Creative Director at Timmermann Group, leading discovery, design and the theme and plugin code on a research-led rebuild. Signature Medical Group is a physician-owned, multi-specialty independent medical group. More than 110 doctors work across some 30 locations in St. Louis and Kansas City. I led a cross-disciplinary team of seven: an account strategist, two content strategists, two art directors and two WordPress developers. We began in October 2014 with an analytics baseline and a competitive benchmark. I prototyped in Sketch from day one. Work moved through staging after the discovery report. Nothing reached production until February 2016.
The existing site was a DevExpress-era ASP.NET WebForms build with no intuitive pathways to content. But the real problem was only visible once the data was in. A blind single-answer survey on the live site found the most common reason for a visit was to find a physician, at 25.3%. Access to health records was close behind, at 24.1%. The analytics baseline, run against October 2014 traffic, found provider profiles had the highest bounce rate on the site, pointing at the doctor locator itself. The most-wanted task was the worst-performing one. Worse, most arrivals came through organic search and averaged three pages before leaving. Users relied on third-party sites to find content that should have been reachable directly.
Research ran first, and the deliverable documented the full method. I recruited participants by Lead User Sampling, the technique von Hippel described in Management Science in 1986. Six active healthcare patients each sat for a one-hour interview. Three knew SMG and three did not. More than eighty people answered a survey. Five SMG business leaders sat for contextual interviews with a participatory design session. Six cognitive walkthroughs covered four direct competitors, Esse Health, BJC, Mercy and SSM, plus two indirect ones in Anthem and ZocDoc. I synthesised the interviews into a five-phase patient journey: Research, Scheduling, Pre-Visit, Visit and Post-Visit. Each phase is marked linear or non-linear, because a shopping phase that loops needs different handling from a visit that runs once. I scored stakeholder requirements MoSCoW, with a confidence percentage on each. The scores exposed a disagreement they did not settle. Online appointment scheduling sat in SHOULD at 75% confidence and in WON'T at 25%.
Discovery ran from October 2014 and put nothing into production for sixteen months. Sketch prototyping started on day one and iterations ran in staging once the discovery report was complete. Only the release waited. Three instruments did the work. A survey went to visitors on the live site. Six one-hour patient interviews fed a journey model. Five contextual interviews with SMG leadership were scored for agreement.
Research
Observation only
“Finding doctors recommended to me who accept patients & my insurance”
Scheduling
Active
“Scheduling an appointment with a healthcare professional”
Pre-Visit
Observation only
“Waiting for and completing intake paperwork”
Visit
Active
“Consulting with the doctor about my health treatment needs”
Post-Visit
Observation only
“Scheduling follow-ups. Paying bills. Updating my loved-ones”
MUST
SHOULD
COULD
WON'T
Participants came from Lead User Sampling, the technique von Hippel described in Management Science in 1986. Six patients sat for interviews, three already familiar with the group and three not. More than eighty people answered a blind single-answer survey. Five stakeholders gave contextual interviews with a participatory exercise inside them. Six cognitive walkthroughs covered four direct competitors and two indirect.
Stated in the deliverable at the time: the sample is not representative of the general population, and this is not a statistically valid quantitative study. It is a contextual exploration meant to surface issues that lead users are especially articulate about.
The planned architecture, drawn before the build. Annotation 1 states the doctor locator is a homepage feature: the earliest retained statement of the input the whole study rests on, made in March 2015, ahead of any code.
Find a Doctor, said twice
Find a Doctor leads the navigation and repeats as the first thing in the page body
The whole clinic on one page
One page holds the entire practice: contact, services, hours, every physician and every location the clinic operates from
A doctor under their own roof
A physician sits under the masthead of their employing practice rather than the group's. Everything the entity markup declares is a field on this page: specialty, procedures, hospital affiliation, address
What the phone drops
At 390px the background video and the desktop navigation are both display:none
The request knows which doctor it is for
Scoped to the doctor whose page it opens from. It captures a requested date and time and how the patient heard about that named physician
Signature was not one brand. It was a group that endorsed practices, and practices that housed independent physicians who had bought into the group. The engagement was a strategic investment in design, marketing and brand, owed to each of them. So the brand architecture had to be expressed twice: once for readers, and once for machines.
