
With over a decade of experience, we are proud to have worked with many outstanding businesses worldwide!

Brand Vision is an award-winning branding, web design and marketing agency, established in 2018, with its head office in Toronto and staffed teams in Chicago, San Francisco and Miami. Roughly 150 brands and upwards of 500 projects have passed through the practice since, and the founding team arrived with over fifteen years of experience already behind them.
The practice was built around a specific and unglamorous problem. A company hires a design studio to build a brand, receives a handsome document, then hands that document to a web shop that had no part in writing it. The identity survives the first two screens and comes apart on the fourth, because nothing in the guidelines anticipated a filter panel, a form error state or a table with nine columns. The site meanwhile gets structured by people who never read the positioning, so the argument the company spent real money settling never appears where a buyer would meet it. Both suppliers did competent work. The client paid for the seam between them.
Owning both ends moves that argument inside the building and into week two, when a strategist, a designer and an engineer are looking at one problem and any of them can still change the answer. A colour pairing approved on a presentation board and failing contrast in a live interface gets caught before sign-off instead of after forty components have been built with it. What has shipped since is the fair way to judge whether that claim holds, because integration is easy to assert and either visible in the output or absent from it.
Where the integrated model is the worse purchase. Three situations, named plainly.
Six disciplines sit under that roof, meaning brand, web design, development, interface work, organic search and marketing, and the people behind them are named rather than described in the abstract. The largest share of the work is still produced in the Toronto practice, where the company started and where the head office remains.
In the overwhelming majority of cases, no. The work is done by the people we employ, and the rare exception is a project needing a specialism nobody sensible keeps permanently on staff, in which case you are told before it happens.
Strategy, design, interface work, engineering, content and search all sit with people on payroll here. That is the answer for very nearly every engagement, and it is the reason the people who scoped your project are the people who deliver it.
The exception exists because a project occasionally requires a craft with its own equipment, its own rate card and its own working rhythm, and commissioning a specialist for it produces better work than asking a generalist to attempt it. When that happens, three things are always true.
That is deliberately the whole policy, because it is the part that stays true. Which specialisms come up and who sits on that list moves over time, so publishing a roster would be out of date by the time you read it. What does not move is that the disciplines you are actually buying are in-house and any exception is disclosed before you sign.
The test worth running on any agency, and it costs nothing. Three questions, asked before a contract exists.
We answer all three in writing, including on bilingual engagements run into the Montreal market, the most common situation where an outside specialist joins a project at all. How we hire onto staff explains why the in-house list runs as long as it does, and for your own project, put the question to us directly.
An agency, and the operating facts settle that more honestly than the word on anybody's website does. Six disciplines under one roof, four offices, ongoing programs carrying a minimum term and a reporting rhythm are agency traits whatever a firm prefers to call itself.
Both words have drifted into marketing, so here is roughly how buyers use them. A studio is small, craft-led and narrow, usually one or two disciplines deep, and the person whose name is above the door is on your project every week. An agency runs several disciplines at once, carries structure and process, and can hold a relationship for years instead of for a project. Plenty of firms call themselves studios because it sounds less corporate while operating exactly like agencies, and plenty of agencies are three people and a network of freelancers.
What the agency shape buys you. One contract covering strategy, design, build and search, so the coordination cost sits with us instead of landing on your desk every Monday. Continuity when somebody is away. The capacity to deliver a project and then keep running the marketing behind it for years. And coverage across time zones, which is the reason the US head office exists at all instead of everything routing through one city.
What it costs you. Structure takes time, and a discovery stage, a written scope, sign-off gates and a reporting cadence are all overhead you are funding. A senior generalist group will not out-craft a specialist inside that specialist's single lane. And no principal is on every project every week, because a firm running several programs at once cannot promise that and mean it.
What a small specialist studio does genuinely better. Worth hearing before you decide.
If your project is one identity for one audience against a settled brief, a good two-person studio is very often the better purchase and you will hear that from us. The useful move for any buyer is to ignore the noun and interrogate the mechanics. How many people, named. Who can overrule whom. What happens the month after launch. Whether a minimum term applies. The range of markets we run in and the finished projects answer that better than a category label ever will.
Most of it is checkable by somebody other than us, and the items hardest to manufacture are the ones a buyer should weight heaviest. Here is the whole list with an honest read on what each one actually proves.
The ceiling on all of it, stated plainly. None of these prove a client's revenue moved. Recognition is a survivorship measure, since the work entered for an award is the work that went well, and the engagements that turned difficult are not in the sample anywhere. A site can win a jury prize and convert badly, and both halves of that sentence happen.
So use the list to decide whether we are worth an hour, then test the harder things yourself. Ask for a reference from an engagement that got uncomfortable. Ask what we would do differently on it now. Read the reviews that are not five stars and see whether the criticism sounds like something you would mind. Anyone running a supplier check on a long-cycle business-to-business purchase follows that sequence as standard, and it is a good habit for everybody else.
There are four situations where you are better off hiring somebody other than us, and we would rather say that on the first call than have you find it out in month two. None of them is a judgment about the business. Each is a case where what we are built to do and what you actually need do not line up.
Somebody whose deciding factor is the lowest number. Price is a legitimate primary criterion and there is a competent supplier at every level of this market. If the total at the bottom of the page is what decides it, we will lose that comparison, and saying so early beats letting you run a process whose outcome you already know.
