Finished programs across five disciplines, each published with the argument that produced it beside the result. What you can browse here sits on top of 500+ delivered projects and something close to 150 brands, and the public selection is a sample chosen to show range instead of to flatter.

The five kinds of work, and what a write-up in each one is trying to prove.

• Branding. Positioning, messaging, naming, identity systems and full rebrands. The case study exists to show what shifted in how a market understood a company, which is considerably harder to photograph than a logo.
• Web design. Marketing sites, stores and custom builds. Here the useful part is the structure decision, meaning what got promoted, what got cut, and why a homepage ended up in that order.
• UI and UX design. Product and application work where usability carries revenue directly. Screens are the least interesting artefact in that file. The flows and the states around them are the work.
• Search and organic visibility. Organic programs, migrations, and the newer problem of being cited inside AI-generated answers instead of only ranked beneath them. Reported against enquiries, pipeline or revenue and never against impressions.
• Digital marketing. Campaign, paid and content programs, usually running on brands the same team built and therefore already understands from the inside.

Names already public run across all five. ArcelorMittal and TeraWulf on the enterprise side. Safeway and Albertsons in grocery retail. First Horizon Bank and the FHN Financial work in institutional finance. McGill University and Alberta Municipalities in the public and institutional world. Flipp in retail technology. Roncelli Construction in the built environment. Dress for Success in the not-for-profit sector.

The checkable part sits outside the case studies, because curation is not evidence. Trading since 2018, with more than fifteen years of practice behind the founding team. A Webby and four Awwwards for craft. Certification with RGD and CGD, which practitioners hold personally and projects never receive. A Clutch rating of 5.0 standing on 64+ interviews the platform verified itself, alongside 250+ five-star reviews across platforms and a Better Business Bureau grade of A plus.

What is missing is every project. A share of the strongest work is covered by agreements older than any interest we have in showing it, which gets dealt with further down this page.

By reading it as evidence of judgment under constraint instead of as a gallery. Almost every portfolio in this industry is attractive, because polish is purchasable and everybody has bought some. The differences that decide your outcome sit underneath the images.

Five tests worth applying to any agency, this one included.

1. Find the problem statement before the pretty part. If a case study opens on the deliverable and never says what was wrong, the agency either was never told or never asked. Both of those are informative.
2. Check whether a constraint is named. Real projects carry a budget ceiling, a stakeholder who would not move, a platform nobody chose, a launch date attached to a trade show. A write-up with no friction in it has been sanitized, and the sanitizing removed the part you would have learned from.
3. Ask what the agency argued for and lost. Every project has one. An agency that cannot name a recommendation the client declined has either never made a difficult one or is managing your impression of them, and the second is more likely.
4. Separate craft from consequence. A site can win awards and sell nothing. A plain page can carry an entire pipeline. Judge those two things independently instead of letting one stand in for the other, because they come from different skills and are frequently held by different people.
5. Look for the same standard twice. Consistency across unlike categories suggests a method. Brilliance in one category and mediocrity everywhere else suggests one good hire who has since moved on.

Read scope before screens, too. A three-week identity sprint and a nine-month enterprise program produce documents that look similar on a page and cost very different money, and comparing them as though they were equivalent is how buyers arrive at a number that has nothing to do with their own project.

The honest limit is that no portfolio can tell you what an agency is like to work with, how many hours a decision took, or how it behaved in the week a deadline slipped. Those things live in references and in the record clients built themselves through long-form verified interviews. If you would sooner have an outsider stress-test the shortlist with you before anything gets signed, an outside read on all of it costs a fraction of choosing wrong.

Because a finished screen is the cheapest part of any project to reproduce and the argument underneath it is the part nobody can copy. Anybody can hire a talented designer for a fortnight. Very few teams can explain why a page is ordered the way it is and defend that order to a board.

