Resource

How to Tell a Good Link Building Agency From a Bad One (and the 7 Questions That Expose Them)

Want more from EMGI in your Google results? Add us as a preferred source.

The short version

  • The strongest signal is whether they try to disqualify you.
  • Ask for link samples, not domain rating averages. DR can be inflated for about five dollars.
  • Ask for the floor and the average. Ours is DR 60 minimum, DR 70 to 75 in practice.
  • Vet the work on link relevance, commercial-page authority and buyer intent, not DR and keyword counts.
  • In 2026, “how do we show up in AI answers?” is a disqualifying question. Our study of 137 SaaS companies across 1,486 buying queries shows why.
  • Guaranteed rankings, DR-only reporting, 12-month lock-ins and revenue-multiple claims end the conversation.

A bad provider has never met a budget they didn't like.

If you remember one line from this, make it that one. Everything below is really just a longer way of explaining how to spot it before you have signed anything.

I run a link building agency, which makes this an awkward piece to write honestly. I am describing the questions I would want a prospect to ask me, including the ones that are inconvenient to answer. But the alternative is another agency blog post telling you to check case studies and look for transparency, which helps nobody.

Prefer watching? The seven diagnostic questions further down this post are covered in the video above. More like it on our videos page.

Why this choice punishes you more than most

Every agency decision matters. This one has a specific trap in it.

Most SaaS teams learn this lesson only after the money's gone. In my experience, it usually costs $15,000 to $20,000, and the agency gets fired somewhere around month four. Traffic climbed, domain rating rose, the report landed every month. Then someone in a board meeting asked how much pipeline came from organic, and nobody had an answer.

SEO results do not arrive in the first few days or the first few weeks. Everyone in the industry knows this, and it is genuinely true. Which means a bad provider can pull the wool over your eyes for a long time, and the honest explanation for slow progress and the dishonest one sound identical for the first three months. The real harm is not lost deals you can point at. It is the invisible pipeline that never entered the CRM.

You cannot solve that by waiting to see results. By the time the results tell you, you have spent two quarters of budget. You have to solve it on the way in, at the point where you are still just talking to someone.

So the whole thing comes down to the quality of one conversation.

The single best signal: do they try to disqualify you?

Here is what I would watch for above everything else.

A good agency is trying to work out whether they can actually help you. That means they ask about your business rather than your SEO. How do you currently get customers. Which pages convert. What is the sales cycle. What does a closed deal look like.

And then, crucially, they remember the answers. You can tell within two calls whether someone is assembling a picture of your business or running a script. When it is real, they refer back to something you said earlier and build on it. When it is not, every call starts from zero.

What you want to see is them qualifying you, or disqualifying you. There should be a version of your business where they say no. We turn work away when the budget is too low to produce anything meaningful, and we say so plainly, because one link a month serves nobody. It does not work for the client and it does not work for us, and it is better to have that conversation at the start than to take the money and disappoint someone for six months.

An agency with no disqualifying criteria is not being generous with you. They are telling you their process does not depend on fit.

The strongest signal is not what they promise. It is whether there is a version of you they would turn down.

Ask them why

The most useful question is also the bluntest. Why would we work with you?

Not to be difficult, but because how someone answers tells you how they think. You are listening for whether they are willing to be honest about what works and what does not, including the parts that do not flatter their own service.

Nobody legitimate is going to promise you results. If someone guarantees positions or guarantees revenue, that is the end of the conversation. Google says this itself in its own guidance on hiring an SEO: no one can guarantee a number one ranking, and you should be wary of anyone claiming a special relationship with Google.

But there is a wide gap between refusing to guarantee outcomes and refusing to explain the mechanism, and the good providers live in that gap. They will tell you what they expect to happen, why, and what would have to be true for it to work.

What case studies actually prove

Case studies matter. They just prove less than they appear to.

Every agency cherry-picks. The big ones, the good ones, all of them. It is the nature of the business, and you would do the same. Nobody publishes the engagement that went sideways.

So discount the graphs and look for the rarer thing: a named client, on the record, describing the working relationship. That is deliberately how we built our case study with Constant Hire. Their Growth Marketing Manager, Guillaume Deschamps:

“I review the links once a month and that’s it, because I trust you. I don’t want to be above anyone saying every day that we need to change anything. It’s really like, please do your work, I know you’re good.”

