Let me tell you something no agency will ever say out loud, because it would be commercial suicide: the pitch is the single worst predictor of what an agency will be like to work with.
I can say this because I've been on both sides. I run a consultancy. We pitch for work. We prepare carefully. We put our best people in the room. We curate the case studies most relevant to the brief. We rehearse. And we're not unusual in any of that - every agency worth its salt does exactly the same thing. That's not dishonesty. It's professionalism.
But if you're a managing partner or COO selecting a digital partner on the strength of what you saw in the pitch, you're making a decision based on the most unrepresentative sample of that agency's behaviour available to you. The pitch is the highlight reel. The first date outfit. The house staged for the Rightmove photos. And just like those photos, everything in it is technically true and simultaneously nothing like what daily life actually looks like.
We've done a thorough pitch process. We've seen the portfolio, met the team, and checked the references. We know what we're buying.
I hear you. And I'm not suggesting you've been naive. But I've watched enough agency relationships go sideways - including ones we've inherited when they've broken down - to know that the firms who end up in trouble almost always say the same thing afterwards: "They were so impressive in the pitch."
The pitch environment is designed by the agency, managed by the agency, and measured by criteria the agency has influenced. Think about that for a second.
The portfolio has been curated for relevance to your brief. The principals presenting are the most articulate people in the business - that's why they're principals. The case studies are the best outcomes the agency has ever produced. The timeline they've proposed has been calibrated to sound achievable without triggering your scepticism. And the references they'll provide, if you ask, are from clients who had the best experience.
None of this is dishonest. I want to be really clear about that. But it means the pitch environment produces a systematic bias. You're seeing the agency's best possible version of itself - prepared, rehearsed, and optimised for the specific concerns they've detected in your brief.
And here's the bit that nobody mentions: you're partly responsible for this dynamic. Agencies pitch the way they do because clients evaluate them the way clients evaluate them. You invited three agencies, gave them two weeks and a brief, and asked them to impress you in 90 minutes. Of course they put on a show. What else were they going to do?
So the question isn't whether the pitch was impressive. The question is: what signals actually predict whether this agency will be good to work with on a Tuesday afternoon in month four, when something's gone wrong and nobody's watching?
I've been thinking about this for years, and I keep coming back to three things. Not pitch things. Working relationship things.
How they behave between meetings. This is the big one. The agency that proactively communicates between scheduled touchpoints - flagging concerns early, bringing solutions alongside problems, treating your time with respect - is demonstrating delivery behaviour that the pitch environment never tests. I worked with a 200-person consulting firm a couple of years back that had been through two failed agency relationships in three years. Both times, they'd picked the agency that looked best in the room. When I asked what went wrong, the answer was almost identical both times: "They were great in the fortnightly catch-ups, but between meetings we had no idea what was happening." The catch-up becomes a mini-pitch. A performance every two weeks. What you actually need is an agency that communicates like a colleague, not a supplier.
How they handle the moment when something goes wrong. Every project has problems. Every single one. I've seen two patterns. The first: the agency communicates the problem clearly and promptly, explains the options, and lets you decide. The second: the agency manages the problem silently, presents a completed solution, and frames it as proactive problem-solving. That second one sounds better, doesn't it? But think about what it really means. The agency made decisions on your behalf without telling you there was a decision to make. They prioritised managing the relationship over managing the work. And you've lost visibility into what's actually happening on your project. That should worry you.
Whether the senior people who pitched are the ones who deliver. This is the classic. I've lost count of how many times a managing partner has told me: "The people in the pitch were fantastic. Then we never saw them again." The agency whose principals are present throughout the engagement is fundamentally different from the agency whose principals win the work and hand it to a more junior team. I'm not saying the junior team is necessarily bad - sometimes they're excellent. But you evaluated and selected based on the senior people. If those people aren't in the work, you haven't bought what you think you've bought.
One thing I'd add here, based on patterns I've seen across engagements: this problem is worse in certain project types. CRM migrations and portal rebuilds are the ones where it bites hardest, because the complexity is front-loaded and the senior people are most needed in the first six weeks - which is exactly when they're busiest pitching the next client.
So what should you actually be looking for? Four signals I've learned to trust - partly from running pitches ourselves, and partly from watching clients navigate bad ones.
They push back on at least one element of your brief. An agency that agrees to everything either hasn't read the brief carefully enough to find something to disagree with, or has decided that winning the work matters more than doing it well. The agency that asks a clarifying question, challenges an assumption, or says "honestly, we wouldn't do it that way and here's why" - that's the one demonstrating the analytical engagement that good delivery requires. I remember sitting in a pitch once where the prospective client looked genuinely taken aback when we told them their timeline was unrealistic. We didn't win that pitch, actually. But the agency that did delivered late. I'm not saying that to be smug - well, maybe a little bit - but the point stands.
