How Marketing Agencies Get Clients Without Referrals

A marketing agency that cannot count on referrals gets clients a different way. It picks one type of client, then contacts enough of those buyers directly, by cold email and LinkedIn, to book sales calls on a schedule it controls.

Referrals stay in the plan as a second engine. Content and SEO come third, because they take the longest to pay back.

In 60 days, LinkedIn and email outreach for one US marketing agency produced 962 replies, 277 sales calls and 34 signed contracts worth more than $255,000. Unlike referrals, outbound can be planned with arithmetic before you send anything, and that math is further down.

Why referrals stop being enough

Most agencies still win clients through referrals. In the AgencyAnalytics 2026 benchmarks report, a survey of 494 agency professionals run by a reporting software vendor, 85% named referrals as a lead source. Only 33% named outbound outreach.

The same report also shows that winning new clients is the biggest operational challenge for 30% of agency leaders, and it has held the number one spot since 2022.

Those two numbers belong together. A referral arrives when a happy client happens to meet someone who needs you. You do not choose the month it lands, the size of the account or the industry it comes from.

At five clients that is tolerable. At fifteen, with payroll that depends on replacing lost accounts on time, it becomes a planning problem. A founder in that spot is exactly one slow referral month away from a revenue crisis.

Referrals do carry one real advantage.

In an August 2026 roundup, Promethean Research, a research and consulting firm that works with digital agencies, reported that clients won by referral and word of mouth kept working with an agency roughly twice as long as those that came from events, networking or outbound.

So keep the referrals coming. Just stop building the payroll plan on them.

What outbound produced in two marketing agency campaigns

Smirnov Consulting Group is a Prague-based B2B outbound lead generation agency that runs cold email and LinkedIn campaigns for founder-led B2B companies and books qualified sales calls.

My team ran both campaigns in the table for marketing agencies, and each has its own published case study.

"Warm email" in the table is email sent after the first contact on LinkedIn, so the email is not the first time the sender reaches out.

Not reported means the published case does not give that figure for the period shown.

US marketing agency: 60 days

Out of 962 replies, 277 became booked calls, about 29 calls for every 100 replies. Those calls produced 34 contracts, roughly 1 contract for every 8 calls. More than $255,000 across 34 contracts works out to an average of at least $7,500 each.

A note on the denominator, because this is where agency numbers often get blurry. These counts are InMails and emails sent, 4,878 in total. Connection requests are not in them, and neither is the number of people reached.

So 962 replies against 4,878 messages, about 20 replies per 100 messages, is a per-message figure. Do not compare it with a reply rate calculated per person.

UK social media marketing agency: 12 months

This one started at 44 calls in month 1 and reached 68 by month 3. Over the full year it booked 714 calls, so months 4 to 12 averaged about 60 a month.

It ran across three niches at once, with no paid advertising behind it.

What the two cases tell you

The two campaigns were built differently. The US one ran LinkedIn plus email for 60 days, with three sequence scenarios. The UK one combined LinkedIn, warm email and cold email for a full year, split across three niches.

The pace differed as well: roughly 140 calls a month for the US campaign over its two months, about 60 a month for the UK one across its year. Without the UK message count, there is no telling how much of that gap is volume.

Read the two results as a range. Neither of them is a promise.

The UK case is also the one with a visible ramp. Month 1 was its smallest of the first three, and calls grew by just over half from month 1 to month 3.

Its later months averaged below month 3 but well above month 1, so month 1 understated what the campaign went on to do.

The contact math for your own calendar

This part is plain arithmetic. I plan with a range: each booked call takes 150 to 250 contacted people, and I use it to size a market in my piece on picking a market before writing any email.

Niche, country and channel decide where a campaign lands in it. Treat it as a planning figure drawn from my own campaigns.

Notice the denominator: people contacted. One booked call per 150 people is about 0.67% of the people you reach, and one per 250 is 0.4%.

The US case counts messages rather than people, and the UK case does not report its message count for the year, so neither fits on that scale. Plan with the range and treat anything better as upside.

Worked through for three targets

Now turn calls into clients. Only the US case has a contract count: 34 from 277 calls, about 1 in 8. At that rate, 10 calls a month would produce a little over 1 new client a month.

That close rate belongs to one case. Your own close rate on past sales calls is the better input, so use it if you have it.

Can a small agency produce that volume?

On paper, yes. In one 2023 financial advisory campaign, each LinkedIn account, paired with its own email accounts for cold emails and follow-ups, reached out to roughly 2,800 people every month.

The 10-calls-a-month line in the table needs 1,500 to 2,500 people, which fits inside one such setup.

For a three-person agency the tighter limit is usually hours. Somebody has to read and answer replies every working day, and at 10 calls a month somebody has to take 10 calls on top of client work.

Then check the market itself. A year at that pace means 18,000 to 30,000 individual buyers, and nobody should hear from you twenty times.

Count the companies in your chosen niche and the plausible buyers in each. If the total falls short, go deeper per company or add a second country.

Choose one kind of client before you add volume

Niching down is standard advice, and most agencies already call themselves specialists. Promethean's 2026 Digital Agency Industry Report, a survey of 1,452 digital agency leaders, found that 86% identify as specialists, 77% by service and 55% by industry.

So a niche on its own does not make you stand out.

The question is whether a stranger can tell, within a few seconds of opening your email or profile, what you fix, for which buyer, what makes your approach different and why they should pick you.

