A consultant's job is partly about saying yes well. It's also — and I think more importantly — about saying no without burning the relationship. The yes work is obvious: build the thing, deliver it, get paid. The no work is quieter and harder to see, but it's what determines whether the next yes is from a merchant who trusts you or one who's looking for a fight.
Four categories of project show up regularly that I refuse to quote. Each one is a real loss in the short term. Each one earns better work in the long term. I'll walk through them with the trade-offs intact, because saying no isn't free — it has its own cost — and the only honest way to describe the discipline is to describe the cost too.
1. The wrong tool for the job.
Roughly once a quarter, a prospect lands in my inbox describing what sounds like an eCommerce project — until the second or third question reveals it isn't. The product catalog is actually a service catalog. The "checkout" is actually a booking flow. The "inventory" is actually scheduled-capacity allocation. Shopify can technically be bent into shape for some of these, but the bend never gets undone, and the merchant ends up paying twice — once for the bend, again when they realize they wanted a different platform.
The honest answer in this category is short: "I'm not the right person for this. You need an operator-focused platform, not a developer optimizing for eCommerce. Here's where I'd look." It costs me the engagement. It saves the merchant a misallocation that could run into five or six figures by the time they course-correct. I'd rather lose the project than be the consultant in the room when the misallocation becomes obvious.
2. Briefs that haven't earned themselves.
The most common version of this: "Make our store look like Competitor X." The brief points at another brand's redesigned site, asks for a near-copy, and treats the design work as the actual deliverable. The thing that's missing is any evidence that Competitor X's redesign moved their numbers. Often it didn't. Often the merchant has been looking at the new site for a month and has confused "feels new" with "works."
The diagnostic-before-estimate principle applies at the brief level too. If the brief hasn't earned itself — if it's reactive, copycat, or untested — the right work is to push back on the brief, not to quote work for it.
The reframe I offer is the diagnostic conversation: "Before I quote the redesign, can we talk about what your data is telling you about why customers leave?" Half the time, that conversation surfaces a tracking issue, a conversion-funnel leak, or a merchandising decision that explains the perceived underperformance better than any redesign would. The redesign quote becomes a smaller, more targeted scope — or it doesn't happen at all. Either outcome is better than building the wrong thing well.
3. Six-week migrations with thirty-stakeholder approval lists.
Enterprise-style projects with hard deadlines and committee sign-off aren't a fit for a solo practice. The work itself isn't the problem — I can do the work. The friction is everything around the work: the weekly status decks, the contract appendices, the stakeholder reviews that translate developer time into political time, the dependency on other agencies whose timelines I don't control.
The honest answer here isn't "I can't." It's "this needs a delivery team, and I'm not one." The merchants who try to fit committee-shaped engagements onto a solo developer usually end up paying twice — once for the solo work, again for the agency they bring in to absorb the committee overhead. I'd rather refer the project at the start than be the person caught in that overhead at the end. I can be a useful second pair of eyes during planning, or a technical advisor inside an agency-led engagement. I shouldn't be the delivery layer.
4. Projects that compound badly.
This is the hardest category to refuse, because the work usually is doable, the merchant usually wants it, and the money is real. The problem is what the project does to the store over the next two years. A quick-fix that adds another layer of conditional Liquid to a section already carrying five conditionals. A custom feature shipped without a maintenance plan, by a contractor who won't be around when it breaks. An app stack where the proposed solution to last quarter's app instability is another app on top.
The honest answer here takes longer, because it has to acknowledge the real problem the merchant is trying to solve, then explain why the proposed shape of the work makes the problem worse on a longer horizon. "What you're asking for would work for three months. Here's what would still be working in two years." If the merchant's timeline can't absorb the longer conversation, I'd rather pass than ship debt I'd be embarrassed to maintain.
What refusing actually buys
The economic logic is straightforward: refusing a $5K bad-fit project preserves capacity for a $20K good-fit project that hasn't arrived yet. The capacity I'd burn on the wrong scope is the capacity I need to be ready for the right one. Three to four bad-fit refusals per year, on a calendar that holds two or three Plus brands at a time, is the difference between a year of reactive work and a year of compounding relationships.
The relational logic is harder to quantify but more durable. Merchants refer the consultants who refused to take their bad-fit work. Robert (anonymized — a prospect from earlier this year whose use case turned out to be operational rather than eCommerce) thanked me for the honest no and asked who he should talk to instead. I sent him to a platform vendor who fit the shape of his problem. Six weeks later he sent me his cousin's underwear brand. That's the second-order economics of a clean no.
The values logic is the simplest of the three. Knowing what's outside your lane is a humbler position than charging for whatever lands in your inbox. I've shipped enough Shopify work over six years to know what I do well and what I shouldn't be doing. Refusing the work I shouldn't be doing is how I get to keep being good at the work I should.
If you're a Shopify operator with a project you suspect might be in one of these four categories, the cheapest version of finding out is a 30-minute call. The honest answer might be "yes, we should work on this together." It might also be "you're right to be unsure — and here's where I'd look instead." Either answer beats the alternative of paying for the wrong project well.