A six month discovery that turned "we need a new website" into a 41 question RFP, five personas, fifteen user journeys, a CMS recommendation and a shortlist of agencies ready to be briefed.
A downstream build that would otherwise have been priced blind, scoped late and changed three times after kick off.
The starting point
A site that everyone agreed wasn’t working. Nobody could agree on why.
The firm had grown into a global practice over the previous decade. The UK and other countries shared a brand, shared a leadership conversation and shared a CMS site that had grown organically since it was built. Over 15,000 pages indexed. A PageSpeed score of 26 on the home page. A Total Blocking Time of 1,240ms when most competitors sat between 140ms and 360ms. A recurring redirect issue the existing agency hadn’t been able to fix.
The marketing lead knew the site was overdue a rebuild. So did the partners. But every conversation about it stalled in the same place. The agency would propose a refresh. The IT team would propose a platform migration. The partners would propose adding their practice area to the home page. Nobody had asked the question that came first.
Who is this site for, and what do they actually do on it?
There was a set of historic personas. There was a brand tracker report. There was a quarterly agency report. There were three different definitions of what counted as a conversion. None of them lined up.
Above all there was no document the marketing lead could put in front of an agency and say “build to this.” Without that document, every quote that came back would be a guess. Every quote that was a guess would create change orders. Every change order would erode the budget.
The job, before anything else got built, was to write the document that wasn’t there.
The brief I was given
Eight objectives. Six months. One person leading the work, with weekly input from the in-house team.
1. Determine the scope and requirements for a new website
2. Run a proper competitor and best practice review
3. Develop a full set of personas with user journeys
4. Build the content matrix from those personas
5. Develop a simplified site structure the US arm could follow
6. Scope the CMS
7. Pull together the work brief, ready to send to agencies
8. Shortlist the agencies who’d receive it
The work needed to finish by the end of May so the new site could be live by the end of October. That was the only hard date. Everything else was a function of getting the right people in the room and asking the right questions in the right order.
The approach
1. Audit, then agree
Three weeks of audit work across the existing UK and US sites. Page count, structure, performance, readability, content health, conversion paths, analytics setup. The current Drupal site had drifted in every direction. The home page scored 26 on PageSpeed. The readability score was 36% Hard pages and 35% Very Hard, the worst spread of any competitor I benchmarked. Internal politics had quietly become the site’s information architecture.
The audit landed as a single working document the marketing lead could take into a partner meeting. Not a deck of opinions. A list of facts the agency couldn’t argue with because they came from the same Google Analytics property and the same Screaming Frog crawl.
Then the conversation that mattered more than the audit itself. With the marketing lead, the BD team and partners from both the commercial and private wealth sides in the room, we agreed what the site was being asked to do.
The existing objectives were brand awareness, inbound enquiries and SEO rankings. None of those were measurable in a way anyone could defend. We replaced them with four that were.
Quality lead generation, measured by conversion rate, named recipient in the BD team and 48 hour response time. Engagement, measured by return users, time on site and pages viewed per session. Automation, measured by the form to CRM to email journey handshake actually working end to end. Brand awareness, measured against the brand tracker the firm already paid for and never used in marketing reviews.
The new objectives had KPIs. The old ones had vibes. That single shift changed what the rest of the discovery had to deliver.
2. Competitor audit that named the gaps
Thirteen competitor sites benchmarked against the same five criteria. Page count. Domain authority. SPAM score. Core Web Vitals. Readability across the Flesch reading ease scale. Then a usability review of each one against a single task: how many clicks does it take a user to get from the home page to a sector landing page or a lawyer biography.
The pattern was clear. Almost every site was built in Sitecore or Optimizely. Almost every site followed the same blocky template. Almost every site led with the insights section even though insights drove a fraction of meaningful traffic. The firm had been told for years that its content was its differentiator. The audit showed that the differentiator everyone was chasing was the differentiator everyone already had.
The opportunity wasn’t to copy the competition. The opportunity was to set a PageSpeed score target above 60, get readability into the fairly easy range, and present services and sectors in a way that beat a three click menu. That moved the design brief from “let’s see what you can do” to “let’s see how you’d solve these four specific problems.”
3. Personas with teeth
Five personas. None of them invented. Each one built from existing client data, BD team input and the firm’s own brand tracker.
The high net worth individual was Mark. Sixties, business and land owner in the North. £10m+ in assets, legacy money, multiple UK properties, trustee for a charity. Risk averse, heritage driven, needs a partner relationship and value for money that he can perceive himself. Comes to the site via referral or a Google search for a named lawyer, lands on a profile page and looks for sectors, case studies and testimonials before he picks up the phone.
The in-house general counsel was Michelle. Fifties, GC for a top retailer, commutes into London. Time poor, meticulous, needs cost transparency and validation that the firm has done the work before. Comes via a directory search or a peer recommendation, lands on a profile page and looks for credentials, sector experience and the ability to book a meeting.