Consider what an image actually proves. It proves somebody had taste and access to a licensed typeface. It proves nothing about whether the company grew, whether the client's own team could maintain what was handed over, or whether the layout survived contact with real content six months later. A portfolio made entirely of pictures is asking you to infer competence from appearance, which is exactly the inference the pictures were built to produce.

So each write-up here carries the situation the company was in and what was measurably wrong with it, the audiences and the commercial decision the work had to support, the options considered including the ones we did not take, the constraint that shaped the outcome, and what the numbers did afterward against a starting point recorded before anything changed. That last item is why brand strategy and UX research appear in the descriptions at all. Without a documented before, every claim of improvement is a story told by the only party with an interest in it.

There is a real cost to working this way and it is worth admitting. Reasoning is slower to read than a carousel, and some visitors want the carousel. It also exposes decisions to disagreement. Occasionally a prospective client reads our thinking on a project and concludes they would have gone the other way, and we lose the enquiry. We would rather have that argument before a contract exists than discover it in week five with a design already approved.

Where a project's reasoning reads thin, it is usually because the client's own information was thin. Sometimes the analytics were never configured properly. Sometimes nobody had ever asked a customer anything. We say that instead of inventing a rationale after the fact, since a retrofitted rationale is the single easiest thing to write in this business and the least useful thing you could read.

Almost certainly, and the harder question is how much that resemblance should count for. The practice has worked in nearly every category we can name, and relevant examples get sent before a call instead of held back as a reason to book one.

The heaviest concentrations sit in a handful of places.

• Real estate and development, where a buyer is making the largest decision of their year and every page has to survive that level of scrutiny
• B2B and professional services, where a long committee purchase means the work has to persuade people who will never speak to you directly
• Technology and SaaS, where the product changes faster than most identity systems designed for it expect
• The not-for-profit sector, where one brand has to hold together across chapters with wildly unequal resources
• Institutions and education, hospitality and food, legal, financial services, industrial and entertainment, all of which are set out properly on the industry pages

Now the uncomfortable half. Sector experience buys three genuine things, which are vocabulary, a head start on regulation and seasonality, and a realistic view of how long a purchase actually takes in your market. It does not buy a better answer to your problem. An agency holding forty websites from your category has a commercial incentive to produce the forty-first, and the conventions it has absorbed include the expensive ones nobody in that industry questions any more.

So when you look at work from your sector, read it for whether a convention got broken on purpose. Sameness across a category portfolio usually means the agency is fluent in that category and has stopped arguing with it. Some of the work here that performed best came from importing an approach that was unremarkable in one industry into another where nobody had seen it.

A better question than whether your sector appears is which parts of your market we would have to study before we could be useful, and how long that study takes. Two weeks of proper reading closes most of the gap. What cannot be closed that way is who signs, what you get compared against, and what a regulator permits, and all three come out of research at the beginning of an engagement instead of being borrowed from somebody else's project.

The selection runs from a few focused weeks of work to programs that ran the better part of a year, and if your budget is modest the smaller projects are the ones worth studying. Large-budget work is easier to make look impressive, which is precisely why it is a poor guide to what a smaller number buys.

A focused piece. A brand refresh inside an identity that still has equity with customers, a microsite supporting one campaign, or the template set that stops a good identity being applied badly. Two to four weeks of work. Several of these read as modest on the page and moved more per dollar spent than anything else in the selection.

One discipline done properly. An identity system, a marketing site of six to ten pages, or a search program. This is where most engagements sit and where scope discipline decides the result. Three things done well inside this band beats seven done thinly, and part of what you are buying is somebody telling you which three.

A connected program. Brand and site together, which comes in under the price of the same two pieces bought in sequence from two vendors. One round of research serves both. One set of messaging gets written. And the design system arrives with the site instead of being fitted around it a year later.

An enterprise engagement. Brand architecture across divisions and regions, a design system dozens of contributors have to stay inside, accessibility compliance treated as a requirement, and a staged migration that protects existing search traffic. Months of work with governance and approval gates. The design is rarely the hard part on these. Alignment is.