Somebody who needs a full-time designer inside the company. Where the work is continuous, small-grained and needs a person in your stand-ups who knows the product without being briefed, an agency is an expensive way to buy that. A salaried hire is cheaper by month five and better by month eight, because accumulated context is the whole asset. Hire one, keep an agency for the pieces that spike, and the standard we hire to is a fair description of who to look for.
Somebody who has already decided and wants execution hands. Some organizations have a strong internal creative lead and a settled direction, and need capable production against it. That is a real way to work. It is poor value here, because most of what you would be funding is an opinion you have asked to stay quiet, and we will offer it anyway, which becomes tiresome by the second review.
Somebody whose actual problem is the product or the price. This is the uncomfortable one. If customers leave because the thing does not do what it promised, or the pipeline is thin because price and perceived value have come apart, sharper positioning and a better site mostly raise the number of people who discover that. Presentation makes a good product easier to choose and a weak one easier to reject. We have said this to companies who were ready to sign, and lost the work, which is the only evidence that the position is real.
There is a fifth that is about timing instead of fit. A company midway through a reorganization, a funding round or a change of leadership should usually wait, because what it stands for is about to move and it will pay for that answer twice.
The quickest way to find out which column you sit in is to describe the awkward version of the situation, budget and internal politics included, and say so in the first email. Sometimes the straight read is that you should not buy anything yet.
Not where two clients would be hunting the same pool of customers in one region at one time, and an overlap of that kind means declining the second engagement. The check runs before a proposal exists instead of after a contract does, and the awkward part is the middle ground, so most of this sits there.
How the check works. In the first conversation you name the companies you would genuinely mind us working for, with geography and segment attached, because a rival on a category list is often no rival for one customer. We hold those names against current and recently closed engagements and answer before anything is written. If an overlap is already live you hear it then, which costs us the pitch and is the only version worth anything.
Why a genuine conflict is a working problem instead of a matter of principle. Positioning is relative, so a strategist building your case against a named rival cannot build the reverse case for that rival the same quarter without serving one of them worse. That person also knows what the other side is about to launch, and goodwill does not unknow it. A firm claiming it can hold both is protecting revenue and calling it a capability.
Where an apparent overlap is nothing, which is most cases. Two firms in one sector serving different buyer pools, segments or regions rarely touch each other. A practice in Toronto and a practice in the Chicago market are seldom in contention for one customer. A manufacturer selling through distributors and one selling direct share a category and little else.
What confidentiality means once a team has depth in a sector comes down to two lists.
The honest trade sits under all of it. A firm with real depth in your category has worked with businesses adjacent to yours, and one promising a clean sheet there is telling you it has no experience in it. Long-cycle categories feel that hardest, which is why firms selling into buying committees raise it early. Two questions are worth asking any agency.

Six steps sit between a first message and a kickoff, five of them move quickly, and the one that decides your start date happens inside your company where we have no influence at all. A few weeks from enquiry to work opening is normal where one or two people decide. Several months is normal where they do not.
How it actually runs, with honest timing against each step.
Where it stalls deserves naming, because it is almost never on our side and it is almost always something familiar. A budget waiting on the next quarterly cycle. One director who has to be comfortable before a finance lead will sign. A board that meets monthly, which is why organizations answering to a board should count backwards from the meeting date and not from the launch date. Or two people who each assume the other owns the decision.
The quickest starts are founder-led companies where one person can approve on the call and work opens the following week. The slowest are organizations with genuine procurement. Neither is wrong, and knowing which one you are is what stops a date being promised that was never available in the first place.
We send one reminder and then leave it with you. A file that goes quiet for a year picks up where it stopped instead of starting over, and the work we actually sell is worth a skim while you decide.
A first call is a diagnosis running thirty to forty-five minutes, the questions travel in both directions, and no deck gets presented. There is no credentials reel and no sequence of follow-ups waiting behind it. The person opposite is a practitioner who would run the work, with nothing riding on which way you answer.
What gets asked, and having a version of each ready is most of what separates a good call from a slow one.
Qualification runs the other way too, and it should. You are working out whether the person opposite has met your problem before, what they would refuse to do on it, and who would be on the project in practice. Ask all three. A firm with no answer to the second one has no point of view to sell you.
Then the call ends one of four ways, and two of them are not a proposal. A scope gets written. Or the recommendation is smaller than what you asked about, which frequently means a costed search review or walking the two journeys that carry your revenue before anybody rebuilds anything. Or the answer is to spend nothing this quarter and revisit once a decision inside your business has settled. Or we are the wrong firm, and you hear that with a name to try instead wherever we have one.
A fair share land in the last three. That is the point of running it this way, and it is why the call is worth taking even where the outcome is that you buy nothing at all.
Nothing here is mandatory, and arriving with none of it costs you nothing except a slower conversation. Six items, each one removing a round of email that would otherwise happen over the following fortnight.
One more thing is worth digging out, and it is the item clients forget they own. Old research, a positioning document from three years ago, personas, win-loss notes, a survey somebody ran and never circulated. Companies routinely hold evidence about how buyers describe them that nobody has opened since the person who commissioned it left. Building on it costs less than repeating it, and now and then it says the opposite of what leadership currently believes.
You can leave the written brief, the vendor shortlist and the budget spreadsheet at home. A brief describes a solution, and choosing the solution is the work you would be paying for. If all you can bring is a sentence about what is going wrong, that is a perfectly good starting position and the call simply spends longer on questions.
Small projects are welcome, and there is a floor underneath which the honest answer is to hire an independent specialist instead. Small here means narrow and finished. It does not mean a thin version of something large.