I call that packaging, and I have put the diagnosis bluntly before: "You might not have a lead gen problem. You might have a packaging problem."

Keep the volume and the effort where they are, fix the packaging, and the results can look completely different. The tactics stay the same. The market finally understands what it is being offered.

The longer version of this argument is in my post on one offer and one audience.

The fix is one clear promise to one clear type of buyer. For an agency, that promise has to fit in one line.

"We help brands grow online" does not. "LinkedIn ads for accounting firms with 20 to 100 staff" does, and it also tells you who goes on the list and which case studies to show.

Hold the position for a year

My rule is to pick one service and one audience and commit to it for 12 months, instead of chasing fifty directions at once. Two different clocks are at work here, and it helps to keep them apart.

The first is the campaign. The UK ramp above is the reason to give a message a full quarter before you judge its replies and calls.

The second is the position. Your profile, website, case studies and outreach all have to tell the same story, and every rewrite of the position resets that work. That is why the commitment runs for a year rather than a quarter.

The UK agency in the table ran three niches inside one campaign. That works as long as each message is written for one kind of buyer. A message aimed at everyone lands with no one in particular.

If you have no paid client results yet, outbound comes second. Getting leads for a proven offer is my job. Building a company from zero is a different job.

Outbound multiplies the proof you already have, and at zero there is nothing to multiply.

Get your earliest clients from direct conversations with people who have watched you deliver. Write each result up with real numbers, then start contacting strangers.

How to start outbound, in order

Start with the list. Take the one kind of client you settled on, find more companies shaped like your current paying clients, and name the person in each who actually signs off. Then count them against the contact table above.

Next comes the sending side. Cold email needs domains of its own, never the one your invoices come from. Each domain gets its authentication records (SPF, DKIM and DMARC), and each new mailbox is warmed up before it reaches a prospect.

That warm-up is the waiting part founders tend to hate. Every address is also verified before it enters a campaign, and I cover the whole setup in my piece on the boring part of cold email.

The setup weeks produce nothing you can show anyone, and rushing them to start sooner only burns the list.

Then the channel order. Open on LinkedIn and follow up by email, so the email is not a cold first touch, and add cold email volume once the LinkedIn side is giving you feedback.

Plan several follow-ups and leave proper gaps between the sends. Of all the follow-ups, the third one is usually what works best for us, so stopping after one message is not a real test.

Once replies arrive, they show which angle, which industry and which job title respond. That feedback is what the first month of sending is really for.

Keep referrals running next to outbound

Referrals still earn their place. Promethean asked 91 agency leaders to rate 34 tactics for revenue contribution on a 1 to 5 scale, and client referrals came out on top.

Source: Promethean Research survey of 91 agency leaders, published August 2026. As the report notes, "Only leaders whose agencies use a tactic scored it."

Partner referrals, events and directory listings all score below client referrals in that table. Paid ads are harder to judge from this data.

Promethean scores pay-per-click and paid social only for agencies with low sales complexity, and 26% of respondents in the AgencyAnalytics report named digital advertising as a lead source. Treat all four as supporting channels and judge each by the contracts it brings in.

Outbound creates referrals too. Plenty of campaign replies are not a clear answer either way. Some point you to a colleague, and that name goes cold if nobody on your side answers it quickly.

So the two channels work together. Referrals bring clients who tend to stay longer. Outbound decides how many new conversations you have next month.

Where content and SEO fit

Content is the slow channel. In Ahrefs' study of how long pages take to rank, an SEO tool vendor's analysis published in 2025 using pages first seen in 2023, just 1.74% of the new pages it tracked climbed into Google's top 10 inside 12 months.

Even in a cleaner sample of new pages with real English content, the figure was 6.11%. The Promethean survey points the same way for agencies: blogs and newsletters scored 1.5 in the table above.

Content still has a job in the first year, just a different one. Before a stranger answers a cold email, they tend to look up the sender: the LinkedIn profile, the website, the case studies. What they find there moves the reply rate.

So in the first months, spend content effort on what a prospect checks after the email: a clear profile, two or three case studies that show hard figures, a home page that says what you do and for whom.

The long-form library can grow later, paid for by the clients outbound brings in.

Mistakes agency owners make when they bring in outbound help

Whether you run outbound yourself or pay a specialist, three questions settle most of the decision. Who will build and check the list? Who answers replies, and how fast? Who runs the sales calls?

If nobody on the team has those hours, outsourcing is worth a serious look. These are the mistakes that show up either way.

Buying volume on top of a vague offer. A cheap lead gen shop blasting out emails will not fix a business with broken foundations. Outbound only carries the offer you give it, so more sending simply shows more people an offer they do not understand.

Counting calls instead of contracts. Booked calls flatter everyone. Ask for three numbers every month: calls booked, calls held, and contracts signed. Of the two cases above, only the US one reports contracts, and neither reports calls held.

Switching providers without changing what you ask for. One founder had hired three agencies before us. They promised a lot, delivered little, and he blamed his niche.

Each of them had gone straight to sending before anyone fixed the offer, which is why a string of failed agencies is not the red flag it looks like.

Before you sign with anyone, ask for the arithmetic: how many people, from how many accounts, reaching which buyers. The eight questions I use to vet a lead gen agency cover the rest, including the answers that should end the conversation.

Settling terms after the start. Who owns the sending domains, the lists and the reply history if you part ways? Settle that with any provider before the first message goes out.