A £12,000-a-month paid search account can send 1,500 visitors to a B2B website, record 30 form fills and still produce only three sales-qualified leads. The dashboard reports a 2% conversion rate. The sales-qualified conversion rate is 0.2%.
A B2B website conversion audit exists to explain that gap. It is not a gallery critique, a generic checklist or permission to rebuild everything. It is a controlled assessment of whether qualified visitors can recognise fit, verify claims, take an appropriate next step and reach your CRM as measurable pipeline.
The score matters less than the evidence behind it. A 51/100 score supported by URLs, session data and CRM outcomes is useful. An 83/100 assembled from opinions is not.
What a B2B website conversion audit should measure
The unit of analysis is the path from a relevant entry point to a qualified commercial outcome. Not a page in isolation, and not the taste of the person conducting the audit.
Write these six facts at the top of the scoring sheet:
- Intended buyer and the other stakeholders who influence the decision
- Offer being evaluated
- Primary next action, such as a consultation, demo or technical assessment
- Exact definition of a qualified conversion
- Typical contract value and sales-cycle length
- Main acquisition channels and device split
If the team cannot agree on a qualified conversion, stop. Sixty form submissions and 18 sales-accepted leads are not 60 conversions. Reporting the former makes performance look 3.3 times better than the outcome sales actually values.
The audit should establish four things: whether the right visitor recognises relevance; whether the claims carry enough proof; whether the next step feels proportionate and works; and whether that action can be connected to pipeline.
It should not prescribe site architecture or a redesign before the evidence is scored. If the findings justify structural work, move to the separate principles for a redesign that improves conversion. Keeping diagnosis and prescription separate prevents a low score becoming a pre-written rebuild pitch.
Build the evidence pack before scoring
Start with 90 days of analytics and CRM data. If that period contains fewer than 50 qualified conversions, widen the window to six or 12 months and note changes in seasonality, offers or channel mix. These are operating defaults, not statistical thresholds.
Pull sessions, engaged sessions and conversion events by landing page, source or campaign, and device. Then join form, call and booking records to sales stages in the CRM. Exclude staff visits, test submissions, duplicate events and obvious spam. Traffic without lead quality is context, not proof of conversion performance.
As a default scope, audit the homepage, the five largest non-brand entry pages, paid landing pages, the primary contact or booking flow, and any path responsible for at least 5% of qualified leads. This is a sampling rule, not a content inventory.
Test the live site in a logged-out browser at desktop and 375–430px mobile widths. Complete each primary path using an approved test record. Check field labels, validation, error recovery, confirmation, calendar hand-off and CRM receipt. Record the URL, device, date and screenshot for every failure.
Use an evidence hierarchy:
- 1. Direct evidence: observed behaviour, analytics, CRM outcomes, call data and usability sessions
- 2. Supporting evidence: technical checks, recordings and platform documentation
- 3. Opinion: stakeholder statements and heuristic judgement
Opinion can create a hypothesis. Repeated observations can support a finding, but not prove its cause. Reserve causal claims for controlled tests, holdouts or directly observed technical failures.
The 100-point B2B website conversion audit scorecard
Rate every subcriterion from 0 to 5. Do not give an entire area one impressionistic rating.
Rating scale
- 0: absent, broken, misleading or impossible for a visitor to verify
- 1: present but creates serious uncertainty or friction
- 2: partial, generic or inconsistent
- 3: adequate for a typical qualified visitor
- 4: strong, specific and consistent
- 5: exceptionally clear, low-friction and consistent
Calculate each result as (rating ÷ 5) × criterion weight. A proof-quality rating of 2 against a seven-point weight contributes 2 ÷ 5 × 7 = 2.8 points.