What genuinely small work looks like at this end of the practice.
Where the floor sits, and the mechanism matters more than a number. Every engagement here carries roughly the same fixed overhead whatever its size, meaning a scoping call, a written scope somebody senior wrote, a contract, a named team, structured reviews, documentation and a handover. That overhead barely shrinks as the work shrinks. Below a certain scope it becomes most of what you are paying for, and at that point you are buying administration at practitioner rates. A capable freelancer or a two-person studio carries almost none of it, gets close to the same outcome, and charges a fraction. We will say that on the call, and the reason is arithmetic instead of modesty.
No dollar minimum is published here on purpose, because the real test is whether the work has enough shape to survive that overhead and not whether it clears a threshold. The qualitative version is this. If the thinking is already done and what remains is a handful of production hours, you are below the line. If a decision still has to be made and defended, you are above it however small the deliverable looks.
Small is genuinely the right purchase for companies with one product and a short runway doing one thing properly before starting the next, for small charities and community organizations where every dollar is accounted for to somebody who will eventually ask, and for any business holding a clear priority against a finite budget. That last description covers most of the good small engagements we run.
Where small becomes a trap is a sequencing problem instead of a size problem, and the next answer deals with it directly.
Starting with one piece is frequently correct, and the strongest version of it is buying a diagnosis instead of a deliverable. An audit ends in a costed roadmap you can hand to us, to your own team, or to an agency you have not met yet, and it costs a fraction of whatever you arrived intending to commission.
That is often the right first purchase for somebody unsure what they need. A paid diagnostic concluding that you should repair six things instead of replacing forty pages has covered its fee before you reach the second section. Where the question is narrower, an SEO audit or a usability review answers it for less again. The test of whether one was written honestly is whether a stranger to the project could act on it with no conversation with us at all.
The harder half of the question is that some small pieces are safe to buy first and some sit downstream of a decision nobody has made yet. Buying the second kind means paying twice.
Safe to start with, because the evidence that settles them already exists.
Where the sequencing trap bites, and each of these gets bought twice.
One question sorts your piece into the right column. If the strategy changed next quarter, would this still be the right thing to have bought. Where the answer is yes, buy it now and buy it small. Where the answer is no, the small piece is a deposit on a second invoice, and settling the decision above it first is usually the cheaper purchase anyway.
Book earlier than feels necessary, because a senior-only team caps how many projects can run at once and the calendar here is a real constraint instead of a negotiating tactic. There is no junior bench waiting to absorb an extra file, so when the group is full you get a date and not an eager yes.
Where it usually sits, without inventing precision. An engagement needing a strategist, designers, a developer, a writer and a search specialist is booked some way ahead, and the busiest stretches of the year push that further out. A narrow piece needing one or two people moves considerably sooner. Any agency claiming identical availability in every month of the year is describing a sales process instead of a schedule.
What can start quickly and what cannot.
Three things worth knowing about how the queue behaves.
If your date is fixed and close, put it in the first message. You will get a straight read on whether that scope and that date can both exist, and where they cannot, the useful conversation is which parts ship by the date and which follow afterward.
The last point is the one companies underrate. Most of the gap between an enquiry and a start is spent on your side, so an organization that settles its internal decision inside a fortnight will begin sooner than one that takes two months over it, even where the second one wrote to us first.

Scope gets settled before any number exists, every phase carries its own price on its own line, and the proposal names in writing the conditions that would move it. The order of those three matters more than the total, because a figure produced before anybody understands the business is arithmetic applied to a guess.
How a quote gets built here.
Which shape is correct depends on whether the value lands or accumulates.
Fixed scope suits work with an ending, meaning a brand program, a site build, an audit. A deliverable, a date, a total, and when it is done it is done.
A monthly arrangement suits work whose value compounds instead of arriving, meaning search, advertising and continuing design support. Pricing that as a project would mean pretending a compounding activity has a finish line. For software and technology companies whose product outruns their own messaging every two quarters, the monthly shape is usually the honest one.
Choosing wrong costs money in both directions. A retainer sold for finite work keeps an invoice alive after the job is finished. A project fee applied to continuing work buys the setup and stops before the return arrives.
The habit worth naming, because it is close to universal, is the agency that produces a number from a written brief and a forty-minute call. It is fast, it wins deals, and it is why so much work gets requoted in week six. We would sooner spend an unpaid week understanding what you actually sell and then quote something that survives contact with the project.
Sometimes, and rarely. Most work here is priced against a defined scope, and on the occasions where hourly genuinely is the right instrument we quote the estimated hours up front so you know the number before anything starts.
How the hourly version actually works when it is used. The work gets assessed, you are told how many hours we think it will take, and the invoice is built against that estimate. If it runs slightly over, that is ours to absorb and it does not appear on your bill. The estimate is the number you agreed to, so the arrangement behaves much more like a fixed price with an honest unit attached than like a meter running in a room you cannot see.
Where hourly is the correct instrument, because it does exist.
Why it stays the exception. A rate card pays a supplier for taking longer and quietly fines them for being good at the work. A designer who has drawn four hundred identities reaches the right answer in a fraction of the time somebody in their second year needs, so billing purely by time means the experienced one earns less for the better outcome. There is a second effect clients feel before naming it, which is that an open meter makes every question cost money, so you stop asking.