Where a budget genuinely sits below the floor for the thing you admired, there are two honest responses and neither is a discount on the full scope. Narrow the work to the piece that compounds, or buy the diagnosis first so the eventual build is not guesswork. Platform choice moves the number more than people expect too, and that decision belongs on the build side rather than in a spreadsheet.

Dollar figures do not appear against individual case studies, because a number without its scope beside it misleads in both directions. On a call we will say which projects sat near your range and what they included. If the answer is that the work you liked cost four times your budget, you hear that plainly instead of receiving a proposal engineered to fit.

In most cases we can tell you, because a benchmark was taken before anything changed and the same measures were read again once it had. Where we cannot, the reason is almost always that a client stopped sharing data after handover, and a result claimed without a starting point is a story.

How the measurement runs, in order.

1. Record the before. Traffic and enquiries by page, conversion on the paths that carry money, search visibility, recognition and preference where a brand study is in scope, and whatever the sales team can say about how a deal currently gets won. All of it captured before a single decision is made, because a benchmark taken afterward is an estimate.
2. Watch the transition. On a website redesign that means daily checks through the first month, since a migration fault caught in week one costs a fraction of the same fault discovered in the next reporting cycle.
3. Retest against the original numbers. Same measures, same definitions, at ninety days and again between six and twelve months where the engagement continues. Comparing fresh figures against a different baseline is the most common way an agency produces flattering results without ever telling a lie.

Taken across engagements, the average increase in leads works out at 250%. The ceiling on that belongs in the same breath. It is a mean drawn from programs of unlike scale and unlike starting condition, several of which were inherited in poor repair, and it describes work already finished for other companies instead of predicting anything about yours. Anything we forecast for you is derived from your own figures, with every assumption sitting beside it so you can argue with the ones you think are wrong.

Two limits matter here. Attribution weakens the further you move from a click, so a brand program's effect on a sales conversation is real and only partly measurable, and we describe that as directional instead of dressing it up as arithmetic. And some outcomes had little to do with us. A company that raised a round, hired a sales team and tripled spend in the quarter it relaunched did not get those results from a website, and saying so is more useful to you than accepting the credit.

Where a client agrees, the write-up carries their numbers and their words. Where they do not, it carries ours with the method described so you can weigh it. The same measurement discipline is available on your own property, either from the search practice or through behavioural research on the paths your customers already use.

Reference calls happen, and the reference gets matched to your situation instead of to our best story. What we ask is that you wait until a scope exists, because the people who take those calls are doing us a favour with their own working time and we would rather not spend it on a browse.

Once a proposal is on the table, tell us what you are actually worried about and you get introduced to somebody who had that exact worry. A client whose project ran long. A client whose internal team had to operate the site afterward with nobody technical on staff. A client who arrived from another agency mid-project. A client who told us no and made us rework a direction. Those conversations are worth more to you than a cheerful generalist, and they are the ones we offer.

There is a structural problem with references and every buyer should know it. No agency hands over an unhappy client, so a reference list is a filtered sample by construction and anybody presenting it as proof is overselling. What it can still do is test consistency. If we tell you our timelines slip when content arrives late, and three unconnected clients describe that same pattern without being prompted, the account holds. If our description of ourselves and theirs do not line up, that gap is the most valuable thing you will learn all week.

For that reason the interviews clients gave directly carry more weight than a call we arranged. Sixty-four and counting, long-form, with named companies describing scope, budget and outcome, collected by a third party. We do not choose who gets asked and we cannot edit what comes back, which is the whole point of them.

Some clients cannot act as a reference at all. That is a contract term and not a verdict on the work, and it is worth knowing before you read a short reference list as a short track record. Geography is rarely the constraint, since the Canadian client base and the US roster are both deep enough to find somebody near you.

Two things we will not do. Publish a client's contact details, and dangle a reference call to move a deal along. Ask early if you want to, and you will get either an exception or a straight explanation of why we are asking you to wait.