They ask hard questions about your internal readiness. An agency that asks about your data quality, your internal bandwidth, your governance capacity, and your stakeholder alignment before committing to a timeline is managing the real project risks. The one that nods along and says "yes, we can do that in twelve weeks" without asking any of those questions is managing the pitch risks. For a mid-market professional services firm doing a website rebuild or a client portal migration, the internal readiness questions are often more important than the technical ones - because the project will stall on your side before it stalls on theirs.
The delivery team is named and introduced. A small agency might not have the capacity to bring the full delivery team to a pitch meeting, and that's fine. But the people who will actually work on your project should be named in the proposal, and ideally you should have some sense they've been briefed on your opportunity before the presentation. If the proposal says "a senior developer" rather than "Sarah, who's been with us for six years and led our last two financial services portal projects," that's telling you something.
They bring problems as well as solutions. The agency that says "here's a challenge we anticipate in your context, and here's how we'd approach it" is being honest with you. The one that only shows successful outcomes and glossy case studies is still in pitch mode. Real work has rough edges. An agency that acknowledges that upfront is one that's likely to acknowledge it during the engagement too.
They agreed to everything without pushback. I covered this above, but it's worth repeating because it's so common and so consistently ignored. If three agencies push back on your timeline and one doesn't, the one that didn't isn't braver or more capable. They just want the work more than they want to do it well.
The timeline is materially shorter than everyone else's. If four agencies estimate 18-22 weeks and one says 12, don't celebrate. Ask yourself: are they misunderstanding the scope, or are they planning to deliver something different from what you think you're buying? Discovery is often the phase that gets compressed or eliminated when an agency is trying to look faster than competitors. And cutting discovery is like skipping the survey before buying a house. It feels like you're saving money right up until the moment you're not.
The price is significantly below the competitive range. A price that's 30% below the next lowest bid is not a bargain. Full stop. It's a signal that the scope has been misunderstood, that corners will be cut to protect margin, or - and this is the one that really catches people out - that the pricing will be revised once the relationship has been established and walking away is expensive. I've seen this play out with a 150-person accountancy firm that went with the cheapest of four bids on a CRM integration. The initial quote was £85k. The final invoice was £210k, spread across a series of change requests that started arriving in week three.
The proposal doesn't name specific people. If the proposal refers to roles rather than individuals - "a project manager," "a UX designer," "a technical lead" - you're buying a capability, not a team. The agency retains the right to staff your project with whoever's available when the work starts, not whoever was implied during the pitch. Ask directly. Get names in the contract.
Two techniques that I've seen dramatically improve the quality of agency selection decisions.
Ask for delivery references, not client references. The agency's client references are curated - of course they are. They're the equivalent of the pitch portfolio. Instead, ask to speak with someone who worked on the delivery side of an engagement that had difficulties. Not a disaster, necessarily - just one where things didn't go smoothly and the agency had to work through problems. A good agency can provide this. They'll have relationships with clients who'll say "yeah, it was bumpy, but they handled it well." An agency that can't provide a reference like that either hasn't done enough work to have encountered problems - unlikely - or doesn't have relationships strong enough to survive them. That's telling.
Request a paid discovery phase before committing to the full engagement. This is the single best risk-reduction strategy available to you. A structured discovery - typically two to four weeks - lets you experience the agency's actual working behaviour before you've committed six figures. You'll see how they communicate, how they handle ambiguity, how senior the people are who do the work, and whether they genuinely understand your context or are still in pitch mode. An agency that welcomes this is confident in their delivery. An agency that resists it is managing the risk that a proper discovery will reveal a scope larger than their proposal assumed.
I'm not trying to make you paranoid about agency selection. Most agencies are run by decent people who genuinely want to do good work. The problem isn't bad intent - it's that the pitch process, as most firms run it, is structurally designed to surface presentation skill rather than delivery quality. And those are different things.
The firms I've seen get this right treat the pitch as the beginning of the evaluation, not the end of it. They use it to create a shortlist, then they test before they commit. They ask awkward questions. They pay attention to how the agency responds when things aren't scripted. They look for the signals that predict what month four will feel like, not what the first meeting felt like.
If you want a structured way to apply this during a live pitch process, we've put together a partner evaluation scorecard that covers the working-relationship signals, the green flags, the red flags, and the pre-commitment tests from this piece - with a scoring methodology and a clear go/no-go output. [Download it here.] There's a companion scorecard for evaluating AI-specific vendors too, if that's relevant - the principles overlap more than you'd think.
And if you're reading this because you're already in an agency relationship that isn't working - I've written separately about what to do when that happens. Because knowing the red flags in advance is useful, but sometimes you're already past that point, and that's a different conversation entirely.