Client case study · Website for a geotechnical firm

Kaveree needed a website that turns a referral into a call.

Kaveree is a principal-led geotechnical engineering firm in Chantilly, Virginia. It competes with national firms for design-build, public and private work. We rebuilt its site around what its buyers fear, showed the engineer they would work with, and tested the result against 15 competing home pages.

The Kaveree site, recorded from the build. Real recordings of the preview site in Edge. The field footage in the site's hero is a dramatized brand film.
Client
Kaveree, Inc., geotechnical engineering, Chantilly, Virginia
Our part
Brand (from June 2026), then the website's strategy, design, build, 3D and films
Status
Reviewed and approved by Kaveree; the site goes live on kaveree.com after launch
Tools
Astro, Blender, HyperFrames, Jev
The old kaveree.com home page: an aerial highway photo, 'Founded on Why? And How?', a Contact Now button and a cookie banner
Before. kaveree.com today. Benchmark score 38 in 100, last of 16.
The new Kaveree home page: 'Know the ground before you price it' over a field film, the principal engineer and the next step
After. The new home page, first screen at the same size. Benchmark score 82 in 100, first of 16.
Brief

The problem and the brief

Most of Kaveree's work starts with a referral. When the referred buyer looks the firm up, the website decides whether that turns into a call.

Kaveree is a small firm. It competes with national and regional engineering companies for design-build, public-sector and private work. We had already rebuilt its brand: the logo, identity, merchandise and a statement of qualifications. Its first website draft was a Framer template that still carried the template's sample testimonials, team and contact details.

The brief was to confirm a referral fast and remove the reasons not to call. Kaveree's leaders asked for a site as calm and technical as the sites of large engineering firms, with no hype.

How buyers choose

The research

62%

of AEC buyers ask a colleague first

80%+

then check the firm's website. References are the check they use least.

Source: Hinge Marketing, How Buyers Buy: AEC. The same study found sellers overestimate price; buyers pick on reputation, senior people and fit.

What buyers write

We gathered forum threads and review pages where engineers, owners and contractors talk about hiring a geotechnical firm: Eng-Tips, Reddit's engineering communities and Google reviews. Jev read 722 passages and judged each one. Was it about hiring a geotechnical consultant or about ground risk? Which theme did it express, and who was speaking?

  1. 722passages read
  2. 299about hiring or ground risk
  3. 159from the buying side
Themes in the 159 buying-side passages
  • Good investigation pays for itself; skimping costs more later55
  • Unforeseen ground, change orders, claims and redesign20
  • Too few borings, or scope cut to win on price11
  • Reports that hand the design parameters back9
  • Recommendations padded to be conservative9
  • Independence and pressure on testing7
  • Trust in senior engineers you can reach7
  • Two reports, two answers on the same ground6

The threads were found by searching for these topics, so the counts show what comes up most often. They are not a survey. Classified by Jev; 30 passages fitted no theme.

From the case film. How buyers choose, and what they write. Competitor names are redacted for client confidentiality.

Four buyers, four fears

  • Design-build teamWe priced the ground blind.
  • Developer or ownerSurprises after closing, and a report that says piles to be safe.
  • Structural engineerThe report hands the design parameters back to us.
  • Public agencyA firm that goes quiet after award.
Logic

The strategy: one section per fear

The home page makes one argument. Ground problems cost least when they are found first; they get found late because exploration is cut, padded or handed down; a senior engineer who knows the local ground fixes that. Each fear gets the section that answers it, with something the visitor can try.

From the case film. Each fear, and the section built to answer it. Competitor names are redacted for client confidentiality.
  1. Fear 1 · Design-build teamWe priced the ground blind.

    What's under your site?

    Pick a county and the site shows the typical ground there, what to watch for and the first exploration step Kaveree would take. The pick follows the visitor down the page and fills in the contact form.

    Agencies and primes score a firm on knowledge of the local ground. Four geologic provinces sit within 60 miles of Kaveree's office.

