Independent field guide

Turn the traffic you already have into sales

The Test Bench is an independent desk for conversion rate optimisation, written for small online businesses: store owners, service businesses and small software teams who cannot hire an agency and should not have to guess. We explain how to measure a funnel honestly, find the leaks worth fixing, and test changes without fooling yourself.

Start with what CRO is and how it works at small scale, or pick a bench below. Every guide opens with a test card: the page type, the goal, the evidence level and the first move.

A small online store's packing bench: stacked parcels ready to ship, a laptop and shelves of orders — The packing bench: where traffic becomes orders.

How the bench runs

01
Baseline
Measure the funnel as it is. No change is judged without a before.
02
Hypothesis
Write down what you expect to happen, and why, before touching the page.
03
Change
One change at a time, big enough to matter, small enough to read.
04
Reading
Run it long enough for the numbers to mean something at your traffic level.
05
Decision
Keep it, drop it, or test the next idea. Record what you learned either way.

Walk the benches

Bench 01 - Calibration benchCRO foundations and measurement What conversion rate optimisation actually is, how to measure a baseline honestly, and how to decide what to work on first when traffic is small. 12 guides Bench 02 - Arrival benchLanding pages The page a visitor lands on decides in seconds whether the visit continues. Message match, one clear offer, and a single job per page. 12 guides Bench 03 - Shelf benchProduct pages Where browsers become buyers or leave. Photos, price presentation, delivery information, reviews and the buy box, in the order visitors actually check them. 12 guides Bench 04 - Till benchCart and checkout The last steps of the funnel, where small frictions cost real orders. Carts that reassure, checkouts that stay out of the way. 12 guides Bench 05 - Price benchPricing pages How prices are presented decides how they are judged. Anchors, tiers, totals shown early, and discount framing that does not train buyers to wait. 12 guides Bench 06 - Capture benchForms and lead capture Every field is a small tax on the visitor. Shorter forms, better questions, honest error messages and enquiry forms that respect the person filling them in. 12 guides Bench 07 - Interrupt benchPopups and onsite campaigns Interruptions can earn their place or burn the visit. Timing, triggers, frequency and offers that respect the page the visitor came to read. 12 guides Bench 08 - Follow-up benchEmail capture and flows The visitors who leave today are tomorrow's easiest sales. Capture that is worth signing up for, and the small set of flows that do the follow-up work. 12 guides Bench 09 - Wayfinding benchSite search and navigation Visitors who cannot find the product cannot buy it. Menus, filters, search results and the zero-results page, treated as selling surfaces. 12 guides Bench 10 - Proof benchTrust and social proof Strangers buy from stores they believe. Reviews, guarantees, policies and proof elements placed where doubt actually happens, and kept honest. 12 guides Bench 11 - Message benchCopy and messaging Words are the cheapest test surface on any page. Headlines, buttons, value propositions and product descriptions written from the customer's reasons, not the seller's features. 12 guides Bench 12 - Split benchA/B testing on low traffic Small stores cannot run textbook tests, and pretending otherwise wastes months. Methods that respect small numbers: bigger swings, longer reads, honest stopping rules. 12 guides Bench 13 - Instrument benchAnalytics and behaviour tools The instruments of the bench: analytics set up to answer buying questions, heatmaps and recordings read properly, and surveys that ask at the right moment. 12 guides Bench 14 - Load benchSpeed and mobile conversion Most small-store traffic arrives on a phone, on a connection the owner never tests. Core Web Vitals in plain terms, and the mobile frictions that cost orders. 12 guides Bench 15 - Shopify benchCRO on Shopify Conversion work inside Shopify's walls: theme sections, product templates, the cart drawer, checkout customisation and what the platform decides for you. 12 guides Bench 16 - WordPress benchCRO on WooCommerce and WordPress Conversion work on the open platform: WooCommerce checkout control, WordPress landing pages, plugin discipline and the speed costs of a full plugin folder. 12 guides Bench 17 - Services benchServices and local businesses When the conversion is a call, a quote request or a booking, not a parcel. Lead quality over lead count, and pages built around how people actually choose a tradesperson or a clinic. 12 guides Bench 18 - Subscription benchSaaS and subscription conversion Trials, demos and pricing pages for small software products, where the conversion is the start of a relationship and activation matters more than the signup. 12 guides Bench 19 - Leak benchFunnel leaks by stage A stage-by-stage walk of where visits quietly drain away: arrival, browsing, product evaluation, the decision, the till, and the return visit that never happens. 12 guides Bench 20 - Audit benchAudits, checklists and mistakes The bench's own paperwork: how to run a conversion audit on a small store, the checklists worth keeping, and the mistakes and myths that waste the most effort. 12 guides

How this desk works

A test bench desk: a notebook with a hand-drawn funnel and line chart, a laptop and a phone

Three rules. First, no invented numbers: this site publishes no case studies and no benchmark figures of its own, because it has no clients and runs no stores. Where a number appears, it comes from published research or platform documentation, is attributed to its source, and carries the date it was checked; worked examples use round numbers and are labelled illustrations. Second, evidence is graded on every guide's test card, so you can see whether a suggestion rests on published research, platform documentation, practitioner consensus, or a method you run yourself. Third, corrections are published, not buried: the corrections page records what changed and why. The desk's position throughout: your store's baseline beats anyone else's benchmark, every time.