Referrals are the dominant channel for most agencies, and treating them as luck rather than as a system is the biggest single miss. Beyond that, the channels that work are narrow and slow: content aimed at one specific buyer, partnerships with complementary agencies, expanding existing accounts, and targeted outreach that references something real. What almost never works for agencies is broad advertising and generic cold email, because the purchase is high-trust and low-frequency. Pick two channels, work them for twelve months, and measure which produced revenue rather than which produced enquiries.
Most agencies get clients through referrals and word of mouth, do nothing deliberate about it, and describe business development as their weakest area.
Both halves of that are true and connected. When the pipeline arrives without effort, no muscle develops - so when referrals thin, there is nothing to fall back on and the response is usually a scramble through every channel at once for six weeks.
This guide covers the channels that actually work for agencies, the ones that mostly do not and why, how to choose two and stick with them, what to do when the pipeline is empty right now, and how to measure any of it.
The honest starting point
Most agencies reading this have one working channel - referrals - which arrived without effort and is therefore invisible as a system.
The useful first move is not adding channels. It is recognising that the channel you already have can be managed, measured and roughly doubled with a handful of well-timed asks. Everything else in this guide is what you build alongside it, not instead of it.
Why agency sales is unusual
Three properties shape everything.
High trust, low frequency. A client might buy an agency relationship three times in their career. There is no learning-by-repetition, so they buy on signals - reputation, referral, proof - rather than on evaluation.
Considered and slow. Weeks or months from first contact to signature, often with several people involved. Channels that work on impulse do not apply.
High value per client. Winning six clients a year can be a whole business. That makes narrow, slow, high-touch channels viable in a way they would not be for a volume business - you do not need reach, you need the right twenty conversations.
Those three together explain why broad advertising underperforms for agencies and why a single well-placed relationship can matter more than a year of impressions.
Two channels, twelve months
The single most useful constraint in this guide: pick two, work them properly for a year, and ignore everything else. Six channels at 15% effort produce nothing, because every channel here has a threshold below which it does not compound.
The channels that work
1. Referrals
The dominant channel for most agencies, and the one most often left to chance.
The mistake is treating referrals as a by-product of good work. Good work is necessary and not sufficient - a referral also needs a specific ask, at the right moment, with an easy way to act.
Our guide to building a referral programme covers the mechanics. The short version: ask after the testimonial, name the profile you want rather than asking generally, give them something forwardable, and close the loop afterwards so they refer again.
This is where to start, before any other channel, because the relationships already exist and the cost is a handful of well-worded emails.
2. Partnerships with complementary agencies
The most under-used high-quality channel available.
A design studio that does not build. A development shop that does not design. A PR firm asked constantly about websites. These businesses talk to your exact buyer, regularly, about needs adjacent to yours - and unlike a client, they generate referrals continuously rather than once per engagement.
Building them is slow and simple: refer work to them first, repeatedly, with nothing expected. Be specific about what you want back. Unlike client referrals, an explicit fee arrangement here is normal and clean.
A related and overlooked source: the agencies you hand over to. A clean project handoff to a competitor generates a startlingly high rate of reciprocal referral, because the incoming team forms a view of you within a day.
3. Content aimed at one specific buyer
Content works for agencies, with one condition: it must be narrow enough that the right reader recognises themselves.
Generic content - "5 tips for better websites" - attracts nobody in particular. Content that names a buyer and a problem attracts exactly the person you want and nobody else, which is the correct outcome.
The mechanism is not traffic. It is that a prospect who arrives at a sales conversation having read three things you wrote is already largely sold, and the conversation starts from a completely different place.
Content compounds and it is slow. Twelve months is a realistic horizon, which is why it pairs naturally with specialisation - narrowing what you write about is the same decision as narrowing who you serve.
4. Expanding existing accounts
Consistently the cheapest revenue available and consistently neglected.
An existing client who trusts you needs no qualification, no trust-building and no procurement conversation. Yet most agencies do the work they were asked for and never propose the adjacent thing.
Two practical moves: put a "what we would do next" section in every client report, and use the wrap-up session at offboarding to give an honest view of what would deliver most value over the next six months. Both are advisory rather than salesy, and both convert.
5. Targeted outreach that references something real
Cold outreach works for agencies in one form: small volume, specific target, genuine observation.
"I noticed your pricing page loads in eleven seconds on mobile, which is likely costing you enquiries" is a real message to a real person. It works because it demonstrates attention rather than reach.