Add the unrounded subcriterion results. Round area rows for display only, then round the final total.
| Area | Weight | Subcriteria and scoring test |
|---|---|---|
| Buyer relevance and proposition | 20 | Audience recognition (6): can the intended buyer identify their situation and the outcome? Commercial specificity (7): is what is delivered, for whom and why it matters concrete? Message consistency (7): do entry, evidence and action points describe the same offer? |
| Conversion path and momentum | 20 | Visible next step (6): is one primary action easy to find? Journey continuity (7): can a visitor move from claim to proof to action without dead ends? Form or booking friction (7): is the commitment proportionate, reliable and clear about what happens next? |
| Trust and risk reduction | 18 | Proof quality (7): are results, clients or credentials attributable? Claim substantiation (6): do material claims include a baseline, scope, timeframe or source? Commercial reassurance (5): are relevant delivery, security, procurement and contact concerns addressed? |
| Offer and qualification clarity | 15 | Fit boundaries (5): is the intended client clear? Engagement clarity (5): can a buyer understand likely scope, process, timing or commitment? Action alignment (5): has the site earned the size of the requested commitment? |
| Usability and accessibility | 12 | Responsive completion (4): does the key task work on mobile and desktop? Readability (3): is the hierarchy scannable? Accessible interaction (3): do labels, keyboard use, focus, contrast and errors support completion? Technical reliability (2): do pages, forms and confirmations work? |
| Measurement and follow-through | 10 | Instrumentation (4): are completed and meaningful partial actions tracked once? Source insight (3): can qualified outcomes be tied to channels and entries? Lead handling (3): are routing, ownership and response expectations operationally real? |
| Market and business legitimacy | 5 | Business legitimacy (2): are company, contact, privacy and cookie details easy to find? Market fit (3): do language, proof and operational promises suit the intended market, including UK terminology where relevant? |
Score clarity, not whether the auditor likes the tone or colour palette. If tone consistency becomes contentious, a workshop-based brand personality framework gives the team shared language; distinctiveness alone does not earn conversion points.
Apply these guardrails:
- If the primary action will not submit or confirm, score form reliability at 0 and cap conversion path and momentum at 10/20.
- If material outcome claims have no attributable evidence, cap trust and risk reduction at 9/18.
- If analytics or CRM evidence is unavailable, mark that criterion unverified. Do not assume a middle score. Withhold the overall total or report the observable subtotal and missing weight, such as 72/90 + 10 unverified.
- If mobile represents at least 20% of sessions, score Responsive completion using the lower of the desktop and mobile results.
- Treat the legitimacy category as a visible buyer-confidence check, not a legal compliance review.
Do not count the same evidence twice. A case study may substantiate a claim; it does not automatically prove the form is proportionate. Write a one-sentence rationale and capture evidence for every rating below 3 or above 4.
A worked conversion audit: 51 points, not a redesign brief
Northstar Cyber is a fictional UK cybersecurity consultancy with 3,200 monthly sessions, a £24,000 average first-year contract and 55% of traffic coming from paid media. Mobile represents 54% of relevant sessions. Its analytics records 40 monthly enquiries: a 1.25% raw conversion rate.
The CRM changes the picture. Twenty-two enquiries are genuine and eight become sales-qualified, so the qualified website conversion rate is 8 ÷ 3,200 = 0.25%.
The homepage promises to “secure your future” but does not identify the buyer or risk it handles. Client logos are named, yet case studies report no outcome, scope or timeframe. A six-field form works on desktop; on mobile, a large video pushes the main action down and the validation message does not identify the failed field. Form submissions are tracked, but no CRM stage returns to analytics.
| Area | Score |
|---|---|
| Buyer relevance and proposition | 11/20 |
| Conversion path and momentum | 11/20 |
| Trust and risk reduction | 8/18 |
| Offer and qualification clarity | 8/15 |
| Usability and accessibility | 7/12 |
| Measurement and follow-through | 2/10 |
| Market and business legitimacy | 4/5 |
| Total | 51/100 |
The site looks legitimate, but a security lead cannot quickly establish fit, verify the claims or see which paid visits become pipeline. That is a measurement and decision-friction problem, not proof that every page needs replacing.
