How do I verify a marketing agency's case study?

Pick one result and ask for the starting number and date range, how it connected to pipeline or revenue, what else changed at the company in that period, and how long it lasted. Then ask to speak with the client or see the numbers in the client's own dashboard. If the agency can answer all of that about one project, the result is credible.

Are agency reviews on sites like Clutch trustworthy?

They're useful for confirming that a client and project were real, since Clutch says it verifies that and labels reviews it can't confirm. They don't independently check the performance numbers a reviewer mentions. Use them as a reason to call a reference, not as a substitute for one.

Should I hire an agency that guarantees results?

I wouldn't. No agency controls your market, the ad platforms and your sales follow-up all at once. Guarantees usually come with a narrow definition of the result or a contract clause that moves the risk back to you.

What results should a SaaS company expect in the first few months?

Expect a clear baseline, working tracking, and early signal from fast channels like paid search within the first couple of months, plus a written plan for what gets tested next. Slower channels like SEO and content take longer. If you want a sense of what that work costs, see how much a SaaS marketing agency costs.

Is a big percentage increase a good sign?

Only once you know the base. A large percentage on a small starting number can be fewer extra customers than a modest percentage on a big one. Always ask for the before and after as plain numbers.

What "proven results" should mean from a SaaS marketing agency

Proven results from a marketing agency should mean a business outcome (demo requests, qualified pipeline, revenue or cost per customer) measured against a stated starting point, over a stated window, for a company like yours, that you can check yourself. Verify it by asking for the baseline, what else changed, whether it lasted, and a call with the client.

Every agency says "proven"

Picture the search. You're the founder or the one-person marketing team at a Series A software company, and you've typed some version of "marketing agency for tech companies with proven results" into Google or ChatGPT. Twelve tabs later, every agency has a logo wall, a "300% growth" badge and the word "proven" somewhere above the fold. None of them look unproven. That's the problem.

"Proven" has become a word agencies put on themselves, like a restaurant calling its own chili award-winning. It might be true. The menu can't tell you.

I'm writing this as an agency owner, so read it with that in mind. My goal is to hand you a test that works on any agency, including mine.

If you're still deciding whether you need an agency at all (versus a fractional CMO or a first hire), start with fractional CMO vs. agency vs. in-house. This piece assumes you've decided on outside help and want to know whose claims to believe.

Take your results to the assay office

Redding sits in old gold country, so here's the picture I keep coming back to. In a gold rush town, a miner could walk in with a sack and swear it was full of gold. Nobody paid him for the swearing. He took it to the assayer, who weighed it, tested it and wrote down what was actually in there. The assay slip was what got honored, not the story.

An agency case study is the miner's sack. Some are full of gold. Some are mostly pyrite with a nice paint job. Your job as a buyer is to run the assay (weigh it, test it, write down what's really there) before you pay for it.

What "proven" should mean for a tech company

For a SaaS or tech company, a result counts as proven when it clears four bars:

It's a business number. Demo requests, sales-qualified leads, pipeline, closed revenue, customer acquisition cost, payback period. Impressions, followers and "engagement" can be leading signs, but on their own they're ore, not gold.
It has a baseline and a window. "Demo requests went from X a month to Y a month between March and September" is a claim. "Demo requests up 400%" is a headline.
It came from a business like yours. A result for an ecommerce brand selling $40 candles tells you little about selling a $30,000 annual contract through a sales team with a 90-day cycle.
You can check it. A named client, a reference call, a screen share of the real dashboard. If none of those are available, the claim may still be true, but it hasn't been assayed.

That last bar is the one most agency pages skip, and it's the one that matters most.

The five-point assay

Here's the test I'd run on any agency's best case study. Ask all five questions about one result, not five different results. A strong agency can answer every one about the same project.

#	Test	Ask	Good sign	Red flag
1	Weigh the baseline	"What was the starting number, and over what window?"	A plain number and dates	Only a percentage
2	Check the metal	"How does this connect to pipeline or revenue?"	They trace it to deals or sales-qualified leads	They change the subject to reach or traffic
3	Look for other hands	"What else changed in that period?"	They name it: a new sales hire, a product launch, a price change	"Nothing, it was all us"
4	Test whether it held	"How long did it last, and how recent is it?"	It held for months and the team that did it is still there	A single great month from four years ago
5	Find a witness	"Can I talk to the client, or see the real dashboard?"	Yes, with some coordination	Every example is anonymous

1. Weigh the baseline

Percentages are the easiest number to inflate without lying. Going from 2 demo requests a month to 10 is a 400% increase, and it's also eight more demos. Going from 200 to 260 is "only" 30% and is probably the bigger business win.