What happens on most engagements instead. Fixed-scope work is priced per phase against named deliverables, agreed before anything starts, and it moves only when the scope moves. Continuing work carries a monthly figure set against the plan and revisited as the plan changes, which is how the continuing side of what we run is charged. Advertising follows the same logic and lands on flat fees on media work, since a share of spend pays an agency to recommend a bigger budget.
The honest cost of a fixed price. It absorbs the risk of the job being harder than expected, so on work that turns out straightforward you would have paid less by the hour. Most clients decide knowing the number in advance is worth that trade.
There is no standard schedule here, because the shape of the project decides it, and a brand identity, a replatforming build and a twelve-month search program are three different commercial shapes. What holds constant is that invoices follow phases closing instead of dates arriving, so what you pay is attached to work that has landed.
The exact schedule, the terms and the split between phases get set for your engagement and written into your agreement. Any firm publishing one set of terms for every project is applying a template that will fit somebody badly.
Why calendar billing goes wrong. A monthly invoice on a fixed-scope project charges the same amount in the month everything shipped and in the month your approver was away. Nothing on the paper distinguishes the two, so the conversation about progress and the one about money never meet. Tying them together forces both to happen at once.
Phases are the unit almost everywhere. Inside the brand practice that means research, positioning, identity and rollout. On a replatforming project it means discovery, architecture, design, build and migration. Each one closes with a written sign-off you gave before any invoice references it. Continuing programs invert the logic, since a monthly service bills monthly, with the report arriving in the same week so you are reading the result and the invoice together.
What should be on the paper, and this applies to every agency you will ever hire.
One question separates the careful firms from the rest. Ask what the final payment is attached to. If the answer is a date, the last stretch of your project is being paid for before anybody has judged it. If the answer is launch, handover and the account transfers being complete, the incentive is pointed at finishing properly instead of at finishing quickly.
We do take one, and the plain reason is that a start date books a team that cannot be in two places, so holding your window costs us the ability to sell it to anybody else. The amount and the conditions attached sit in your agreement instead of on this page, since both move with the size and shape of the engagement.
Capacity is the actual mechanism, and it is finite in a way clients rarely picture. This is one senior team, and it is the same team whether you sit in Toronto or in a market we cover remotely. When your project is scheduled, other work is declined or pushed for that window. If the client then goes quiet for five weeks, the window is gone and cannot be resold at short notice, and somebody carries that cost.
The second function is less comfortable to say out loud. A deposit filters. A meaningful share of enquiries are genuine interest attached to a budget nobody has approved yet, and this is where that becomes visible. A company that cannot release a first payment generally cannot release a decision either, and learning that in week one is better for both sides than learning it in week seven with a designer sitting idle. We would sooner lose a project at that gate than staff one that dissolves.
What you get in exchange, which should be specific instead of a feeling.
Concede the obvious. A deposit protects the agency more than it protects the client, and any firm claiming otherwise is selling. What makes it fair is the structure around it, so press on the specifics before you sign anything. Find out what happens to the money if you cancel before work starts, whether it is credited against the first phase or charged on top of the total, and what you are owed when the agency is the one that slips. For brands selling food and drink working to a production and listing calendar, the date being held is frequently worth more than the money itself.
Five things legitimately move a figure after signing, every one of them a change to what the work is, and the agency having underestimated its own job is not among them. All five get raised before the work happens instead of surfacing on an invoice afterwards.
What does not justify a change order, and this is where to hold any agency to account. We mispriced it. The work was heavier than we expected. Discovery ran long. Those are estimation errors, and the risk of a fixed price being wrong belongs to whoever set it. If we scoped redesigning an existing site and the legacy codebase is worse than our audit suggested, that is our read and not your invoice.
Whoever you hire, insist that nothing extra gets built before it is agreed in writing with a number attached, so declining the change and keeping the original scope stays available to you. Insist too that a requote arrives as a decision with doing nothing among the options. A mid-project requote on a rebrand is frequently a sign the strategy phase was thin, and that is worth saying out loud.
Both happen, both get handled as ordinary business events, and how each one works depends a good deal on the project and where it has reached. Companies lose a funding round, change a chief executive, or find a larger problem than the one they hired for. None of it should become an argument about paperwork.
Pausing. Work stops at the nearest completed phase instead of mid-sentence, because a half-built design system is worth less than nothing to whoever picks it up. Everything finished is delivered and documented, so restarting is a briefing instead of an excavation. A pause is not free, though. People booked on your window get rebooked, so resuming means a new date in a schedule that has moved on. Pausing an identity change already in flight is the costliest version, since a company living in two brand states at once is worse off than in either.
Cancelling. The engagement settles at the last completed phase. You are invoiced for what exists, the files and accounts transfer, and nothing is held back to make leaving awkward. Concepts you saw and declined stay with us, and third-party licences keep their own terms.
The six-month minimum on ongoing programs. The one hard commitment in the arrangement, and it protects you from the commonest way a company concludes this work does not function, which is funding the build-up and leaving before the compounding starts. For a product company mid-repositioning, two quarters is close to the minimum honest read anyway. Where a business cannot commit to it, meaning restaurant and grocery clients running season to season, a fixed-scope piece is the better purchase.
The part that matters more than any clause. Almost every one of these conversations ends with the work continuing in a different shape instead of stopping. A budget gets rephased, a scope narrows to what this quarter can carry, a program pauses at a sensible boundary and picks up when the business is ready. We would far rather find that version with you than enforce a term, and that is most of the reason clients stay and a fair part of why the record reads as it does, meaning a 5.0 rating on Clutch and well over 250 verified five-star reviews. A firm reaching for the contract before it reaches for a solution is telling you what the relationship is.