Because a meaningful share of it sits under agreements that have nothing to do with how good it was. Confidentiality is the ordinary condition of enterprise and pre-launch work, and the practical result is that a public portfolio understates the practice behind it.

What stays out, and why.

• Work covered by a non-disclosure agreement, meaning a signed promise not to reveal it. Standard on enterprise engagements, and frequently covering the existence of the relationship as well as the output
• Work that has not launched yet, including naming projects waiting on trademark clearance and products still inside a release schedule nobody outside the company has seen
• Work for early-stage companies in a contested position, where publishing positioning ahead of launch hands a competitor a free brief
• Internal systems, portals and tools that were never meant to be looked at from outside the organization that commissioned them
• Work where the client has since changed direction, and showing the previous identity would embarrass them for no benefit to anyone
• Work we are permitted to show privately and not to publish, which is more common than people assume and is one reason a call covers more ground than a page can

That leaves the public selection skewed toward clients with permissive marketing teams and launches that went out on schedule. It is a sample, and it is not a ranking of our best projects. Under an agreement in the other direction, confidential work can be walked through in a meeting where the material stays in the room, and that closes most of the gap for a serious buyer.

The reverse of this is worth saying too, because it cuts against us. An agency with an enormous public portfolio may simply work with less careful clients, and that is not a compliment to either side of the arrangement. Volume of published work is a weak signal and it is the one most agencies compete on.

Permission gets asked for instead of assumed. A client who says no stays out even where the work would have been the strongest thing here. And when a client leaves us for somebody else, the case study stays up, because the work happened and a record that only holds current relationships is a marketing document instead of a record.

Some of it, and the honest complication is that the projects we are least proud of are often the ones the client was happiest with. A portfolio is a record of what an agency was permitted to do, and permission and quality are different things.

Two kinds of work sit at the edge of the selection or outside it.

Work compromised into ordinariness. Where a recommendation was overruled, a scope was cut to fit a number that had already been set, or a committee removed the one decision that made the thing work. What ships is competent and forgettable. We do not lead with those, and we will describe them plainly when somebody asks what a constrained version of a project looks like, because that description is genuinely useful to a buyer working inside a hard limit.

Work that has aged. Sites built five or six years ago to conventions that have since moved. Identity systems that were correct for a company at the time and would be wrong for it now. Some of those stay up deliberately, because implying an agency has only ever produced current-looking work is a claim no practice with real history can support.

What we will not do is publish a case study nobody here can stand behind in order to fill a gap in a category list. If you ask whether we have done a particular kind of project and the truthful answer is once, badly, years ago, that is the answer. The alternative is you discovering it in month two, at which point it costs you money instead of costing us an enquiry.

The sharper version of this question belongs on a call anyway. Ask what went wrong on a recent project and what changed in the process because of it. Ours are boring and specific. A visual identity shipped without the day-to-day templates that would have kept it applied correctly, which is why collateral is now scoped alongside the system instead of after it. And a rebuilt site launched before the content plan was finished, which is why content readiness is now a gate and not a hope.

An agency that answers that question with a story about a difficult client has just told you how it will describe you.

The people named on the case study, and the reason that is worth stating at all is that plenty of agency portfolios contain work the agency did not do. Attribution in this industry is loose enough to be worth interrogating on any shortlist.

People who have since left. A portfolio accumulates and a team does not. A project shipped four years ago by a designer who has moved on legitimately belongs in an agency's record, and it tells you very little about what arrives next month. The question that resolves it is which of the people who made the work you liked are still there, and it is a fair thing to ask anyone.

Subcontracted production. Common, rarely disclosed, and the most frequent reason a launched site looks nothing like the pitch. Nothing core leaves this building. Strategy, design, engineering and search are all done by our own senior people across the four offices. The exceptions are named specifically in a proposal, meaning specialist photography, video, illustration in a particular hand, and trademark counsel where clearance is needed.

