Hiring
How to vet an ecommerce agency or operations partner
Every provider deck shows logos, results and enthusiasm — none of it predicts your experience. The vetting method that does: failure-mode questions, artifact requests, reference calls that ask the right thing, and the contract terms that reveal confidence. Usable on us.
Choosing an ecommerce partner is a decision most companies make on the least predictive evidence available: the sales experience. Decks show logos and case studies; calls show enthusiasm and fluency; proposals show ambition. All of it is the provider at their best, and none of it predicts the only thing you are actually buying — how they behave in week thirty, when a listing is suppressed, a specialist quits, or a number disappoints. This piece is the vetting method that targets week thirty directly.
Declared interest: we are a vettable party — an operations partner with published pricing — and everything below is usable on us in a first call. That is rather the point: a vetting method a provider would not welcome is a vetting method telling you something.
Key takeaways
- —Vet for failure handling, not success stories: how a provider behaves when work breaks is the product you are buying — and the questions that reveal it are askable in an hour
- —Demand artifacts over adjectives: a real (redacted) status report, an SOP sample, an onboarding plan — what they show you is what exists
- —Reference calls work when you ask about the worst month, not the relationship
- —Contract terms are confidence made legible: notice periods, exit terms and IP ownership tell you how the provider expects to be judged
- —Match the provider type to the work first — vetting a strategy agency for operations grind fails both of you
- —Weight team truth over sales polish: who exactly works your account, and what happens when that person leaves
The failure-mode questions
- —'Walk me through the last time a client's account had a serious incident — what happened, hour by hour?' Grade for specificity and ownership; polished vagueness is a rehearsed non-answer
- —'Who reviews the work before I see it, against what standard?' A named layer and checkable standard, or a promise wearing a process costume
- —'What happens the week my assigned specialist leaves?' The answer should include a named backup and documented SOPs — continuity is engineered or absent
- —'Tell me about a client you lost and why.' Providers who have never lost one are new or lying; the honest story tells you how they fail
- —'What would make you turn down this engagement?' Real providers have a fit test; universal enthusiasm is a sales quota talking
- —'When a metric disappoints, how do I find out?' You want proactive disclosure with a diagnosis, not dashboards you must interrogate
The artifacts to request
Talk describes; artifacts prove. Ask for a redacted weekly or monthly status report from a real client — its honesty gradient (does it surface problems or only wins?) predicts yours. Ask for an SOP sample: providers who run on documentation can show it in minutes; providers who improvise will offer to 'put something together'. Ask for the onboarding plan they would run on your account — week by week, with what they need from you. And for operations work specifically, ask how they would audit your current catalog before touching it: the provider who wants to look before quoting is pricing your reality; the one who quotes instantly is pricing a template.
Reference calls that actually work
References are hand-picked, so the standard 'are you happy?' call returns the expected yes. The questions that get past curation: 'What was the worst month of the relationship, and how did they handle it?' 'What do you wish you had known before signing?' 'What do they need from you to perform — and how much of your time does that take?' 'If they raised prices 20% tomorrow, would you stay?' The last one is the sharpest satisfaction instrument available. Also worth doing: one un-referenced check — the industry is small, and a peer who used the provider off the reference list is worth three curated calls.
Contracts, pricing and the red-flag list
- —Notice periods signal confidence: monthly-rolling terms after an initial period mean the provider expects to be judged on the work; long lock-ins mean they expect to be judged on the contract
- —Exit terms in writing: credentials returned, documentation handed over, transition support — the offboarding you negotiate while friendly is the one you get
- —IP and access ownership: your accounts, your data, your SOPs — a provider who holds credentials hostage structurally has told you the ending in advance
- —Pricing legibility: line items you understand beat blended mystery retainers; published pricing beats quoted-per-prospect
- —Red flags in any tier: guaranteed rankings or revenue, logo walls without contactable references, instant quotes without discovery, enthusiasm unaccompanied by a single clarifying question about your catalog
Interview for week thirty: ask how they fail, demand artifacts over adjectives, call references about the worst month, and read the contract as a confidence signal — any provider worth hiring will pass gladly, and the ones who bristle have answered your question.
Frequently asked questions
The failure-mode set: walk me through your last serious incident hour by hour; who reviews work before I see it; what happens when my specialist leaves; tell me about a client you lost; what would make you decline this engagement; how do I learn about disappointing numbers? Specific, owned answers predict the relationship — polish without specifics predicts the sales cycle.
Ask past the curation: the worst month and how it was handled, what they wish they had known before signing, how much of their own time the provider needs, and whether they would stay through a 20% price rise. Add one un-referenced check from your own network if you can — the uncurated data point outweighs three arranged calls.
Long lock-ins without performance outs, vague or absent exit terms, provider-held credentials and accounts, blended retainers you cannot decompose, and guarantees of rankings or revenue nobody can honestly make. Strong signals run the other way: monthly rolling or a short explicit initial term, written offboarding, your ownership of accounts and documentation, published pricing.
The deliverable is continuity, so the vetting weights shift: review layers, named backups, SOP depth and credential handling matter more than creative portfolios; the provider-type match matters first (operations grind is most agencies' weakest layer); and the artifact test is a catalog audit approach rather than a campaign deck.