    Kaveree's ground finder: a regional section from the Valley and Ridge to the Coastal Plain, with Frederick County picked and its typical profile, what to watch for and the first step listed
  2. Fear 2 · Developer or ownerSurprises after closing, and a report that says piles to be safe.

    How many borings does a site need?

    The visitor sets the number of borings and places them on a grid or by the geology. The drawing shades the ground a report would be guessing at and splits the cost of exploring from the cost of finding the problem later.

    It is the theme buyers write about most: 55 of 159 buying-side passages say good investigation pays for itself.

    Kaveree's borings drawing: six borings placed after reading the geology, the interpreted rock line against the real one, and guessed ground shaded
  3. Fear 3 · Structural engineerThe report hands the design parameters back to us.

    What lands on your desk

    Four sample report pages: the sealed cover, the boring plan, a boring log and the foundation recommendations, with the values an engineer designs to.

    Reports that punt the parameters back came up in 9 of the 159 passages. Showing the deliverable answers it before anyone asks.

    Kaveree's report stack: sample foundation recommendations with bearing pressure, settlement and slab values
  4. Fear 4 · Public agencyA firm that goes quiet after award.

    The engineer who scopes your borings signs your report

    Dr. Sushant Upadhyaya's title block with his licences and his signature, drawn on screen. The engineer who scopes the borings signs the report and takes the field calls.

    Agencies choose on qualifications and fear staff swaps after award. Most competitor home pages never name an engineer.

    Kaveree's principal engineer title block: Dr. Sushant Upadhyaya, PhD, PE, D.GE, his licences, his signature and 'Your site in Loudoun'
  5. And the clockThe funding has a date on it.

    What fits in your window?

    Kaveree's team told us clients work against fixed dates: funding windows, school summers, bid deadlines and emergencies. The visitor picks what sets the deadline and the weeks they have, and the schedule shows what finishes before the date and what follows it.

    A typical investigation takes three to six weeks from notice to proceed, so the deadline is often the first question.

    Kaveree's deadline planner: bid or award date picked, one week available, a schedule of borings, lab tests and the report against the date

The proof sits below these sections: six DOT design-build projects drawn as sheets, each with the owner, delivery method, location and Kaveree's exact scope, then the firm's certification numbers in one title block.

Kaveree's project sheets: six design-build projects for VDOT, ALDOT and THEA, with a highway interchange drawing and the firm's figures
Drawing set

The design system

The site is drawn like an engineering drawing set, the format Kaveree's buyers read every day. Every image is Kaveree's own drawing or 3D model.

  • Navy #203562Type and linework
  • Green #40B843The one thing that matters in a drawing: groundwater, the boring, the active choice. Never a large fill.
  • White paperThe primary colour, at the client's request

Type. Geist at regular weight for headlines, Geist Mono for drawing labels only.

Motion. Each animation shows a process: a boring drilled to rock, a log writing itself, a signature. All of it stops for visitors who ask for reduced motion.

Choosing the hero with Jev

We tested first-screen concepts with Jev through four lenses, including Kaveree's CEO and two buyer types. A wide film band with the headline over its lower corner scored highest. A full-screen takeover scored lowest.

  • Film band, headline over the film98 in 100
  • Full-screen takeover70 in 100
The Kaveree home page hero on a phone: Know the ground before you price itThe Kaveree home page on a phone, scrolled to the principal engineer and the ground finder
From the case film. How it looks, and the first-screen test. Competitor names are redacted for client confidentiality.
Method

The benchmark

We scored the new home page against 14 competitors and the old site, with the same questions and the same extraction for every page.

Each page was rendered in Edge with its navigation, footer and cookie banners removed, so every site was read the same way. Jev answered 15 questions per page: would four kinds of buyer contact this firm; is the page clear; does it prove its claims; is it distinctive; does it show the people, the local ground and a next step; does it look established; how broad is it; and three needs clients raise most (responsive, capacity, budget).