Four questions to put to any agency before signing, read in the contract instead of taken on a call.

Partly, and the honest split is that four things hold on every engagement while the shape around them gets rebuilt for the problem in front of us. An agency running an identical sequence on every client is selling a template and charging for judgment, which is a comfortable business to operate and a poor one to buy from.
What is genuinely fixed, whatever the work turns out to be.
Everything else moves. Depth is the first variable, since a category with three competitors and a category with two hundred need different amounts of evidence before anyone is entitled to an opinion. Sequence is the second. Content leads and design follows it on a site whose actual problem is that nobody can find anything, and the order inverts where a fixed launch date means the visual direction has to land while the writing catches up.
The third variable is which phases exist at all. A company that already ran a serious round of user research does not need us repeating it, so we read theirs and start from it. A business with settled positioning and genuine internal agreement can skip most of a discovery phase and should be told so. A multi-property hospitality operator usually needs a brand architecture phase nobody else needs, meaning the question of how a parent brand and its individual properties relate to each other.
There is a simple test of whether an agency adapts anything. Ask for two past proposals from projects unlike each other, and see whether the phase names change or only the client name does.
As often as the project needs, agreed at kickoff and written into the scope instead of left to settle by accident. Some engagements run two or three check-in calls a week. Others are better served by a call every couple of weeks with a channel in between, and the difference is the work and not the client.
Cadence is a design decision and the work decides it. A brand sprint with daily decisions and a compliance reviewer in the loop needs more contact than a long build phase where the useful conversation happens at either end. Getting it wrong either way costs you, because too little hides problems and too much turns your week into status meetings you are paying for.
The four layers, each doing a different job.
Now the part most agencies leave out. Quiet stretches in the middle of a design or build phase are real. Two people can be a fortnight into the engineering half of a project with nothing worth showing you, because a half-assembled template teaches you little and invites comment on decisions that are not finished. An agency filling that gap with activity updates is manufacturing reassurance you are paying for.
So expect the middle of a phase to be quieter than either end. What you should never accept is quiet that outlasts whatever cadence you agreed, because that standing session is the floor and it landing on schedule is the signal that nothing has gone wrong.
If it slips with no explanation, treat that as a fault and say so to the person running the work. If a straight answer does not come back inside a day, use the second name, and get that name at kickoff from any agency you hire. Regulated categories carry an extra loop, which is why clinics and wellness brands build the compliance reviewer's turnaround into the rhythm from the start.
Figma for design, working prototypes for anything with behaviour, whichever content management system your own team will operate, and a shared tracker holding the schedule and the open decisions. Tool names matter far less than the principle under them, which is that you should finish an engagement holding access to the systems your assets live in and not a folder of exports.
Interface and layout design. Figma, structured as a component library with named styles instead of a pile of loose frames. Organized that way, a colour change propagates instead of being repeated ninety times by hand, and your next designer can work inside it. You get a seat during the project and keep it afterwards.
Identity artwork. Vector source files handed over editable, plus the production formats a printer, a sign fabricator or an embroiderer will ask for. An identity you cannot open is one you cannot extend, and flattened images are where that goes wrong.
Prototypes. Anything with behaviour gets prototyped before it is built, because a static screen cannot show what happens on error, on empty or over a slow connection. That is where the behaviour of an interface gets settled, and clicking through one produces sharper feedback than looking at a picture of it.
The platform the work ships on. Webflow, WordPress, Shopify or WooCommerce, chosen against who operates the site after launch instead of what we enjoy building in. Anything custom is built on our own engineering bench and documented well enough for a developer who has never met us to take it over.
Measurement. Analytics, tag manager and search console, registered to your company from the day each is created.
Tracking and documents. A shared board carrying the schedule, the open decisions and who owes what by when, plus a document space holding the scope, the notes and every written approval. The product name matters least. What matters is that you can look at it without asking anybody, and that nothing important lives only in somebody's inbox.
Where we bend and where we do not. We will work inside your ticketing system and messaging platform where you have them, because making a client change tools for one supplier is slow and rude. We will not run design in something that cannot hold a component library, since that cost arrives on your desk long after we have gone. Before signing with anybody, establish whether you get a seat in the design file, whether the platform account is in your name, and what transfers at handover. Those answers matter most on the product side of the practice, where what you own is a living system somebody has to keep extending.
Unlimited in most cases, unless your agreement names a number, which happens on smaller and tightly defined pieces of work. Where it is uncapped, and it usually is, the reason is that a revision count is a budget for disagreement and creative work gets worse when disagreement is rationed.
What a cap does to your side. You start saving notes. Something gets approved that you have quiet reservations about, because a round feels like currency and spending it on a small thing seems wasteful. The reservation does not evaporate. It resurfaces in month three attached to work that is now expensive to unpick.
What a cap does to our side, which is the half nobody mentions. Once rounds are counted, every comment becomes a question about whether it falls inside scope. The agency starts defending the previous version instead of improving it, and a review turns into a negotiation. Nobody produces good work from that posture.
Structure is what holds an uncapped arrangement together instead of arithmetic.
Phases with a gate at the end of each one. Research, positioning, identity and rollout on a brand program. Discovery, architecture, design, build and launch on a site. Each ends with a written approval from the named approver.
Unlimited inside the open phase. While a phase is live the work continues until the answer is right. There is nothing to count and no reason to hold anything back.