Projects with several vendors. We have built sites on another agency's brand guidelines, and handed finished identity systems to internal teams who built their own site on top of them. Where that happened, the write-up says which part was ours. An agency claiming a whole outcome it contributed one component to has told you something about its reporting habits before you have signed anything.

Juniors working under a senior name. The pitch team and the production team being different people is the standard substitution in this business. Here the people who scoped a project ran it through to launch, which follows directly from only employing senior practitioners and is the reason our openings are few and slow to fill.

On any project in the selection we can tell you who was in the room, what each of them did, and whether they still work here. Where a case study turns on interface and interaction work or on engineering specifically, the person who built it can join your call and answer for it directly, which tends to end the question faster than any claim on this page.

Tell us, because it is useful information about your taste and a weak brief for your project. That style was an answer to a different company's problem, with a different audience, a different competitive set, and a different list of things it had to avoid looking like.

Here is the commercial reason it matters. Positioning is relative, and an identity does part of its work by being unlike whatever sits next to it on a shelf or in a set of search results. Borrow the visual language of a company in an unrelated category and it can land beautifully. Borrow it from a company in yours and you have paid to look like the market leader, which is the cheapest available way to be understood as the alternative to them. That is a positioning problem wearing a design costume.

So references get taken apart instead of copied, to find the property you actually responded to.

• Whether it was restraint, confidence, warmth or precision, which are strategic positions and do transfer between categories
• Whether it was a specific device, meaning a typeface, a colour palette, a motion behaviour or a photographic treatment, which usually does not
• Whether the thing you liked is a solution to a problem you also have, which is the only test that finally matters

Categories with strong visual conventions are where this goes wrong most visibly. Food and drink brands arrive holding a competitor's packaging more often than any other sector, and the honest response is that the reference proves the convention works, which is simultaneously the strongest argument for breaking it. Occasionally the right answer is to follow the convention, and we will say so, because being legible on a shelf beats being interesting in a presentation.

Bring the references anyway. Three or four, including one you actively dislike, with a sentence on why for each. That is a better input than a written brief, and an identity system built from it will hold up under questioning in a way a copied one never does. What we will not do is reproduce a look because you found it. If that is the engagement you want, a production studio will serve you better and charge you less.

Send the two or three projects that stopped you, plus a paragraph about where your own business currently is, and we will tell you plainly what a comparable engagement would involve. No written brief required, and the first call is a diagnosis instead of a pitch with slides.

1. Say which projects, and say why. The reason matters more than the selection. Admiring the structure of one site and the tone of another carries more information than most written briefs do.
2. We read before we talk. The site, the existing brand material, current search visibility and the three competitors you actually worry about all get read beforehand, so nobody burns the first twenty minutes of a call on material that was public all along.
3. A working call. Somewhere between half an hour and an hour on what you are trying to change, what is in the way, the deadline that is real and the budget that is real. You get a straight assessment whether or not you appoint us.
4. An itemized proposal. Scope, phases and price set out per component, so a line can come out without reopening everything else, and so you can hold it against another agency honestly. One total against another total is not a comparison.

On timing, a proposal usually arrives inside a week of that call. A tightly scoped identity sprint lands in two to four weeks. Add research and strategy and it becomes one to three months. Naming stretches further again, because trademark screening moves at its own pace. Sites typically run eight to twelve weeks. Search carries a six-month minimum, since the early months are cost and the return lands later, and real movement in organic visibility takes somewhere between four months and a year. Where an agency guarantees you the first page inside a month, it is either not telling you the truth or bidding on terms with no traffic behind them.

Your side of it needs three things, meaning analytics access if a site is in scope, any brand or research documents you already paid for, and one person with authority to approve. What slows it down is unresolved authority, meaning several people who can object and nobody who can decide.

Sometimes the honest recommendation is smaller than the project you admired, or a different project entirely, and occasionally it is that you should not spend the money this year. You hear that on the call and not after a contract. Start the conversation when you are ready, and the full range of what it could involve is worth reading first.