What we look for, how to prepare, and how we score it.

A channel recommendation should trace back to the client's economics. A good strategy document opens with the target. Pipeline, revenue, CAC ceiling, payback period. And works down to channel and budget allocation, with a reason each channel fits that business, that margin, and that sales cycle. An agency that jumps straight to "we'll run Google and Meta" without connecting it to the client's numbers is executing tactics, not running strategy.

Paid media: account structure, budget setting and reallocation, creative and audience testing, and the difference between platform-reported ROAS and blended MER. SEO: technical audit ability, content production system, and how links are actually acquired (we can tell manual outreach from a PBN), plus a current view on AI Overviews and LLM-driven answers intercepting clicks. Content, social, email/lifecycle, CRO, creative: depth in what you claim, shown through real work rather than a capabilities slide.

Tracking configured correctly (GA4 and GTM, conversions firing cleanly, ideally server-side tagging), and numbers that reconcile to a source of truth like the client's CRM or Shopify, not just the ad platform. A 4-level answer: a test readout that states the hypothesis, the control, the spend or sample threshold set before starting, and why you called it when you did.

"Reduced blended CAC 28% while holding lead volume flat over six months" is a business result. "Generated 3.2M impressions" is activity. Lead with numbers a CFO would recognize: CAC, CPL, ROAS or MER, pipeline created, revenue influenced, retention. Impressions, reach, and follower counts are diagnostics, not outcomes.

The fastest way to a low score is over-attribution. If revenue grew 40% during your engagement but the client also launched a product, ran a Q4 promo, and had normal seasonality, claiming the full 40% is a red flag. And the reference will usually say so. A strong agency isolates its contribution, ideally with something like a geo holdout or a conversion lift study, and at minimum with an honest statement of what else was in play.

Client tenure is the hardest number to fake. We look at average length and logo churn. A roster of relationships past 18 to 24 months tells us more than a single spectacular three-month case study. A years-long client is often your strongest single piece of evidence.

Did the numbers in the case study match what you saw on your side? When something underperformed, did they tell you first, or did you find out on your own? Did they stay in scope and budget? Have you referred them to anyone, and would you hire them again?

Every agency has a channel that flopped or a quarter that came in under target. Variance is normal. We grade what you did next. Whether you raised it early, diagnosed it honestly, and changed course, or whether you went quiet and hoped.

The most common complaint about agencies is silence between invoices. We want a defined cadence, and reports that explain what you did, what it means for the client's business, and what you're doing next. A stack of platform screenshots with no interpretation scores lower than a shorter report that says what the numbers mean in revenue terms.

Strong agencies flag a dropping ROAS or a stalling campaign in week two, before the client notices. We ask references specifically about this. Surfacing problems early is one of the clearest markers of an agency worth trusting.

Promising fast SEO or content results sets up a trust failure you can't recover from. Naming realistic timelines up front, and being specific about when the client should expect to see movement, reads as experience.

If you're offshore, show us how you handle it: defined overlap hours, a response commitment (for example, a reply within one business day and same-day acknowledgment during overlap), a named point of contact, and a clear escalation path. Done well, timezone difference stops being a factor. Left vague, it's the first thing that goes wrong.

The best account managers speak the client's business back to them. A CMO wants to hear the revenue impact, not the click-through rate.

You should work inside the client's own accounts through granted or delegated access. Their Google Ads, Meta Business Manager, GA4, CRM. Not by collecting the client's passwords and not by running everything through accounts you own, which locks the client in and concentrates risk. Roles should follow least privilege, so a junior media buyer doesn't hold domain access they'll never use.

When an engagement ends or a team member leaves, access should be revoked promptly and verifiably. "We probably still have logins to a few former clients" is a failing answer. And a more common one than agencies admit.

If you touch media budgets, we look at how you handle client spend. Budgets, any fee or markup, and reconciliation against platform invoices should all be transparent. Buying media at one price and billing the client higher without disclosure is the single fastest way to lose a client's trust. And it's a fail here. Transparent spend handling, where the client can see exactly what went to the platform and what went to you, is a selling point.

Customer lists, email databases, and CRM exports carry real obligations under GDPR and CCPA. If you're processing a US or EU client's customer data from another country, you need to state plainly where that data lives, who has access, and which subprocessors touch it. "We just download the list" is not an answer that passes.

An approval workflow before anything publishes under the client's name prevents the off-brand post or the policy violation that gets an ad account suspended. Your process should survive a person leaving: a project management system of record (Asana, ClickUp, Monday, or similar), documented accounts, and a continuity plan, so the client relationship doesn't live entirely in one person's head.