Reopening a closed phase priced as a change of scope. Not a punishment and not a refusal. It gets quoted before anything is rebuilt, and declining it and keeping the approved version stays available. Reopening an identity after rollout across real touchpoints has begun is the costly version, because every applied asset moves with it.
Where a number is named. Smaller and tightly bounded pieces are occasionally scoped with a set count, because the work is defined enough that the number is honest instead of restrictive. It sits in the agreement and it is never something you find out later. Even then it is a starting point and not a wall. Whatever the paperwork says, we are flexible about it, because the thing being protected is a project that succeeds and not a technicality somebody gets to win.
What makes the uncapped version work is the feedback process underneath it, which is deliberately detailed. Work is presented in batches with the reasoning attached, comments are gathered from everyone at once instead of arriving one person at a time, and each round closes with what changed and why written down. None of it functions if approval is treated as a formality while every question stays quietly open, and on product work the unresolved thing underneath is usually the states and edge cases nobody specified.
One named lead who is personally producing part of your project, and the question that actually matters is what that person can settle without leaving the call. A contact who has to check everything is a relay, and a relay quietly adds a day to every question whether anybody intends it to or not.
Your lead here can change a design decision, reorder work inside an open phase, call something wrong and rebuild it at our cost, bring another specialist in for a day, and answer a technical question about the build without consulting anybody. Those decisions make up most of a working week, and answering them inside the session instead of two days afterwards is where a schedule is quietly won or lost. What that person cannot settle alone is anything touching the total, the contract or the launch date. Those go to whoever owns the commercial side and come back in writing, usually inside a day.
The account manager model deserves a fair hearing, because it is not a trick. A dedicated coordinator gives you faster replies, somebody whose entire job is your project, and a person who chases internally so you do not have to. On a program with six workstreams and forty people that coordination is necessary, and a firm of that size without it would serve you worse. The cost is translation. Every question gets relayed out and every answer back, and reasoning is what degrades in transit. You receive the conclusion and lose the argument behind it, which is the part you were paying for.
Ours is the other shape and it carries its own cost. Somebody producing the work is sometimes unreachable for two hours. If what you want is a person whose whole job is being available, we are the wrong purchase and a larger firm is the right one.
The mirror question gets asked far less often and matters as much. Who is our daily contact on your side, and can they decide anything. An operator running several locations, common among hotels and destination brands, often has three regional marketing leads and no chair, which costs more elapsed time than anything on our side.
Three things worth putting to any agency before you sign.
On product and interface engagements the first question decides most of it, since whoever ran the usability sessions is the only person who can explain why a screen is shaped as it is.
Almost never the design or the build, which are the predictable parts, and almost always the waiting stacked between them. Production time can be estimated within a few days. Elapsed calendar cannot, because most of it goes on decisions that have nothing to do with craft.
Ranked by the time each one genuinely costs, heaviest first.
Three habits remove most of this, and none of them costs money.

Two practices bought as one engagement do land at a lower total than the same two bought a year apart, and that stops being a saving the moment the second practice is something your company has no room to absorb. Both halves of that are true, and the second half is the one agencies leave out of the pitch.
Where the efficiency genuinely comes from, since it is structural instead of a discount.
No bundle percentage gets published here, because there is not one to publish. The lower figure comes out of a smaller scope, and it appears line by line in the proposal, so you can see which items disappeared and why.
When buying together is the more expensive decision. Three situations, and any one of them is enough on its own.
A bundle should never be the reason a component enters a scope. If a line would not survive being priced and defended on its own, it does not belong in the total.
Copy is a priced deliverable here with a writer's name against it, and it decides more of the commercial outcome than the layout does. What gets written and what gets declined both need stating, because the word copywriting covers several unrelated trades.
What is written as standard.
What is declined, and what to buy instead.
Now the uncomfortable part, and it is the most common way a good project quietly underperforms. A site is designed to carry a hundred and eighty words in a section. Eighty arrive the night before launch, written by whoever had capacity that week. Nothing about the design is wrong and the pages simply say less than they were built to say, and six months later somebody concludes the redesign did not work.
So the copy call belongs in the first fortnight instead of the last. Who writes each page, against which date, reviewed by whom, and what happens if that date slips. Where the budget cannot carry professional writing across forty pages, fifteen pages written properly is the better purchase, and we would sooner build the smaller site than fill the larger one with filler.
We art direct photography and video and we commission the crew, which is a different business from running a production house and is worth separating before anybody quotes you. Art direction decides whether the images look like your company. Operating the camera is a trade of its own.
What art direction actually means on a project.
The crew is commissioned. Photographer, videographer, stylist, producer, motion designer and whoever else the job calls for, working to our direction and itemized in the proposal so you can see who does what and at what cost. In the Los Angeles market we serve that bench is unusually deep, which is one of the few genuine advantages of shooting there.
The part clients least want to hear is about money. Imagery is regularly the largest line nobody budgeted. A brand refresh gets costed carefully and the photography meant to fill it is treated as something that will turn up. It does not turn up. Stock gets substituted, the launch looks like the category average, and the identity underneath takes the blame.
A strong system survives an average typeface and a slightly wrong blue. It does not survive tired photographs, because pictures are what a customer reads as evidence about the business. That is why imagery direction sits inside the system alongside the applied design work it appears in, and why venues and touring productions budget a shoot before they budget a campaign.
Where we stop is scale. A directed commercial with a crew of thirty and a broadcast deliverable belongs with a production company, and we will help brief one.
