The cheapest conversion lever on a mid-market store is almost never a new tool. It is the one the store already owns and has quietly broken. I have audited enough enterprise checkouts in the last ten years to say this out loud: the biggest, most repeatable lift most teams can book this quarter is still an easier guest checkout.
Nobody wants to hear it. The marketing lead wants a personalisation platform. The CTO wants a replatform. The CFO wants a line item they can defend on a board slide. “Make guest checkout work properly” fits none of those slides. It is also where the money actually is, and it is where the published research has pointed for over a decade.
So this piece is deliberately narrow. One expert, one operational problem that costs money every day, and three levers a buyer can ask their agency about on Monday morning. The expert is Christian Holst, co-founder and research director at the Baymard Institute, whose team has been measuring checkout UX for more than a decade and whose figures sit behind most of the industry statistics you have read about this subject.
Holst’s own line: still offer the guest checkout, even when you want the account
Here is the catch-22 Holst keeps rediscovering in testing, in his own words, from a 2026 Q&A on checkout optimisation: “To be clear, you should still offer a guest checkout. Not all users will care about ‘collecting points’. But describing the benefits of the discount program near the optional password field is a great way to increase voluntary account creations.” (Wingify, 17 April 2026.)
Two things sit inside that sentence that are worth pulling out, because they are the entire argument. The first is that account creation and checkout should not be the same step. The second is that the account is still worth having, so the aim is not to abolish it but to decouple it from the moment the customer is trying to pay. The teams that keep conflating the two are the ones running the highest abandonment rates in their category.
Baymard puts the number at 70%, and names the drivers to stop
The headline figure from Baymard’s rolling meta-analysis of 50 separate studies, last updated in September 2025, is a 70.19% average cart abandonment rate. Baymard’s own quantitative survey of US shoppers, excluding the 43% who were “just browsing,” breaks the reasons down roughly like this: 40% extra costs too high, 20% delivery too slow, 19% trust concerns on card details, 18% site required an account, and 17% the checkout was too long or complex. (Baymard, cart abandonment rate data.)
Three of those five sit inside the checkout flow itself and can be moved by the store team without touching product, logistics or brand trust. The account requirement is the cleanest one to fix, which is why it is the first lever, and why Holst’s team keeps writing about it ten years after the ASOS case made it textbook.
Lever one: stop asking for a password before you have taken the money
The reference case here is still ASOS, and it is still the sharpest proof you can hand a sceptical CFO. James Hart, then ecommerce director at ASOS, reported that removing the explicit “register” language from the checkout login screen and replacing it with a simple “continue” button produced “a 50% decrease in abandonment of the site at this page”. The account was still created in the background; what changed was the language and the sequence. (Econsultancy, 23 February 2015.)
Fifteen years on, you can still visit mid-market checkouts that force a password field as the second step. Some of them are running paid acquisition at double-digit cost per click. Fixing this is one engineering ticket and a copy change. It is the single highest-return change most teams I review have on the table, and most of them are not going to get to it this quarter because it is unglamorous.
A note on sequencing that is often missed: the ASOS fix did not abolish the account, it moved it. The customer still ended up with one, created silently during payment. If the business genuinely needs the account for CRM, loyalty or returns management, this is the pattern. The business keeps everything it wanted. The customer keeps the lower friction. Nobody argues over “guest versus registered” because the question was wrong.
Lever two: six to eight fields, not twenty-three
Holst’s second published position is on the form itself. On a 2019 piece he updated through 2025, he writes that “sites can achieve a checkout flow with as little as 6-8 form fields (for a guest checkout flow incl. the credit card fields)”. (Baymard, Christian Holst, Research Director.) Baymard’s own benchmarking of US checkouts puts the typical average at roughly double that. The gap between the two is one of the most consistent findings in their decade of research, and it is one of the cheapest to close.
What usually sits in the twelve extra fields is a mix of three things: fields the business thinks it needs but never queries (second phone number, company name on a B2C flow, “how did you hear about us”), fields that could be inferred (city and state from postcode), and fields that could be shown later (gift wrap options at the point of fulfilment, not at the point of pay). None of them pays for itself in incremental revenue. All of them cost you a measurable share of completions.
