The Redesign Discovery Process: Questions Before Wireframes

A redesign discovery phase that records no baseline and no target has produced zero useful success criteria.

A number by itself can still be decorative: it can reward cheaper traffic, a looser lead definition or a convenient reporting window.

The B2B website redesign discovery process is where success criteria become falsifiable or stay decorative, and the tell is whether anyone writes down a number they could later be wrong about.

The B2B website redesign discovery process starts with evidence

The brief usually arrives as a verdict: the website feels dated, the proposition looks unclear or sales dislikes the leads. Those observations can start an investigation. They cannot define success.

Build the baseline from five named sources where available: GA4, Google Ads, Search Console, the CRM and call tracking. Definitions matter more than dashboard polish. An illustrative evidence ledger might show 31 submitted forms versus 10 CRM-accepted enquiries. Both figures can be accurate while answering different questions.

Settle the discrepancy before setting a target. If two systems differ by more than 10% on the same defined outcome, stop target-setting, choose the system of record and document the reconciliation rule.

The window also needs a decision. Compare a 30-day snapshot with a 90-day baseline; if seasonality or long sales cycles make them materially different, use the longer defensible window. Do not let the post-launch team select whichever period flatters the result.

The wider questions of persuasion and proof belong within conversion-led B2B redesign principles. Discovery has the narrower job of deciding which outcomes, pages and definitions those principles must serve.

The honest limit here is that post-launch movement is confounded by four things: sales follow-up, offer changes, seasonality and media mix. Discovery cannot prove that design caused every change. It can record the controls, exceptions and competing explanations.

Evidence turns redesign ambition into a testable commercial brief.

Seven artefacts the B2B website redesign discovery process must produce

A workshop deck is not an artefact merely because somebody exported it as a PDF. Seven outputs must survive the meetings, and each needs a completion test.

1. Baseline and evidence ledger. Record five fields for every metric: name, value, source, window and owner. It is finished when a second person can reproduce the figure without asking its author what was excluded.
2. Audience and decision map. Give each priority audience four decisions: immediate job, buying objection, required proof and next action. It is finished when the sponsor has ranked the audiences and explicitly named those the site will not prioritise. Lanteria’s broad Microsoft 365 HR capability set required routes for multiple stakeholder audiences; Lake Erie Shores needed stay and ownership audiences held coherently under one site.
3. Current-to-future page inventory. Assign every URL one of five statuses: keep, merge, rewrite, retire or create. It is finished when every current URL has a status, owner and destination. An illustrative inventory reading 84 current URLs versus 46 proposed destinations communicates a decision; “simplify the website” does not.
4. Page-intent briefs. Lock five fields for every priority page: one primary query, one buyer job, one working title, one primary action and one proof obligation. It is finished when none of those fields contains “and” as a way of avoiding a choice.
5. Content and proof gap register. For every material claim, record four fields: required evidence, current asset, owner and create-or-remove decision. It is finished when unsupported claims have either acquired a credible source or been deleted from the proposed message.
6. Measurement and acceptance specification. Give every success measure four fields: event name, system of record, qualification rule and evaluation window. It is finished when marketing, sales and the implementation team can describe the same accepted outcome. Testing alternative experiences later belongs to structured B2B conversion optimisation, not discovery.
7. Decision log. Record five fields for every consequential choice: decision, owner, evidence, date and reopening condition. It is finished when no unresolved item can change the sitemap, page purpose or success target without a named decision-maker.

AfriCap Hub’s catalogue, filtering and registration journey demanded a different architecture from Savgen’s technical, multi-industry offer across brand and site. The useful artefact captures why those structures differ; it does not impose a favourite sitemap on both.

Finished artefacts leave no room for interpretive hand-offs.

Four decisions to lock before any wireframes

Layout is an expensive place to discover that stakeholders disagree about the page’s purpose. Four gates should close before UX and UI design starts arranging content.

1. Lock the page inventory. Approve the current-to-future map, including merges and removals. If more than 10% of priority URLs still lack an agreed status, pause wireframes and reconcile the inventory.
2. Assign one query and one job per page. One query means one primary search intent, not one literal keyword. If a page serves more than one primary query or more than one buyer job requiring different proof and actions, split the page or nominate one job as secondary.
3. Lock the titles. Approve three title surfaces: navigation label, on-page heading and SEO title. They may differ, but they must describe the same page. If three approvers submit three materially different page meanings, return the brief to the audience decision instead of asking a designer to blend them.
4. Lock the primary actions. Name three action fields: call to action, destination and tracked event. If fewer than 80% of priority pages have all three, stop layout work and finish the action map.

These decisions do not prohibit refinement. They make changes visible. A later change to page purpose becomes a recorded scope decision rather than silent rework disguised as design feedback.

Page purpose must be settled before page shape.