Frequently, and the arrangement lives or dies on a boundary written down before anybody starts, because the failure mode here is never hostility. It is two capable groups each assuming the other owned something.
Three shapes cover almost every version of this.
We lead and your people execute. We settle the strategy, build the identity and the site, and hand over a system your team runs day to day. Correct where the internal group is strong on production and thin on the decisions above it.
Your team leads and we fill one gap. You own the plan and we supply a discipline you do not employ, meaning search, interface design, or design capacity through a heavy quarter. Here we work to your cadence inside your tools, and we make no attempt to reorganize a process that already functions.
We build the system and train the people who will run it. The deliverable is a working setup plus the teaching. Templates, components, documented rules, a governance model and recorded sessions. Institutions ask for this most often, since a communications office inside a school or university usually holds real capability and no framework to hold it together.
What has to be written down, and it takes about an hour.
The failure mode in practice is dull and expensive. Nobody writes the boundary, two months pass, and a page exists twice, a campaign goes out carrying the old mark, or a component gets rebuilt by whichever side reached it first. Nobody behaved badly and the money is gone anyway.
One political point that gets left unsaid. An agency making an internal team look unnecessary is resisted quietly until the engagement fails, so the work is scoped to make your people faster and more visible inside their own company. The templates they build from do more for that than any strategy document. A fair test at the twelve-month mark is how much of the output your own people produced without asking us anything.
Regularly, and the whole arrangement turns on one question, which is whether a single party owns the standard everybody else builds against. Where somebody owns it, three suppliers produce one coherent company. Where nobody does, they produce three versions of it inside a year and no single decision was to blame.
Most companies past a certain size already hold a public relations firm, a media buyer, a search supplier or a development partner, and replacing all of them is seldom the right recommendation.
How drift actually happens, because it is never dramatic.
What prevents all of it is unglamorous. One documented standard, versioned and dated, somewhere every supplier can reach unaided. A named owner of that document. Change requests routed to the owner instead of settled locally. And the mark with its approved lockups and clear space distributed as files, since a rule nobody can find gets replaced by a guess.
Who arbitrates. Your side does, and it should be one named person with the authority to close a discussion. The standard owner writes the recommendation with its reasoning, the other suppliers argue it once, and your person decides. An agency arbitrating between suppliers it competes with is a conflict of interest wearing a process.
We work either way. When we hold the standard, the other suppliers get the files, the reasoning and somebody to call. When another firm holds it, we build to their system and do not reopen decisions we would have made differently, with two exceptions we always raise, meaning accessibility failures and anything creating legal exposure. Handing a finished site to an incumbent development partner is a normal ending here instead of a lost sale.
The honest cost is that coordination is real work. Briefing three suppliers, reconciling calendars and reviewing their output takes hours somebody funds, so it gets its own line in the scope instead of being absorbed quietly. Against a release and touring calendar where a promoter, a ticketing platform and a publicist all publish, that line is not small.
Seven things get declined here consistently, and each one has a category of firm that does it properly. A capability list with no edges is a sales document, so this is the other half.
One point about the referrals. Nothing is paid in either direction for pointing you somewhere, so the name you get is the one we would use on our own work. Where we have no good name, you hear that instead of a placeholder.
The reason to publish this is that the alternative reads worse. An agency saying yes to all seven is either subcontracting quietly at a markup or learning the discipline on your budget, and you find that out in month four.

Nothing about the outcome and a great deal about the work, because no agency controls the auction, the algorithm, your competitors or the people who answer your phone. Anyone guaranteeing you a position, a conversion rate or a revenue figure is either saying something they know to be false or has picked a target so undemanding that reaching it proves nothing.
Read a guarantee closely and it usually dissolves. A promised first-page ranking attaches to a phrase with no buyer behind it. A guaranteed lead volume counts submissions of any quality, including the ones your sales team deletes on sight. A promised quantity of placements in link building counts pages that exist to sell placements and that no human reads. Each of those promises gets kept. What was promised was worthless, which is exactly why it could be promised.
What does get committed here, all of it written down before anything starts.
Measured only where a genuine before and after both exist, the average lift in leads comes out at 250%, a pooled figure from past programs of very unlike sizes and very unlike starting points, and it forecasts nothing whatsoever about yours. Movement worth showing a board in the organic side of the practice still takes four months to a year, and no commercial arrangement compresses that.
The fair test of any agency on this point is simple. Ask what they will refuse to guarantee and why. Vagueness there is the answer.
Long enough for real traffic to find what testing did not, with the exact window written into your agreement, and the line that matters far more is the one between a defect and a change. Almost every unpleasant conversation about post-launch work starts there, and almost nobody puts that line on paper before it is needed.
A defect. Something we built that does not do what the approved specification said it does. A form that appears to submit and delivers nothing. A template breaking on a common handset. A redirect that was mapped and never fired. A filter returning the wrong products on a Shopify build. Those get repaired at no charge, and we want to hear about them whether or not a window is still open.
A change. Something working exactly as specified that you have since decided you want different. A new page type. An extra field on a form. A layout you now prefer to the one you approved. That is chargeable, and it should be, because the alternative is a supplier pricing every project on the assumption of unlimited free revision after delivery, which every client funds and only some clients use.
Written down the distinction looks obvious, and it collapses under pressure, since the honest version of most disputes is that the specification was ambiguous and each side read it the way that suited them. The protection is a specification precise enough to be wrong against, agreed before build, which is why scopes here name states, error behaviour and edge cases instead of page counts alone.