Lever three: an accelerated path, with the customer’s details already on file
Once the account requirement is gone and the form is honest, the next question is whether to add an accelerated option for repeat buyers. The most recent and most honestly reported figure here is PayPal’s own case study of Black Forest Décor, published 13 September 2024. In an A/B test running from 3 April to 15 June 2024, their Fastlane cohort converted at roughly 86% against the non-Fastlane guest cohort at roughly 74%, with checkout time dropping from about 3.9 minutes to under 2 minutes. Dustin Dupus, COO and CFO at Black Forest Décor, is quoted on the record: “We are big fans of A/B testing. We wanted to test Fastlane’s effect on conversion, while ramping slowly and analyzing the data.” (PayPal case study, Black Forest Décor.)
Two caveats a buyer should hear. First, the baseline matters: a 74% completion rate on a guest cohort is already very high, which suggests the gain comes partly from Fastlane and partly from a flow that was already clean. On an unoptimised flow a Fastlane-style accelerator will not rescue the broken fundamentals above it. Second, Shopify’s widely cited “Shop Pay lifts conversion by up to 50% versus guest checkout” number comes from a 2023 external study whose methodology, sample size and control group were never published; the figure Shopify printed in its own Princess Polly case study was a far more modest 4.1%. (Clean Commit, Shop Pay conversion analysis, citing Shopify.) The honest advice: expect single-digit to low double-digit lift from accelerated checkout, and build the flow underneath first.
The lever that still does not work: the 43% who were never going to buy
The honest symmetric point, in a format that is supposed to tell a buyer what to do and what not to bother doing: 43% of abandonments in Baymard’s own study came from shoppers who said they were “just browsing” or “comparing prices.” No checkout change touches those. They are a product problem (does your merchandising let a browser self-qualify out early?), a pricing-transparency problem (can they see landed cost before adding to cart?) and a retargeting problem (can you earn a second visit?). Trying to solve them with checkout UX is a category error, and I have watched more than one well-funded project aim at exactly the wrong line in the data.
Read against that, the five in-flow drivers Baymard measures (shipping cost, delivery speed, trust, account requirement, flow complexity) are the ones a checkout team can actually move. Those are the levers, and they are the ones agencies get paid to pull. The rest sit upstream.
What this looks like inside a WooCommerce store
Translating this into a build on WooCommerce is less work than most teams assume, which is why the frustrating thing about it is that so many stores are still running the broken version. In rough order of effort:
- Allow guest checkout at the setting level. WooCommerce ships with an account requirement toggle under Settings → Accounts & Privacy. On most stores we audit, “Allow customers to place orders without an account” is off because somebody turned it off three years ago to grow the mailing list. Turn it back on. Catch the email after the order, in the order confirmation flow, with an opt-in that is honest about what the account gets them.
- Trim the checkout fields. The
woocommerce_checkout_fieldsfilter lets a developer hide, re-order, or make optional every default field. On a B2C store on a single market, a safe starting point is email, name, address line one, postcode, city, country, phone. That is seven. If the business wants company name, second address line or an order note, add it back when there is evidence it earns its place in incremental revenue, not before. - Add one accelerated path, not three. Shop Pay, PayPal Fastlane, Stripe Link, Apple Pay and Google Pay all install cleanly on WooCommerce via their own extensions (WooPayments, PayPal Payments, Stripe for WooCommerce). Pick the one that matches your customer base and payment processor. Running all five at once produces decision paralysis in the shopper and operational noise for the finance team. The right posture is: default flow is honest and short, one accelerated button is offered above it.
- Measure the two moments. Enhanced ecommerce events in GA4 (
begin_checkout,add_payment_info,purchase) will tell you the drop between the start of checkout and payment entry, which is where the account-requirement penalty lives. Baseline those two numbers before you make the changes, so you have an honest before-and-after. In my experience the first lever movesadd_payment_infowithin a week.
None of the above is novel. All of it works. The reason it is still worth writing is that the mid-market stores I review are, as a rule, still breaking at least three of the five Baymard-measured in-flow drivers. The reason is not technical. It is that fixing them has no champion in the room: it is nobody’s quarterly target, and it does not show up on a vendor slide. That is where an outside eye earns its fee.
If your checkout has not been audited against this list in the last year, that is where I would start, before any discussion of personalisation, replatforming, or a new payments stack. The question I would put to your current supplier is simple: in the last twelve months, which of the five Baymard drivers did we move, by how much, and what did it cost to ship? If the answer is a shrug, you already know what the next conversation is about.
Last modified: October 17, 2026