So ask for the starting number. That goes for us, too: our homepage says we helped Adora POS grow demo requests 875% over a seven-year relationship. Hold that claim to this exact test. Ask me what the starting point was, what else changed at Adora along the way, and whether you can talk to them. A result you can't question isn't proven yet, no matter whose website it's on.

2. Check the metal

Tech marketing has a long chain between a click and a closed deal, and every link is a place where a result can be real but useless. Plenty of agencies have legitimately tripled a client's form fills by loosening who counts as a lead, then watched sales ignore all of them.

Ask how the result connects to the sales side. A good answer sounds like "these demo requests became sales-qualified at about the same rate as before, so pipeline grew with them." A weak answer is a new chart about website traffic.

3. Look for other hands on the pan

Results rarely have one parent. In the same quarter the agency launched new ads, the company might have shipped a big feature, hired two account executives, raised a round that got press, or cut its price. Any of those can move demo numbers on their own.

An honest agency will tell you what else was going on, and it will still be able to explain why it thinks its work mattered. I trust "we think our landing page rebuild drove most of it, and the new sales hire helped close more of them" far more than "it was all us." The first one sounds like someone who looked.

4. Test whether it held

A spike is easy. Turn up ad spend, run a discount, get a lucky mention in a newsletter, and one month looks heroic. What you want to know is whether the number stayed up after the push ended.

Also ask how old the result is, and whether the people who produced it still work there. Agencies change. A great case study from 2021 tells you what a different team did in a different market, with different ad costs and different buyers.

5. Find a witness

This is the closest thing to an assay slip. Ask to talk to the client behind the result, ideally the person who hired the agency and watched the numbers. Ask them what the agency was like to work with when things went badly, because every engagement has bad months.

Review sites can help here, as long as you know what they check. Clutch, for example, says its team verifies each review "to ensure no conflict of interest and that the project occurred," and labels the ones it can't confirm as Not Verified. That tells you the client and the project were real. It doesn't audit the numbers in the review, so treat it as a witness to the work, not to the results.

Why so many results can't be checked

Not every anonymous case study is hiding something. Some clients sign contracts that keep the agency from naming them. Some don't want competitors to know what's working. That's fair.

But anonymity should make an agency work harder on the other four tests, not skip them. "A Series B fintech company" plus a real baseline, a real window, a frank list of what else changed and a reference you can call is still checkable. "A leading SaaS brand saw explosive growth" is just the sack.

There's also a legal floor worth knowing about. The FTC's rule on fake reviews and testimonials, finalized in 2024, bans testimonials that misrepresent who the reviewer is or their experience, along with undisclosed insider reviews. It was written with consumer reviews in mind, so it won't cover every B2B case study, but it tells you where regulators think the line is: a testimonial has to come from someone real who actually had the experience.

Red flags that should end the conversation

Some answers are disqualifying on their own:

Guaranteed results. "We guarantee 50 demos a month" or "page one in 90 days." Nobody controls the buyers, the algorithms and your sales team at once. A guarantee usually means the fine print redefines the result.
Results only in their own reporting tool. If the proof can't be shown in your own Google Ads, analytics or CRM account, you're looking at a painting of the gold.
Every case study is from a different industry. Range is fine. But if nothing in the portfolio sold a considered B2B purchase, you'd be paying for their first try.
No bad months. Ask about a campaign that failed and what they learned. If they can't name one, they either haven't run many or don't look closely at the ones they ran.

If you want a fuller list of interview questions to go with the assay, we keep a printable one in questions to ask a marketing agency.

Proven for them, or proven for you?

Even a result that passes all five tests is a record of the past. The more useful question is whether the agency can show you how it would find the gold in your business.

So in the sales process, ask them to diagnose something real. Share your demo page, your last quarter's numbers or your ad account (read-only is fine) and ask what they'd test first and why. You'll learn more from 30 minutes of how they think than from a dozen polished case studies. An agency that can explain its reasoning out loud is one you'll be able to check every month after you sign, too. If you've already hired someone and the monthly reports feel thick but empty, these are the signs your agency isn't working.

If you'd like a second opinion on an agency proposal you're weighing, or want to put us through the same five tests, book 30 minutes with me. Bring the case studies. I'll bring a magnifying glass.

Here's to clients who ask for the assay slip, and agencies happy to hand it over. 🍻