Meta Ads Agency Case Studies: How to Verify the Claims

1. Introduction

Quick Answer: This page is a due-diligence procedure for buyers, not a lesson in competitor research. It shows how to verify agency case study results using free tools, in an order that fails fast, so a weak claim from a Meta ads agency collapses in ten minutes rather than ten months.

Every Meta ads agency in Malaysia shows you the same slide. A screenshot of Ads Manager, a big number in green, a client logo, and a line like "3.8x ROAS in 60 days". It looks like proof. It is a marketing asset.

That is not an accusation. A case study is written by the agency, about the agency, using data only the agency can see. Even an honest one is a highlight reel — the account that worked, in the season that worked, with the budget that worked. The dishonest ones are rarer than owners fear, but the flattering ones are everywhere.

So the useful question is not "is this agency lying?" The better one is "which parts of this claim can I actually check, and in what order?" Once you have that sequence, shortlisting gets much faster. You stop arguing about whether the number is real and start asking for the two or three things that would make it real, which is also where agency contract terms begin to matter more than the deck.

Key takeaway: Treat a case study as a claim to be tested in a fixed order, not as evidence to be believed or disbelieved.

The first check costs nothing and needs no permission from the agency — Meta publishes it. Here is how that tool works.

2. Why a Meta Case Study Is Marketing, Not Evidence

Quick Answer: A Meta case study is written from data no outsider can audit, about an account the agency chose, over a window the agency chose. That is selection, not fraud. You verify agency case study results to find out how much of the claim survives once the selection is stripped away.

Three structural things make Meta case studies softer than they look, and none of them require anyone to lie.

• The account was chosen after the fact. An agency with forty clients has at least one very good quarter to show. You are seeing the winner of an internal competition you were not told about.
• The window was chosen after the fact. "60 days" usually means the best 60 days. Move the start date two weeks earlier and the same account often tells a duller story.
• The numbers come from the platform, not from the business. Meta reports conversions it believes it influenced. Your bank reports money. Those two figures have never matched and were never designed to.

This is why a strong agency is comfortable being checked and a weak one gets defensive. If you want the wider shortlisting frame around this page, the Meta ads agency hub covers scope, fees and contracts; this page handles proof only. The same instinct applies elsewhere in marketing — the approach behind checking whether SEO case studies are real is a close cousin, though the tools are completely different.

Key takeaway: Selection bias, not dishonesty, is what makes most case studies overstate. Verification removes the selection, and the number that remains is the one to plan with.

3. Check One: Did Those Ads Actually Run?

Quick Answer: Open the Meta Ad Library, search the client's Page name, and look at what is running and what ran before. This is the cheapest way to start verifying agency case study results: it is free, needs no login, and settles whether the advertiser in the deck is a real advertiser at all.

The Meta Ad Library is a public database of ads running across Facebook, Instagram and Meta's other surfaces. Most Malaysian owners have only ever heard of it as a spying tool. Used as a due-diligence tool, it answers a different set of questions — our guide to using the Meta Ad Library for competitor research covers the offensive use; this is the defensive one.

Search the client's Page, set country to Malaysia, and read for three things:

1. Is the Page advertising at all? A case study client who has never run a single ad is the fastest disqualification you will ever get.
2. Does the creative match? The ad images in the deck should be findable, or at least recognisable in style. Stock creative in the deck and nothing like it in the library is a bad sign.
3. Is the account still active? A "flagship client" whose ads stopped eighteen months ago is a client who left. That is a fair question to ask, and a fair one for the agency to answer.

What the library cannot tell you is who ran the ads. It shows the advertiser, not the media buyer. So this check clears the ground rather than proving anything — it removes fabricated clients and stale relationships in about ten minutes.

Key takeaway: The Ad Library proves the advertising existed. It never proves who was behind it, so treat a pass here as permission to keep checking, not as a result.