What gets watched in the opening weeks, because that is when genuine faults appear.
One reason no single figure is published on this page. A WooCommerce store carrying its own server, its own updates and its own share of card-data compliance deserves a longer watch than a five-page site on a hosted platform, and one number covering both would be tidy and untrue. Ask any agency to define a defect in writing, and ask who decides when the two of you disagree.
Six things, every one of them unglamorous, and together they are the difference between a site that was excellent in March and one that is mediocre by December. Decay here is never dramatic. It arrives a tenth of a second and one untagged image at a time, which is why nobody notices until a year has passed.
Who honestly does not need one. A small stable site on a fully hosted platform, with content changing a few times a quarter and somebody internal comfortable in the editor. The vendor patches itself, no plugins are drifting, and a monthly fee for a mailbox nobody writes to is a poor purchase. That business needs a name against the site, a reminder to look properly each quarter, and a number to call when something breaks.
Where the calculation flips is volume and consequence. For consumer brands selling direct, an hour of broken checkout on a Friday evening costs more than a year of support, and that arithmetic is the entire argument. Everything else is preference, and working out which of the two describes you takes one conversation.
You do, on an account opened in your company's own name at the start of the project and billed to you directly by the provider, so an agency relationship ending is never a threat to a live site. We work inside your account under access you grant and can withdraw in a minute.
Concede the other arrangement first, because it has a real case. A business with nobody technical genuinely prefers one invoice and one number to ring, and most firms who resell hosting do it to be helpful. The problem is structural instead of moral. The account carries their name, the billing relationship is theirs, the domain renewal notice reaches their inbox, and your live site sits with a company you may one day be arguing with. Nobody sets out to build that hostage situation. It assembles itself the moment a relationship sours, or the agency is acquired, or the one person who understood the setup resigns.
What to establish before agreeing to any hosting arrangement.
The practical difference between the two shapes of hosting is who carries the standing obligations.
Fully hosted. Servers, patching, availability, certificates and most of the card-data problem sit with the vendor, which is what a subscription to something like a fully hosted store platform actually buys. Control over parts of the stack is the price, and for most businesses that trade is obviously correct.
Self-hosted. An open-source store you run yourself, or a comparable WordPress build, hands you total control together with the server, the updates, the backups, the firewall and the uptime. That is an operating job and never a free upgrade, and where nobody owns it the stack drifts silently out of date until an incident announces it.
Access hygiene deserves the same attention as ownership, which is why larger organizations provision access through named accounts inside their own identity system. Any small company can copy that cheaply, and shared logins are the first habit to drop.
Signing one is routine here, and a mutual form is the version worth insisting on, because by the second conversation confidential material is travelling in both directions. Where you want the paperwork settled before a first substantive discussion, that is the sensible order and it is what we would ask for with the positions reversed.
Why mutual instead of one-way. A one-way document quietly asserts that only your information is sensitive, which stops being true the moment anybody describes a method, a fee structure or another client's situation. Making it reciprocal costs a paragraph and removes the awkwardness of raising it later.
On speed. Send yours, or ask us for a form, and expect a turnaround measured in a day or two. Where it stretches into weeks, that is usually two sets of counsel negotiating a document neither party will open again, and the useful move is starting from whichever version your own organization has already approved. On engagements that go through legal review the confidentiality terms tend to arrive inside a larger agreement anyway, which makes the separate document a formality.
What it means day to day, signature or no signature. Work in progress stays invisible, meaning no process shots, no unreleased screens inside a pitch, and no naming you as a client until you have said it publicly. Access is individual and narrow, with the minimum permission that lets somebody work and removal when the engagement closes. Commissioned specialists are bound before they see anything, which is our arrangement to make and not yours to chase. And where the existence of the conversation is itself the sensitive item, it gets handled that way from the first email.
The portfolio, which is the commercial half of this and usually goes unsaid. Publishing work is worth real money to a studio and it is a large part of why an agency pushes a project past the brief. So a client who will never permit the work to be shown should say so in the first meeting. It changes nothing about how the work gets done and it removes the uncomfortable request that otherwise lands eighteen months later. For property and development launches and unannounced consumer products, an embargo with a date attached is usually the right middle position, meaning nothing published until you go public and a proper write-up afterward.
One boundary that does not move. We do not draft your agreement and we do not advise you on its terms. Reading what we can accept and marking up what we cannot is our part of it, and the opinion on whether it protects you is your own lawyer's to give however tight a timeline has become.
Then you leave, and the arrangement is built so that costs you a handover session instead of a recovery project. Companies change agencies for reasons with nothing to do with the work, meaning a budget cut, a new marketing director arriving with their own relationships, or a real change in what the business needs. Treating that as a betrayal is how firms end up sitting on a domain and earning the reputation that follows.
What transfers, and most of it never has to move because it was yours from the first week.
The handover itself is a working session plus a written document. Dropping a folder into a shared drive is a transfer of the problem and not a handover. There is a call with whoever picks the work up, the incoming agency included where one exists, because refusing that helps nobody and costs us an hour. A file transfer takes minutes and the thinking behind those files does not travel on its own.
So put one question to every agency you are considering, this one included, and make them answer it in specifics. What happens on the last day. Which accounts do you hold. How long does a transfer take and who performs it. What is documented and what lives only in somebody's memory. What stops working the moment your access is revoked. Larger organizations settle this by agreeing a documented offboarding path before the first credential is issued, and nothing prevents a smaller company asking for the same in a single paragraph.