What does not work is volume. Fifty personalised messages a month beat five thousand templated ones, because the templated version signals that you have not looked - and an agency that has not looked is exactly what nobody wants to hire.
6. Speaking and community
Slow, high-trust, and effective in niches with a genuine community. Industry events, meetups, podcasts, associations.
The return is rarely from the room. It is from the artefact - the talk, the episode, the recording - and from being the person who spoke, which is a durable credential in a small community.
A note on effort
Every channel below that works requires sustained, unglamorous effort over months. Every channel that fails is available immediately. That is not a coincidence, and it is the most useful thing to understand about agency business development.
The channels that mostly do not work
Worth being direct, because agencies spend money here.
Broad paid advertising. The buyer is not in-market when they see it, the purchase is considered, and the audience is enormously wider than your actual market. Narrow, intent-based search advertising can work for specific high-intent terms. Broad awareness advertising for a ten-person agency almost never does.
Generic cold email at volume. Covered above. It also damages the brand you are trying to build, because being the agency that sends templated mail is a position.
Directories and lead marketplaces. Occasionally useful, and the buyers there are usually shopping on price and speaking to five suppliers. The economics rarely work once you count the proposals written for deals you never had a real chance at.
Social media as a primary channel. Useful for staying visible to people who already know you. Rarely a source of new demand on its own, and it consumes far more time than it appears to.
The pattern: breadth fails, specificity works. Anything that reaches many people shallowly performs badly; anything that reaches few people with genuine relevance performs well.
Choosing two
The most common business development mistake at an agency is doing a little of everything.
Six channels worked at 15% effort produce nothing, because every channel here has a threshold below which it does not compound. Two channels worked properly for twelve months produce a pipeline.
Choose based on where you already have an advantage. A founder with a strong network should work referrals and partnerships. One who writes well should work content and speaking. One with deep domain expertise should specialise and let content do the work.
Always include referrals. Cheapest, highest-converting, and it exists whether or not you manage it.
Give it twelve months. Content and community both take six to twelve months to show anything. Abandoning at month four - which is when the effort has been made and the return has not arrived - is the single most common reason agencies conclude a channel does not work.
When the pipeline is empty right now
The channels above are the twelve-month answer. Sometimes the question is this month.
In order of speed:
Contact every past client. Not a campaign - individual messages to people you worked with. "It has been a while, how did the site end up performing?" A meaningful proportion of these turn into work, because their needs have moved on and nobody has asked.
Ask your best three clients for a referral, specifically. Named profile, forwardable line, low commitment.
Contact every agency you know. Tell them what you are looking for and what capacity you have. Agencies frequently have overflow they are struggling to place.
Propose the next thing to current clients. The adjacent work you know they need.
Revisit lost proposals from the last twelve months. Circumstances change, budgets reset, and the supplier they chose may not have worked out. A short, unembarrassed message converts more often than expected.
All five are relationship-based, because relationship-based is the only thing that moves quickly. Anything requiring reach takes months.
The compounding point
A channel worked for three months produces almost nothing. The same channel worked for eighteen produces a pipeline. That asymmetry is why consistency beats intensity here.
Measuring it
Two rules and one trap.
Measure revenue, not enquiries. A channel producing thirty enquiries that close at 3% is worse than one producing four that close at 50%, and enquiry counts flatter the broad channels systematically.
Record the source of every enquiry, in one field, consistently. Almost no agency does this, and without it every conclusion about channels is anecdote.
The trap is attribution. A client who read two articles, met you at an event and then got a referral from a colleague records as a referral. The referral got the credit; the content and the event made it convertible. This is why judging content purely on directly attributed revenue reliably undervalues it, and why the honest approach is to look at whether the total pipeline is growing rather than at precise per-channel attribution.
Building the habit rather than the campaign
The reason most agency business development fails is not channel choice. It is that it happens in bursts - a fortnight of intense effort when the pipeline thins, then nothing for four months once work arrives.
That pattern produces the worst of both. The effort never compounds, because nothing runs long enough. And the bursts happen at exactly the wrong moment, when you are least able to sustain them and most likely to accept badly-fitting work.
The fix is small and constant rather than large and occasional. Two hours a week, every week, regardless of how busy you are. That is enough to publish something monthly, maintain half a dozen partner relationships, and stay in contact with past clients - which between them is most of what works.
Put it in the calendar as a recurring block and treat it with the same seriousness as a client meeting. The reason it gets skipped is that it never has a deadline, and the only defence against that is a standing commitment.
Measure the habit, not just the outcome. Did the two hours happen? Over a quarter that is a number you control, unlike enquiries, and it is the leading indicator of everything else.