Guillaume Deschamps, Growth Marketing Manager at Constant Hire

Guillaume Deschamps, Growth Marketing Manager, Constant Hire

Video case studies move me more than written ones, because it is harder to fake a person on camera than a paragraph on a page. But even there, be clear about what you do not know. You do not know the full context, how long it ran, how much they spent, or whether the testimonial came with a discount attached. A client who was given money off in exchange for a video is not lying. They are just not a neutral witness.

And I would not rule out an agency simply because it has no case studies. Everyone starts somewhere, and smaller agencies frequently care far more about the outcome because they cannot afford not to.

Here is how I would weigh each kind of proof they put in front of you.

What they show you What it genuinely proves What it cannot tell you
Written case study They have run this play before How many engagements went the other way
Video testimonial A real, identifiable client went on record The budget, the timeline, or whether a discount bought the video
Sample of live links Their actual quality bar, not their claimed one Whether yours will be placed to the same standard
Results on their own site They are willing to test on themselves Whether their niche is easier than yours
A named point of contact Who you will actually be dealing with Whether that person stays after the sale

If you want to see what the format looks like from the outside, we keep a roundup of SaaS link building agencies that includes us, which should tell you how much weight to put on any listicle including its author.

Ask to see the actual links

This is the question most founders forget, and it is the one that cuts through metric arguments fastest.

Ask for a sample of links they have built that are relevant to you. If they cannot show yours, ask for a different client's. They may be limited by NDAs and they may not be able to show you everything, but they should be able to show you something. Some agencies go further and let you pre-approve every placement before it goes live, which is a different structure entirely, but worth knowing exists.

Then look at where those links actually sit. Not the domain rating. Whether the site ranks for the things its own business cares about.

If you are evaluating two agencies and the only real difference between their pitches is the domain rating number they quote, you are not comparing quality. You are comparing who is more comfortable quoting a manipulable metric. Domain Rating is Ahrefs' own logarithmic score based on a site's backlink profile, and because it is calculated from links it can be inflated with a Fiverr service for about five dollars.

For what it is worth, our own published minimum is a domain rating of sixty with a five thousand monthly organic traffic benchmark, and the average of what we actually place lands closer to seventy or seventy-five with thirty to fifty thousand visitors. I am giving you both numbers deliberately, because an agency quoting only its ceiling is quoting the exception, and an agency that will not tell you its floor does not have one.

Quality metric Stated floor Delivered average
Domain rating DR 60 DR 72
Monthly organic visitors 5,000 40,000

The floor is the promise. The average is the product.

Our thinking on where that bar should sit is in our piece on high authority placements.

Are they testing on themselves, or on you?

A question I like: do they research and test these things themselves, or is your site going to be the test?

There is nothing wrong with an agency learning. There is something wrong with an agency learning on your budget without telling you. If they can point to their own research, their own site, their own experiments, you are buying informed judgement. If every answer is theoretical, you are funding their education.

Related, and often overlooked: find out who you will actually be working with. I prefer to work with the founder of an agency where possible. With a larger agency that is impossible, and that is fine, but you want someone involved rather than someone who hands you off. The person selling to you has just spent hours learning your business. If they disappear after signature, that knowledge goes with them.

What are the seven questions to ask an SEO agency?

The short version: ask about link relevance, commercial-page authority, buyer intent, AI visibility, demo-driving pages, competitor AI presence, and pipeline reporting. In my experience running EMGI, those seven expose more in one call than a 38-question checklist ever will.

I'd ask these before hiring anyone in this industry, including us. They aren't trick questions, they're diagnostic. A good agency answers every one confidently and specifically. An agency still optimising for the old playbook, and the wrong scoreboard, will squirm.

1. What percentage of our links come from publications our ICP actually reads?

This question exposes how your agency thinks. Are they chasing domain rating and link volume, or are they building semantically relevant links that help Google and AI systems understand your site? A DR70 domain looks good on a slide. What actually matters is whether your buyers would ever read the page the link sits on.

Here's a concrete example. Say you sell therapy practice management software to licensed psychologists running their own small practices. So where do those buyers actually live? What do they read? Psychology Today, for a start. A good answer names publications like that and connects them to your ICP.

The red flag: an answer built entirely on DR thresholds and monthly link counts, with no mention of who reads the sites. That usually means they're building links for their own internal dashboard, not your bottom line.