Every question was asked in three wordings, because a different wording moved an answer by 25 points out of 100 on average. Asking the same wording twice moved it by less than 1 point. That makes 1,020 judgments across 17 texts: 16 home pages plus the new page cut to the runner-up's length of 2,700 characters, to check that length wasn't doing the work.

Score out of 100: the average chance, across 12 questions asked three ways, that the model answered yes (the three client needs are scored separately). The films show the same scores as decimals. Competitor names are redacted for client confidentiality.
  1. Kaveree, new site82
  2. New site, cut to the runner-up's length78
  3. Competitor (name redacted)61
  4. Competitor (name redacted)60
  5. Competitor (name redacted)56
  6. Competitor (name redacted)56
  7. Competitor (name redacted)53
  8. Competitor (name redacted)51
  9. Competitor (name redacted)51
  10. Competitor (name redacted)50
  11. Competitor (name redacted)49
  12. Competitor (name redacted)48
  13. Competitor (name redacted)46
  14. Competitor (name redacted)46
  15. Competitor (name redacted)44
  16. Competitor (name redacted)42
  17. kaveree.com, old site38

Jev (jev-latest), 2 October 2026. 90% bootstrap range: new site 78 to 86, runner-up 47 to 74. The new site ranks first under every grouping of the questions and under each wording alone. Median competitor: 50.

Where it leads, and where it trails

The lead comes from three things most engineering home pages leave out. These are also qualities the site was designed for, so part of the lead is built into the test. On looking established, the field still scores higher.

QuestionNew siteField average
Shows who you'll work with9015
Knows the local ground9332
Gives a clear next step8732
Looks established6985
The benchmark, in six short chapters. Competitor names are redacted for client confidentiality.
Measured

Results so far

Scores are a model's reading of each home page's text, from 0 to 100. They are not enquiries or traffic: the site had not launched as of 2 October 2026.

Benchmark rank
1st of 16 home pages, under every question group and every wording
Benchmark score
82 in 100 (90% range 78 to 86). Runner-up 61, median competitor 50, old kaveree.com 38: the new page scores more than double the old one
Length-matched
78 in 100 when cut to the runner-up's length, still first
Leads
Shows who you'll work with: 90 in 100, competitors average 15. Local ground: 93 against 32. A clear next step: 87 against 32
Trails
Looking established: 69 in 100, competitors average 85
Hero test
Film band 98 in 100, full-screen takeover 70
Accessibility
axe WCAG A/AA clean on 40 of 40 pages
Client review
Kaveree's CMO reviewed every page before sign-off. Her one request was a full copy check, which we ran: 149 edits across 40 pages
Speed
First paint 1.0 s on emulated mobile 4G with the CPU slowed 4x
From the case film. Speed, accessibility and what gets measured. Competitor names are redacted for client confidentiality.
“The new templates have also reduced the time required to prepare qualifications and marketing materials for new opportunities.”
Dr. Sushant UpadhyayaCEO, Kaveree, in a verified 5.0 Clutch review of our brand work, which came before the website
Read the verified review on Clutch
After launch

What's next

  1. Launch once Kaveree signs off, with every old URL redirected.
  2. Measure with privacy-friendly analytics: county finder picks, the borings drawing, report views, and each contact path started and submitted. Review at 90 days.
  3. Add Kaveree's proof: a reply-time commitment, one written case study with numbers and real field footage.
  4. Run the benchmark again after launch and set it beside real enquiries.
3:54

The whole case, as one film

We first made these films so Kaveree's leaders could see why each part of the site exists before they signed it off. This is the same material as the clips above, end to end.

The whole case, as one film. The research, the fears, the sections, the benchmark and the proof, in the order of this page. No voiceover; every figure names its source on screen. Competitor names are redacted for client confidentiality.

Want your buyers' fears answered before they call? Talk to us about website design and development.

Curious how your site compares? Run the same audit on your site

Take the 3-min scorecardBook a 30-min call