Agencies with consistent pipelines are rarely doing anything clever. They are doing something small every week for years, which is considerably harder than doing something clever occasionally.
Qualifying, so the pipeline is worth having
A pipeline full of the wrong opportunities is worse than a thin one, because it consumes proposal time you never had a chance of converting.
Four questions worth answering before writing anything:
Is there a budget, roughly? Not the exact figure - a range. A prospect who genuinely does not know is usually not far enough along to buy.
Is there a decision-maker in the conversation? If you have never spoken to anyone who can approve spend, you are being used to build a business case.
Is there a deadline or a trigger? "Why now" is the single best predictor of whether a deal closes. Without a trigger, it slips indefinitely, however enthusiastic everyone is.
Are we the right fit, honestly? Declining work you are not suited to is a positioning decision, and it protects the margin and the reputation that make the rest of this work.
Our guide to client intake covers building this into a repeatable process rather than a judgement call made under the enthusiasm of a first conversation.
Why the slow channels are the defensible ones
A structural point worth understanding, because it explains why the effective channels are the uncomfortable ones.
Anything that works quickly is available to everyone. Paid advertising, cold outreach at volume, directory listings - these can be switched on this week by any competitor with a budget, which means whatever advantage they produce is temporary and contested.
The channels that take twelve months - reputation in a niche, partner relationships built by giving first, a body of published work, a client base that refers - cannot be bought or copied quickly. That is exactly why they take twelve months, and it is also why they keep working once established.
The practical implication: an agency that only ever runs fast channels is permanently competing on spend and never accumulates anything. One that runs a slow channel consistently for two years has something a better-funded competitor cannot simply outbid.
Which does not mean ignoring the fast ones. It means recognising that they buy time rather than building position, and that the slow work has to happen alongside them rather than after them - because the moment there is time for it is the moment you least need it.
What to do in the first ninety days of trying
A realistic sequence for an agency starting from nothing but existing relationships.
Weeks 1-2: harvest what exists. Message every past client individually. Ask your three best clients for a specific referral. Contact every agency you know. This is the fastest available revenue and it costs only time.
Weeks 3-4: fix the shopfront. Rewrite the homepage to speak to one buyer. Add two case studies with actual numbers. If a prospect referred to you looks you up and finds a generic site, the referral works harder than it should.
Weeks 5-8: pick the second channel and start. Content, partnerships or targeted outreach. One, not three. Publish or contact something every week without exception.
Weeks 9-12: build the referral habit. Add the testimonial and referral asks to your project close-out as scheduled steps with an owner, so they happen without anyone remembering. Covered in building a referral programme.
At the end of ninety days you will have some revenue from the first two weeks, a site that converts a referral properly, one channel with a small amount of compounding started, and a process that produces referrals from every future project.
What you will not have is a full pipeline from the new channel, because that takes six to twelve months. Expecting it at ninety days is the single most common reason agencies abandon a channel that was working.
The one thing that beats every channel
Do excellent work and finish it properly.
That sounds like a platitude and it is mechanically true. Every channel above is a multiplier on reputation. Referrals require someone willing to vouch. Partnerships require agencies confident recommending you. Content converts because the work behind it is real.
Which means the highest-return business development activity available to most agencies is not a channel at all - it is the handoff, the offboarding and the follow-up that turn a completed project into a testimonial, a case study and a referral path. Those cost hours, not budget, and almost nobody does them.
Tracking it without a CRM
You do not need sales software to run this well. You need four fields, consistently filled in.
Source. How they found you, in one word. The field almost nobody maintains and the one that makes every other conclusion possible.
Stage. Conversation, proposal sent, won, lost.
Value. Approximate is fine.
Reason, if lost. In their words rather than your interpretation.
A spreadsheet is entirely adequate for an agency winning fewer than fifty clients a year. What matters is that every enquiry gets recorded on the day it arrives, because retrospective reconstruction is where this data goes wrong.
Reviewed quarterly, those four fields tell you which channels produce revenue rather than noise, what your real conversion rate is, and how long your sales cycle actually takes - which is usually longer than anyone thinks.
The summary
Agency business development is narrow and slow, and that is a feature. You do not need many clients, so you do not need reach.
Start with referrals, because the relationships already exist. Add one more channel that suits your actual strengths. Work both for twelve months before judging.
And do the two things that require no strategy at all: propose the next piece of work to clients you already have, and message every past client to ask how it went. Both are available this afternoon, and both convert better than anything else on this page.