2. Which of our commercial pages have their own links, not just the homepage?

Most agencies default to building links to your blog, because it's easier, or to hammering the homepage over and over. But your homepage isn't always where buyers land or convert. Your use case pages, integrations pages and feature pages need authority signals too.

Back to the therapy software example. That company might have an insurance billing page, a clinical notes page, a practice management software page and a use case page for clinical psychologists, each optimised for one specific keyword. That page infrastructure builds topical authority and puts the right page in front of the right buyer. But unless you're a huge established brand, those pages won't rank just because you launched them. They need links.

These are also the pages LLMs reference, and where a buyer reads, understands your solution, then books a demo or starts a trial. A good answer maps links to specific commercial pages: "this quarter we're building authority to your integrations hub and the psychologist use-case page, and here's why those two". The red flag is a link profile where everything points at blog posts and the homepage.

3. Are we optimising for buyer intent or just search volume?

Are the people searching your target terms buying, or just learning? Here’s keyword data I pulled from DataForSEO for the video, from the practice management category. "Practice management software" gets 6,600 searches a month at a keyword difficulty of 74. "Therapy notes software" gets around 260 searches at difficulty 65. "Simple Practice alternatives" gets roughly 90 searches at a difficulty low enough that two or three backlinks can break into the top positions.

That 6,600-search head term sounds amazing in a pitch deck, but you'd be fighting category incumbents with serious money behind their brands, and most of those searchers are learning, not buying. Don't read it as 6,600 ready buyers.

Now look at a term like "best practice management software for psychologists". A junior strategist skips it for being too small. But that searcher has their credit card in hand. And someone searching "Simple Practice alternatives" already knows the category and believes something better exists. That is quite literally some of the most qualified traffic on the internet.

A good answer sounds like: "we'd start with the alternatives and best-X-for-Y terms your buyers actually convert on, and earn the head term later once the commercial pages have authority".

The red flag: a strategy deck sorted by search volume, top to bottom.

4. How does our brand show up in AI answers?

Most agencies can't answer this, and the gap tells you everything. LLMs don't just pull from an index of pages the way Google does. They lean on training data and on patterns of which brands are consistently mentioned and consistently credible in a topic.

Getting recommended comes down to context and repetition: being mentioned, repeatedly, on surfaces that are topically relevant to your category. Your Ahrefs DR doesn't matter here. Whether you're actually visible in the contexts that matter does, and that's much harder to fake.

This gap is measurable. At EMGI we studied 137 SaaS companies across 1,486 buying queries in 18 categories to map who actually gets cited in AI answers, and the difference between visible and invisible brands came down to relevant off-page presence, not domain rating.

When you ask this question, you're really asking something bigger: do you understand that buyer behaviour has shifted, and that the 2018 playbook can't work anymore? A good agency says "here's what we're doing about it: a monthly prompt panel on your buying queries, share-of-voice tracking against named competitors, and a map of which sources the AI answers actually cite so we know where to earn mentions". Because this is a newer topic, weaker agencies haven't rehearsed an answer. Vague hand-waving, or "AI search is just SEO", is your red flag.

5. Which organic landing pages actually generate demo requests?

This tests plain commercial competence. Does your agency know which pages drive demos and trials, versus which pages just drive traffic? Some of that data won't be immediately accessible to them, which is fair, but that's exactly why the conversation needs to happen. You should be identifying the lead-producing pages together and pointing the campaign at them.

A good answer sounds like: "these three pages produce most of your demo requests, so this quarter's content and links are aimed at them, and here's how we'll track it". We've broken down how to connect link building to revenue in our guide to SaaS link building ROI if you want the attribution mechanics.

The red flag: an agency that talks about sessions and rankings but has never asked which pages convert. If they don't know what a demo is worth to you, how can they prioritise anything?

6. Where are our competitors showing up in AI search that we aren't?

A serious agency in 2026 monitors your share of voice in AI answers against named competitors, not just your keyword rankings against them. If ChatGPT, Perplexity and Google's AI answers recommend three rivals on your money prompts and never mention you, that's invisible pipeline leaking every single day. Would you accept not knowing that?

The surfaces doing the recommending might surprise you. Across the same 1,486 B2B SaaS buying queries, our research found Reddit appears on Google's AI-enhanced search page 81.6% of the time, rising to 94.1% on bottom-of-funnel "best X" queries, with Google's AI answer directly quoting a Reddit thread in 12.1% of queries.