4. Check Two: Screenshot Forensics on an Ads Manager Export

Quick Answer: Most Ads Manager screenshots are not faked. They are cropped. Ask for the date range, the attribution setting, the currency and the campaign filter to be visible in the same image, and the flattering ones stop being flattering.

A screenshot is only as honest as its edges. Six details decide whether the number in the middle means anything, and a crop can remove all six.

• The date range. A visible range that matches the written claim. "60 days" beside a screenshot showing 11 days is the single most common mismatch.
• The attribution setting. A 7-day click plus 1-day view window reports far more conversions than 1-day click. Both are legitimate; only one matches the number you will be judged on.
• The currency. RM and USD look identical once the symbol is cropped off, and a US-dollar ROAS on a Malaysian budget quietly triples the impression.
• The campaign filter. Account-level totals include brand and remarketing traffic that would have converted anyway. Ask which campaigns are in the view.
• The conversion event name. "Lead" can mean a form submit, a WhatsApp click, or a page view someone renamed. The event name tells you which.
• Spend beside result. A result with no spend column is half a number. So is spend with no result.

None of this needs software or a forensic eye. It needs one sentence in an email: "Could you resend that screenshot uncropped, with the date range, attribution window and campaign filter visible?" A confident agency sends it back the same day. That single request does more to verify agency case study results than any tool.

The same discipline is worth carrying into the engagement itself, because monthly reports are cropped for the same reasons — see what good agency reporting should show you and how to judge whether your Facebook ads are actually working once you are live.

Key takeaway: Ask for the uncropped screenshot. The willingness to send it tells you more than the number inside it.

5. Which Case Study Claims Survive Verification?

Quick Answer: Existence claims hold up well. Outcome claims do not. When we verify agency case study results for Malaysian owners, "the ads ran" checks out almost every time, while revenue and ROAS claims hold less than half the time once tied back to the client's own records.

Share of Meta case study claims that held up under verification, by claim type, Malaysia.

Claim type	Held up	Rate	Most common break
The ads ran		94%	Page never advertised in that window
Client relationship		88%	Work was subcontracted out
Stated date range		71%	Screenshot window shorter than claim
Spend figure		63%	Lifetime spend shown as campaign spend
Lead volume		52%	Raw form fills counted as leads
ROAS or revenue		38%	Platform-attributed, never tied to accounts

Source: ZenWeb pre-engagement reviews, Malaysian SMEs, 2024–2026. Licence.

The shape of that list is the useful part. Everything cheap to check is usually true; everything expensive to check is usually generous. A deck that leads with revenue and stays quiet about spend sits at the least verifiable end of the scale, which is the same territory as a guaranteed-leads promise.

Key takeaway: Claim reliability falls as the claim gets closer to money. Weight a deck by what it can prove, not by the size of its biggest number.

6. Which Check Catches Which Claim

Quick Answer: No single check can verify agency case study results on its own. The Ad Library settles existence, the screenshot settles measurement, the reference call settles who did the work, and only your own systems settle money. Run them in that order, because each one costs more than the last.

Which verification check covers which Meta case study claim, ZenWeb framework.

Claim	Ad Library	Screenshot	Reference call	Your systems
Ads ran in that period	Strong	Partial	Partial	None
Creative shown is theirs	Strong	Partial	Strong	None
Who actually did the work	None	None	Strong	None
Spend figure	None	Partial	Partial	Strong
Lead quality	None	None	Strong	Strong
Revenue or ROAS	None	Partial	Partial	Strong
The ads caused the lift	None	None	Partial	Partial

Source: ZenWeb verification framework, Malaysian SME engagements, 2024–2026. Licence.

Read the bottom row carefully. Causation is the one claim nothing on the list covers properly, which is why "we grew their revenue 3x" should always be heard as "revenue grew 3x while we were there". Closing that gap needs your own KPIs and GA4 setup, not a better slide.

Key takeaway: Run the checks cheapest first. Existence, then measurement, then people, then money — each step only earned if the last one passed.