Three more personas filled out the picture, each one written with a name, an age, a location, a job, a set of priorities and a set of statements they’d make before and after using the firm. Three user journeys per persona. Fifteen journeys in total, each one pointing at the same handful of pages: the lawyer profile, the sector landing page, the case study, the contact form.
That gave the site structure its centre of gravity. Lawyer profiles and sector pages were doing the heavy lifting. Insights and thought leadership were not. The content matrix that followed treated those four page types as the priority and everything else as supporting.
4. What users actually do on a law firm website
This was the part that should have been obvious but wasn’t, because nobody had asked the question directly until the discovery did.
When the BD team and senior partners were asked what users came to the site for, in priority order, the answer was clear and consistent across the commercial and private wealth sides.
See what services the firm offers. See what sectors it works in. Read lawyer biographies. Get contact details.
Reading articles, downloading thought leadership, listening to podcasts and watching videos all sat well below those four. In some cases they barely registered. The insights section, which the firm had spent years investing in, accounted for a small fraction of meaningful interactions. That finding alone changed the priority list for the new site.
It also gave the marketing lead something to take back to partners who’d been pushing for their byline on the home page. The answer to “where’s my article?” became “on your biography page, where the readers are.”
5. CMS scoped against actual needs
Four CMS platforms reviewed against the firm’s real requirements: WordPress, Statamic, Umbraco and Drupal. Each one assessed against programming language, ease of use, design flexibility, content management, SEO features, scalability, security, multilingual support and cost.
The recommendation wasn’t a particular CMS. The recommendation was that the choice mattered less than the agency picking it. A good agency could deliver a great site. A bad agency could deliver a poor site on the most enterprise grade platform in the market. The CMS decision belonged with the build partner, not ahead of them.
That framing changed how the RFP was written. Instead of mandating a platform and ruling out agencies who didn’t work in it, the RFP asked each agency to recommend a CMS and defend the choice. The agencies who had a defensible answer revealed themselves quickly. The ones who didn’t, did too.
6. The RFP
Forty one questions across company information, account management, communications, technical capability, SEO, accessibility, security, headless options, costs, IP and timeline. A 16 page document with an appendix table the agencies filled in line by line. Hard word limits on every answer. A confidentiality clause that meant agencies who withdrew had to delete their copy in writing.
The RFP did three things the marketing lead hadn’t had before.
- It set the same questions for every agency, so the responses could be compared like for like.
- It made the questions specific enough that bluffing was difficult, with word counts that punished waffle.
- And it carried an appendix of the firm’s own audit findings, personas and user journeys, so every agency was responding to the same brief and not inventing their own.
Six months into the discovery, the marketing lead had a 16 page document she could send to a shortlist on the morning of 17 April and trust the responses she’d get back on 13 May.
What this means in practice
The build budget for a project like this typically lands somewhere between £200k and £600k.
The discovery work paid for itself in three ways before the build even started. First, by making the agency responses comparable. Second, by removing the change order risk that comes from an underspecified brief. Third, by letting the firm walk away from any agency whose answer to the persona section came back as “we’ll do that in discovery,” because the discovery had already been done.
The agencies who pitched against the RFP did so on a level playing field. The shortlist was drawn from a longlist that had already been filtered against the firm’s actual needs, not a generic brief. The presentations ran on a question set the firm had defined, not a sales pitch the agencies had prepared.
The new website was on track to go live by the end of October.
What was delivered
- Scope and requirements document |Full audit of UK and US sites, with measurable objectives
- Competitor benchmark |13 sites scored on performance, readability and usability
- Personas | 5 personas, named, with priorities, statements and channels
- User journeys | 15 journeys, three per persona, mapped to page templates
- Content matrix | All ~17,000 current pages audited and tagged for reuse
- Site map | New structure designed around services, sectors, bios, contact
- CMS recommendation | Four platforms evaluated, decision framework provided
- RFP document | 16 pages, 41 questions, appendix table, confidentiality clause
- Agency shortlist | Longlist filtered to a shortlist with named reasons for each
What this means if you’re reading this
If you’re a marketing lead at a firm being asked to commission a new website, the question you’ll be measured on isn’t whether the new site is nicer than the old one. It’s whether the work it took to get there is defensible when a partner asks why it cost what it cost.
The way to make it defensible is to do the discovery before you brief the agencies, not during the build with them. Five personas built from real client data, not invented at a workshop. User journeys mapped to actual pages, not aspirational ones. A competitor audit that names the gaps in specific terms. A CMS scoped against your real needs. An RFP with hard questions and hard word limits.
A discovery like this doesn’t replace the build. It makes the build possible to price properly and to compare across agencies on the same terms.
If you do this work first, you’ll spend less on the build, the build will land on time and the partner conversation about budget will be a different conversation than it would have been.
About this engagement
Six months. Weekly review calls with the in-house marketing lead. Independent of the firm’s incumbent agency. Followed by an agency selection engagement which is the subject of the next case study.
For a confidential conversation about a similar engagement at your firm, get in touch.