Type
Three faces, three jobs
Signature Medical Group is an independent, multi-specialty physician group serving greater St. Louis, greater Kansas City and Bolivar, Missouri.
AaAaAaAa
Find a Doctor · Why Signature? · Community Outreach · Programs
Montserrat · headlines and every control
Rich directory listings
Medical group
An independent, multi-specialty physician group serving greater St. Louis, greater Kansas City and Bolivar, Missouri.
Orthopedic clinic
A Division of Signature Medical Group. EMG and nerve conduction studies, MRI imaging, orthopaedic surgery, pediatric orthopaedic surgery, sports medicine.
Orthopedic surgeon at Kansas City Bone & Joint Clinic
A division of Signature Medical Group. His memberOf names the practice that employs him; the group's endorsement reaches him through the clinic.
Organic results
signaturemedicalgroup.com
Signature Medical Group | Physicians Making A Difference
Signature Medical Group has the best doctors St Louis and Kansas City residents could ask for. Use our website to find the top doctors in St Louis and Kansas …
signaturemedicalgroup.com › practice › mid-county-orthopedic
Mid County Orthopaedic Surgery & Sports Medicine | Signature Medical Group
Mid County Orthopaedic Surgery & Sports Medicine is a specialty practice provider of orthopedic surgery and sports medicine services in the St. Louis, MO …
signaturemedicalgroup.com › doctor › robert-p-bruce
Robert P. Bruce | Signature Medical Group
Robert P. Bruce is part of the Signature Medical Group network of top physicians in the Kansas City, MO area practicing orthopedic surgery.
Shared on Facebook
I wrote the theme in PHP on Roots Sage, and the plugins that gave the group its content model and its editing experience. The team moved the content onto it, on a migration plan we wrote together.
Search existed because finding a physician was the most common reason for a visit, and the hardest thing to build. Saving and sharing existed because the content strategy treated the site as the group’s record of itself.
Roots Sage compiles assets/ to dist/ with a content hash in every filename, so browsers can cache the bundle forever and still pick up a change the moment it ships. It is also why the analytics taxonomy is recoverable: thirty event calls survive in one archived file, main-7541c5ed.js
A page per physician: specialty, procedures, hospital affiliation and the practice that employs them
A page per clinic: contact, services, hours, its physicians and every location it operates from, with a paginated archive at /practice/ that the post type provides
Browse the 2017 archive →The browse layer above the entities. The only tier that never carried typed markup
Nutrition Edge, Strong Start and the trainer network, each with its own page
Typeahead over every physician and practice name, plus facets for specialty, zip, insurance and language. The hardest thing on the engagement, and the one the survey said mattered most. Still working in the archive
Lets a patient save a physician, practice or post to return to. Source of the Bookmark Tracking events
Sharing a highlighted passage as well as a whole page. Source of the two Share Tracking labels
Stack
Read from the 8 October 2016 capture: script filenames, handles and plugin slugs. Select2 is here because it is what made the search facets usable.
The build followed and launched in February 2016, by which point I had left Timmermann Group. I did the discovery, the design and the theme and plugin code, and planned the content migration with the team who carried it out and shipped it. Working that way is what taught me that development belongs alongside design and content strategy, not after them. Everything below is scored against what went live. It was a custom WordPress theme replacing the .NET site. It ran on the design system I authored and reused across clients: a 10px root, a 32px baseline, spacing on a 4px grid at 86% adherence, and named colour utilities. The discovery report itself is the more durable artifact. It states its sampling frame, cites its methodology, and gives confidence intervals for every survey figure. Its limitations section says plainly that the sample is not representative. It calls the study a contextual exploration with no claim to statistical validity. That was written into a client deliverable in 2014, eleven years before this portfolio was built to argue the same thing.
The research figures (the 25.3% and 24.1% survey splits, the October 2014 analytics baseline) come from the retained discovery report, which states its own sampling frame and calls itself a contextual exploration with no claim to statistical validity.
The durability claim is read from the Internet Archive and the live site, not from client analytics: no post-launch measurement was instrumented against the new structure, and I had left Timmermann Group before the February 2016 launch. What is shown is that the architecture survived a 2021 redesign and replatform, not what it moved.
