CRO checklist. What a full engagement covers, and the conditions under which it is the wrong purchase, is set out under the CRO Program and website conversion optimization.