If this analysis exposes wider gaps in your site, we can turn the evidence into a focused redesign brief — book a call

What discovery is not: three things that do not work

Polished workshops can still produce no usable constraint. Three common substitutes repeatedly fail to move the outcome.

1. A stakeholder interview round that produces sentiment and no decisions. Statements such as “the brand should feel more premium” or “customers need clarity” create no priority, owner or acceptance test. Interviews become useful only when conflicting observations end in a recorded choice.
2. Moodboards before architecture. Early colour, typography and imagery discussions anchor stakeholders to an aesthetic before the team knows which audiences, evidence and actions the interface must support. When visual character is genuinely unresolved, a practical brand personality framework should produce bounded traits and exclusions, not a collage of admired brands.
3. Competitor screenshot galleries presented as strategy. Screenshots reveal visible conventions, not the competitor’s buyer mix, content constraints or commercial reasoning. Copying their navigation can reproduce a solution to a different problem while making the result appear reassuringly familiar.

More interviews, more sticky notes and more reference sites do not compensate for absent decisions. Activity is not progress when nobody is accountable for choosing.

Sentiment without a recorded decision is not discovery.

Five demands before signing and one worked success statement

Commercial vagueness becomes expensive after the proposal is accepted. When appointing a website design and redesign partner, hand the agency five pre-signature demands.

1. Recent case studies showing comparable decision complexity, with an explanation of what was decided and why.
2. The process in writing, including decision rights, conflict resolution and approval ownership.
3. Exact deliverables by week or month, stated as inspectable outputs rather than workshop names.
4. The reporting cadence, including who receives decisions, risks and unresolved dependencies.
5. Response times for routine and blocking questions; require sponsor escalation when a blocker remains unanswered for two working days.

Our Lanteria, AfriCap Hub, Savgen and Lake Erie Shores work demonstrates different architecture problems, not a universal template. Recent case studies should expose that reasoning. Attractive screens alone cannot show whether discovery was rigorous.

The success statement must then connect the evidence to a test. Full investment and payback modelling belongs in a website redesign ROI model; the discovery document only needs a controlled, falsifiable outcome.

Take a transparently illustrative Manchester cyber-security consultancy with five labelled inputs:

• Input A — fixed Google Ads spend: £8,000 per month.
• Input B — recorded baseline: 10 CRM-accepted enquiries per month from Google Ads landing-page sessions.
• Input C — target: 14 accepted enquiries per month.
• Input D — comparison windows: 90 days before launch versus days 31–120 after launch.
• Input E — qualification rule: a target-sector business with a named need, accepted by sales, excluding duplicates, tests and spam.

The arithmetic is reproducible:

Baseline enquiry cost = £8,000 ÷ 10 = £800

Target enquiry cost = £8,000 ÷ 14 = £571.43

Baseline 90-day total = 10 × 3 = 30 accepted enquiries

Target 90-day total = 14 × 3 = 42 accepted enquiries

The complete success statement is:

For this illustrative Manchester cyber-security consultancy, redesign success means increasing CRM-accepted enquiries attributed to Google Ads landing-page sessions from the recorded baseline of 30 during the 90 days before launch to at least 42 across days 31–120 after launch, while Google Ads spend remains £8,000 per month and the CRM qualification rule remains unchanged.

Our falsifiable claim is that the redesign will reach 42 accepted enquiries under those controls, and 41 or fewer proves the claim wrong.

A signed target gives every design decision a burden of proof.

FAQ

How many people should hold discovery decision rights?

Use one accountable sponsor and no more than four contributing leads. If six or more people can veto a decision, name one tie-breaker before the first workshop and move observers to written review.

How much historical data is enough?

If the previous 90 days contain fewer than 30 qualified outcomes, extend the baseline to 12 complete months. If the offer or qualification rule changed during that period, use the post-change window and record why comparability was sacrificed.

Should full copy be written before wireframes?

Write complete copy for the three pages with the highest technical, regulatory or proof burden before finalising the component system. If any page needs more than two distinct evidence formats, let the content structure drive the layout.

Must the CMS be chosen during discovery?

Only choose a replacement when at least two documented needs cannot be met by the current system. Compare the current and proposed platforms against five criteria: editorial governance, integrations, performance, security and ownership cost.

Clear ownership makes remote discovery no less rigorous.

Summary

• Reject any discovery proposal that does not name its seven finished artefacts.
• Reconcile the system of record when matching outcomes differ by more than 10%.
• Pause wireframes when over 10% of priority URLs lack an agreed disposition.
• Split any page carrying more than one primary query or buyer job.
• Record a baseline, target, window and disproof condition before approving layout work.

Wireframes should begin only after evidence, ownership and tests align.

Actualyse designs and rebuilds B2B websites that turn research visits into qualified pipeline. Book a call to talk through where yours stands.