This is why directories, reviews and Reddit matter as off-page work now. They shape how machines perceive your brand, much like they shape how Google perceives it. A good answer includes a competitor gap analysis across AI surfaces and a plan to close it. The red flag is an agency that has never run one, because you can't close a gap you've never measured.

7. What does your monthly report tell us about pipeline?

The most important question, and ideally the last one you ask. To be fair to agencies: it's not their job to single-handedly increase your revenue, and I'd argue it's misleading when one promises "we'll add X leads". But they should be working towards pipeline, and the report should show commercial understanding: marketing qualified leads, sales qualified leads, demos booked, and how the work connects to them.

What you'll more likely see is "we raised your domain rating, built X links, you rank for Y more keywords". By themselves those are vanity metrics: the agency reporting what they did rather than the impact they produced. The CFO finds out eventually if SEO spend isn't producing organic inbound. A good answer treats reporting as a shared conversation between agency, in-house team and leadership: how is this work influencing pipeline, how many more demos are we booking? The red flag is a report you could swap onto any other client's letterhead.

How they answer the measurement question

That seventh question deserves a fuller answer, because how an agency measures tells you almost everything.

A bad answer promises a number and a date. By month three you will have this much traffic, or this many demos, or this many signups.

A good answer explains how the work connects to pipeline, and is honest about where attribution gets murky. Here is how we measure ours.

Page-level keyword positions. If we build links to a specific page and that page moves, we can reasonably attribute the movement. This is actually easier for a link-only engagement than for a full service one, because we are not touching everything else at the same time.

AI citation tracking. We take a baseline set of buying prompts and re-run them over time to see whether the brand appears more often. This is newer and it is becoming a bigger part of what we report on.

Share of voice and trajectory. How frequently you appear in your category, and whether the direction is consistent. Not one month's number.

Inquiries from organic search. Are they happening, and are they increasing.

Whether the client stays. The most honest metric in the industry. Nobody keeps paying for months and years if it is not working, and retention is harder to fake than a screenshot.

Each of those has a ceiling, and an agency worth hiring will tell you where it is.

Metric What it shows What it cannot show
Page level positions in Search Console Whether the pages being targeted actually moved Whether that movement became revenue
AI citations against a baseline prompt set Whether you are named more often in AI answers Which citation influenced a purchase
Share of voice in your category Direction of travel over months Anything useful from a single month's reading
Inquiries from organic search A real pipeline signal Clean attribution across channels
Client retention The most honest proxy available Anything at all in the first quarter

The AI citation column is the newest of these and the one most likely to be sold badly. The citation gap study above is the evidence behind why we track it against rankings rather than traffic.

What you will not get from me is a revenue claim. I genuinely dislike it when agencies say they did SEO for a company and now that company has four times the revenue. There are so many factors feeding that number. The product changed, the sales team changed, the market moved, they raised money, they hired. Claiming the lot is not confidence. It is a category error, and anyone in the industry reading it knows that.

The question Worth hiring Walk away
How do you get customers? Asks, and remembers the answer next call Only ever asks about your SEO
Who do you turn down? Has real disqualifying criteria Has never met a budget they didn’t like
Can I see the links? Shows samples, unprompted Quotes a domain rating average
Have you tested this? Tests on their own properties first Your site is the first test
How will we measure it? Explains how links reach pipeline Promises traffic by month three
How would you know it failed? Has an answer ready Claims 4x revenue from SEO alone

One of those answers is worth knowing in advance: most agency inventory comes from exchange networks, and what a partner pool actually is tells you what you are really buying.

Which red flags should end the conversation?

Some answers shouldn't just cost an agency points on your scorecard, they should end the call. Why end the call rather than just dock points? In my experience the patterns below reliably predict a wasted budget. Note that they're behavioural patterns, not brands: plenty of decent people work inside broken agency models.

One of our own clients hit exactly this red flag before finding us. Guillaume Deschamps, the Constant Hire marketer you heard from earlier, has built links himself, so he knew exactly what he was hearing:

“We’ve had agencies telling us, oh, you should do 25 links a month. And it’s just never going to happen, and it’s never good.”

Guillaume Deschamps, Growth Marketing Manager at Constant Hire

Guillaume Deschamps, Growth Marketing Manager, Constant Hire

What sensible volume looks like instead: he went from 8 to 44 inbound sales calls a month on nine links a month.