Each design input is a specification, not a prediction. The design outputs are what shipped. The verification method is how the archive confirms it. Outcome is a different question: what happened afterwards? For most rows, nobody measured. Where a decision survived the 2021 replatform, run by another team, that survival is the only independent evidence in this record.
| ID | Hypothesis | Design Output | Acceptance Criteria | Verification Method | Outcome |
|---|---|---|---|---|---|
| H-001 | Finding a physician must lead the site rather than sit inside a department structure Evidenced | Find a Doctor is the first primary-navigation item and a search control is present above the fold on the homepage | Wayback capture inspected at 1280px desktop and 390px mobile | Retained 2021 | |
| H-002 | Access to health records must be reachable without using search Evidenced | A records control is present in a persistent patient-resources group on every template | Wayback capture inspected across homepage and physician templates | Retained 2021 | |
| H-003 | Navigation must be shaped by audience rather than by the organisation chart Evidenced | Primary navigation carries no more than five items and none of them names an internal department | Wayback capture of the primary navigation before and after | Retained 2021 | |
| H-004 | Every practice must state its membership of the group on its own page Evidenced | The group endorsement is present in the practice masthead and in machine-readable markup on the same screen | Archived markup inspected for the endorsement line and for memberOf | Lost 2021 | |
| H-005 | Physicians and practices must be resolvable as entities by a search engine rather than as prose Evidenced | Each physician and practice page carries typed markup naming the entity and its relationships | Archived markup inspected for schema.org types on both templates | Lost 2021 | |
| H-006 | The physician recruitment argument must be a section of the site rather than a careers listing Evidenced | Recruitment content is reachable from primary navigation and is not filed under Careers | Wayback capture of the primary navigation and the utility bar | Demoted 2021 | |
| H-007 | Value propositions must be demonstrated rather than asserted Mixed | Accessible, affordable and independent care are stated in body copy, and each is carried by D-013 (named patients telling their own outcome story), D-007 (physician credentials published as typed data rather than asserted) and D-009 (figures the group could source) | Archived body copy checked for the three stated claims, and the page checked for the surfaces carrying the fourth | Lost 2021 | |
| H-008 | Community and programme content must be reachable without displacing the primary task Evidenced | Programmes and outreach occupy their own primary-navigation groups and not the homepage above the fold | Wayback capture of the navigation menus | Lost 2021 | |
| H-009 | A patient on a phone must not be made to download the homepage background video Assumed | The background video is suppressed below the mobile breakpoint | Archived page rendered at an iPhone viewport and at desktop and the computed styles compared | Not applicable | |
| H-010 | Appointment demand must be captured at the physician level without committing the group to live scheduling Evidenced | A request control is present on every template, and on a physician page the request form is scoped to that physician and captures a requested date and time | Wayback capture of the utility bar and of the request form on a physician page | Retained 2021 | |
| H-011 | The rebuild must raise provider recruitment and retention Evidenced | A measured change in recruitment or retention attributable to the site | Recruitment and retention figures are held by the client rather than observable from the site | Not measured | |
| H-012 | The site must report what patients were looking for, not only where they stopped Evidenced | Every search facet selection and every off-site conversion is recorded as a named event with the chosen value as its label | Theme JavaScript read from the archive: 30 ga('send','event') calls in main-7541c5ed.js, plus an inline handler on practice phone links | Lost 2021 |
content model
information architecture
structured data
structured data
brand expression
delivery
Endurance is a measure most design work never gets held to. Awards are given at launch, and revenue is credited to the campaign next to it, but neither asks whether the thing still works once the people who built it have gone. This split is that question answered on one project. Six decisions outlived the team, the platform and the CMS. Five did not. Find a Doctor still leads the navigation, ten years on, because it was the most-wanted task in 2014 and nobody since has found a reason to move it.
Robert Duebelbeis
Creative Director
Discovery research, design, and the theme and plugin code. Planned the content migration with the team.
Content migration onto the new theme.
Content migration onto the new theme.
Andrew Mullens
Content migration onto the new theme.
Melissa Gall
Content migration onto the new theme.