The commercial case can still be modelled. If tested fixes add two qualified leads in a month, the close rate remains 25%, first-year revenue is £24,000 and gross margin is 60%, expected first-year gross contribution from that month’s added leads is 2 × 25% × £24,000 × 60% = £7,200.
If the uplift repeats monthly, deals close at the same rate and sales-cycle and cash-collection timing are excluded, a £9,000 implementation has a modelled run-rate payback of 1.25 months. That is not a forecast. It exposes the assumptions that must be validated.
Turn the audit score into an action queue
Use the total as triage, not a verdict:
- 80–100: strong observable foundation; run controlled optimisation tests
- 60–79: credible but materially leaky; address high-weight gaps
- 40–59: trust, qualification or journey friction is likely suppressing enquiries
- Below 40: repair basic clarity, access and measurement before buying more traffic
Rank findings with point deficit × exposure × evidence confidence ÷ effort. Score exposure as 3 for paths handling over 50% of relevant sessions, 2 for 20–50% and 1 below 20%. Use confidence of 1.0 for direct evidence, 0.7 for repeated supporting evidence and 0.4 for opinion.
Score effort as 1 for up to one week, 2 for more than one week to four weeks and 3 for more than four weeks.
For Northstar, the eight-point measurement deficit scores 8 × 3 × 1 ÷ 1 = 24. The five-point usability deficit scores 5 × 3 × 1 ÷ 1 = 15. The ten-point trust deficit scores 10 × 3 × 0.7 ÷ 2 = 10.5. Instrumentation goes first because it improves every decision that follows; mobile form repair follows because the failure is directly observed.
Broken enquiries, accessibility blockers that prevent the primary task, and confirmed privacy or security failures bypass the ranking formula. They are immediate fixes.
The audit defines the problem, not the supplier. Some gaps need analytics or CRM work; others need copy, design or development. Broader website design and redesign expertise may be relevant when the conversion path is structurally weak. Examples of website project delivery can help a team calibrate whether a contained build is enough.
If external support is required, use an evidence-led question list for comparing redesign agencies before reviewing proposals. For each approved action, record the evidence, affected path, owner, expected metric, guardrail metric and review date.
Test one material hypothesis at a time where traffic allows. Before release, use the common B2B redesign failure patterns as implementation QA rather than expanding this audit into another strategy exercise.
After 250 relevant sessions or 30 form starts, recheck technical and usability criteria. Treat that as a review trigger, not proof of commercial improvement. Update commercial scores only when qualified-outcome data has matured, then run the full audit quarterly or after a material launch or traffic change. Keep the source mix and scoring rules consistent, otherwise the comparison is theatre.
FAQ
How long does a B2B website conversion audit take?
As a planning estimate, a focused audit of 10–15 conversion-critical paths takes two to four working days once analytics and CRM access is available. Multi-market sites or broken data joins take longer.
What is a good B2B website conversion audit score?
Eighty or more indicates a strong observable foundation, not guaranteed commercial performance. Read the total alongside qualified conversion rate, pipeline value and cost per qualified opportunity.
Can the audit work without much conversion data?
Yes, but the result is provisional. Widen the period, use observed task completion and user sessions, and label measurement gaps. Do not turn low volume into fabricated certainty.
Does a low score mean the website needs a redesign?
No. It means high-weight friction deserves investigation. A broken mobile form, vague proof and missing CRM feedback can often be corrected without rebuilding the whole site.
Summary
- Define conversion as a qualified commercial outcome, not any form fill
- Build the evidence pack before scoring opinions
- Rate subcriteria from 0–5 and calculate a weighted score out of 100
- Prioritise point deficits by exposure, confidence and effort
- Re-score with the same rules after enough relevant evidence has accumulated
Actualyse designs and rebuilds B2B websites that turn research visits into qualified pipeline. Book a call to talk through where yours stands.