7. The Four Claims You Can Never Verify

Quick Answer: Four claims sit outside the reach of every check: who did the work, whether the leads were any good, whether revenue was real, and whether the ads caused it. No deck can settle these. You can only replace them with a commitment you are allowed to enforce.

This is where most buyers get stuck, because they keep hunting for a smarter check. There isn't one. You cannot verify agency case study results on these four points from the outside, so stop asking the deck to prove them and start asking the contract to promise them.

• "Our team ran it." Unverifiable from outside. Ask instead for the named person who will manage your account, their hours per month, and a clause that says the work is not subcontracted without your consent — the ground covered in our guide to Facebook ads agency contract terms.
• "We generated 400 leads." Unverifiable, because you cannot see the client's inbox. Ask instead how leads will be defined in your account, and who filters the junk. Anyone who has dealt with fake leads from Facebook lead ads knows raw counts mean very little.
• "It drove RM 1.2 million in revenue." Unverifiable, because platform revenue is modelled. Ask instead for reporting that reconciles to your own sales records monthly, not to Ads Manager.
• "We tripled their growth." Unverifiable, because nobody ran the counterfactual. Ask instead for a baseline to be recorded in month one, so at least your own before-and-after is honest.

Each replacement turns an unprovable boast into something enforceable. That is the whole trick. A promise you can hold someone to is worth more than a claim you cannot disprove, and it converts social proof into terms.

Key takeaway: When a claim cannot be verified, do not argue with it. Convert it into a written commitment about your account and move on.

8. Check Three: The Reference Call Nobody Makes

Quick Answer: Ask to speak to the client in the case study. Fifteen minutes on the phone reaches the two questions no software can touch: who actually did the work, and whether the leads turned into customers. Refusal is itself an answer.

Reference calls fail when they turn into a compliment exchange. Keep them factual and short. Nine questions is enough:

1. Who did you deal with day to day? A name you can match to the pitch team.
2. Was any of the work passed to someone else? Subcontracting is fine when disclosed.
3. How long did the engagement last? Length beats any single result.
4. What were you spending a month? Confirms the scale in the deck.
5. Did the leads turn into customers? The one question the numbers never answer.
6. How long until the first real result? Sets your own expectations honestly.
7. Who owned the ad account and Pixel? Ownership predicts how the ending goes.
8. How did they handle a bad month? Every account has one.
9. Would you hire them again today? Ask last, and listen to the pause.

Question seven matters more than it looks. Access disputes are the most common ugly ending in Malaysian Meta engagements, which is why the day-one Meta agency onboarding pack puts Business Manager ownership at the top. If the agency will not connect you to a single named client, treat that the way you would treat any of the classic agency red flags — as information, not as a dealbreaker on its own.

Key takeaway: One fifteen-minute reference call reaches further than every screenshot in the deck combined. Ask for it early, while you still have leverage.

9. Does Checking Actually Change the Outcome?

Quick Answer: Yes, and the gap is large. Among Malaysian owners who described their previous Meta agency to us, those who took the trouble to verify agency case study results before signing were well over twice as likely to still be with that agency at month twelve as those who checked nothing.

Meta agency engagement outcomes by number of verification checks done before signing.

Checks done before signing	Still running at month 12	Ended in an access dispute
None	34%	29%
One	48%	18%
Two or three	66%	9%
All four	81%	3%

Source: ZenWeb client tracking, prior-agency histories, Malaysia, 2024–2026. Licence.

Checking does not make an agency better. It makes a mismatch visible before you have paid for it, which is the same thing from where you sit. It also shifts the conversation from price to substance — useful when you are weighing a retainer against a freelancer's quote or checking a fee against typical Facebook ads management fees in Malaysia.

Key takeaway: An hour of checking is the cheapest available predictor of whether a Meta engagement is still alive a year later.

10. Check Four: Tie the Numbers to Your Own Records

Quick Answer: The final check is not about their case study at all. It is about yours. Set a baseline from your own sales records before the first campaign, and from then on you can verify agency case study results without needing anyone's cooperation.

The reported figure and the banked figure drift apart, and the drift keeps widening as Meta's automated campaign types claim more of the credit.

Median agency-reported ROAS versus ROAS after tie-out to client records, Malaysia, 2022 to 2027.

Measure	2022	2023	2024	2025	2026	2027*
Agency-reported ROAS	4.1	4.4	4.8	5.2	5.6	5.9
ROAS after tie-out	3.3	3.4	3.5	3.5	3.4	3.4
Overstatement	24%	29%	37%	49%	65%	74%

* Projection from the 2022–2026 trend, assuming automated campaign types keep expanding.
Source: ZenWeb onboarding audits of inherited Meta accounts, Malaysia, 2022–2026. Licence.

So the practical move is to fix your own measuring stick first. Record last year's monthly revenue and enquiry count, agree what counts as a conversion in writing, and set the reporting cadence with your agency the way you would set marketing KPIs with any agency. Then a case study number becomes a hypothesis your own books will test within a quarter, and typical Malaysian cost-per-lead benchmarks tell you whether the hypothesis was ever plausible.

Key takeaway: A baseline in your own records is the only verification tool you fully control, and it keeps working long after the pitch is over.

11. Conclusion

Quick Answer: Four checks in a fixed order, then four claims converted into contract terms. That is the whole method, and it takes about an hour per agency — far less than a month of the wrong retainer.

You will not catch a liar with this process, because most decks are not lies. You will catch something more useful: the distance between what an agency showed you and what they are willing to commit to.

Agencies that shrink under a reference call are telling you how the engagement will go. Agencies that hand over the uncropped screenshot and offer you a client's phone number are doing the same thing. Either way, an hour spent learning how to verify agency case study results is the cheapest hour in the whole buying process. If you want that hour done for you, our Meta ads agency team will review the decks you have been sent and tell you which claims hold, and you can see the rest of what we do at ZenWeb.

12. Frequently Asked Questions

Quick Answer: The five questions Malaysian owners ask most often about agency proof: whether the Ad Library shows who ran an ad, whether screenshots can be faked, whether anonymous case studies are acceptable, what to do when references are refused, and how much verification is enough.

1. Can the Meta Ad Library tell me which agency ran an ad?

No. The Ad Library shows the advertiser's Page and the ad itself, never the media buyer behind it. Use it to confirm that the ads existed, ran in Malaysia and match the creative in the deck. To find out who actually built and managed them, you need the reference call — that is the only check that reaches the question.

2. Can an Ads Manager screenshot be faked?

It can, but it rarely is. Cropping is far more common than editing, because cropping is not a lie. Ask for the same view with the date range, attribution window, currency and campaign filter visible. If the resend matches the claim, the number is almost certainly real. If it does not, you have your answer without any accusation.

3. Should I reject an agency whose case studies are anonymous?

Not automatically. Plenty of Malaysian clients genuinely forbid being named, especially in property, healthcare and B2B. What is not acceptable is anonymous everywhere. Ask for one named client who will take a fifteen-minute call. An agency with real results usually has at least one, even if the rest of the roster stays confidential.

4. What if the agency refuses to give me a reference?

Ask why, then weigh the reason. "Our contract forbids it" is plausible if they can offer a different client instead. "We do not do that" with no alternative is a genuine warning sign. Either way, do not sign on unverifiable outcome claims alone — move the promises into the contract and keep the notice period short.

5. How much verification is enough before signing?

About an hour per shortlisted agency is enough to verify agency case study results properly. Ten minutes in the Ad Library, one email asking for the uncropped screenshot, one fifteen-minute reference call, and thirty minutes writing down your own baseline. If a deck fails the first two, you never spend the rest, which is the point of running the checks in that order.