Red flag What it usually means
“We guarantee page-one rankings” Nobody controls Google. Guarantees are a sales tactic aimed at buyers who don’t know that yet.
“We have secret techniques we can’t share” Either there’s nothing behind the curtain, or there’s something risky behind it. Both are your problem later.
Link reporting that’s only DR numbers They’re optimising their dashboard, not your revenue. Relevance never made it into the model.
No request for access to your Search Console They plan to report on their activity, not your outcomes. Real diagnosis needs your data.
12-month lock-in before any results exist The contract is doing the retention work the results should be doing.

That last row is why EMGI contracts include a 90-day walk-away clause. I'd rather earn month thirteen than have a contract enforce it. Any agency confident in its work can offer you something similar, so ask.

How much should an SEO agency cost?

In our experience, credible SaaS SEO engagements mostly land between $3,000 and $15,000 per month depending on scope, with focused link building at the lower end and full-service strategy, content and links at the upper end. Anyone quoting a few hundred dollars a month is reselling volume work that fails the seven questions above.

So why do the cheapest and most expensive options fail for the same reason? Price reflects overheads and positioning, not whether the agency understands your ICP and pipeline. A $12,000 retainer producing irrelevant DR60 links wastes money faster than a $4,000 one earning relevant placements. Our pricing philosophy in one line: charge for work that can plausibly influence pipeline, and be able to show which work that is. For month-by-month expectations, see our SaaS link building cost guide.

Eight more questions for the call itself

If you take nothing else into the call, take these, alongside the seven diagnostic questions above.

  1. Why would we work with you?
  2. What kind of client do you turn down?
  3. Can I see a sample of links you have built, ideally for someone like me?
  4. What is your floor for quality, and what is your actual average?
  5. Do you test these approaches on your own properties, or will mine be the first?
  6. Who will I actually be working with after I sign?
  7. How will we measure this, and at what point should I expect to see something?
  8. What would make you tell me this is not working?

That last one is the one I would listen hardest to. An agency that has never considered how it would know it was failing has not thought very hard about how it succeeds.

A closing note on the awkward bit

Declaring the bias

I am aware of the position this puts me in. I have written a list of questions designed to make agencies uncomfortable, and I run an agency.

But the providers who are good at this want these questions, because they are the ones where the honest answer is also the impressive one. The providers who are not good at it want you to compare domain rating averages and look at a case study.

Print the questions out and send them ahead of the call if you like. If a single one gets a defensive answer, that is your signal. It is much cheaper to find out on a call than in month six.

One filter beats every question on this list: ask to see their actual work. We published a complete SEO audit example, run live on Cal.com, precisely so you can judge the thinking before you ever book a call.

And if you are still at the stage of deciding whether to hire anyone at all rather than run this yourself, that is a different question, and I have answered it separately in the honest maths on buying versus building.

You are also welcome to point every question here at us. What we actually deliver on a link building retainer is written out in public, including the monthly volume and the quality minimum, so you can check our answers against the list before you ever get on a call.

Want to see how your keyword mix, link profile and AI visibility hold up against these questions? Book a Strategy Call and we'll answer every one on the spot, live, in 15 to 30 minutes.

Frequently asked questions

What questions should I ask a link building agency?

Ask why you should work with them, what kind of client they turn down, for a sample of links relevant to you, their quality floor and their actual average, whether they test on their own properties, who you will work with after signing, and what would make them tell you it is not working.

Should I trust an agency's case studies?

Treat them as evidence, not proof. Every agency cherry picks its best results. Video case studies are harder to fake than written ones, but you still do not know the budget, the timeline, or whether the testimonial came with a discount attached.

Is domain rating a good way to compare agencies?

No. Domain rating can be inflated cheaply and says nothing about relevance. If the only difference between two pitches is the DR number quoted, ask instead whether the linking sites rank for the terms their own businesses care about.

How should a link building agency measure results?

Page level keyword positions on the pages being targeted, AI citation tracking against a baseline set of buying prompts, share of voice, and inquiries from organic search. Be sceptical of any agency attributing revenue multiples to SEO alone.

How long should I give an SEO agency to show results?

Think in quarters, not months. In our experience you should see leading indicators by month three: relevant links live, commercial pages moving on buyer-intent keywords, early impression growth. Revenue impact typically follows later. What you shouldn't accept is three months of activity reports with no evidence of direction.

Should an SEO agency guarantee rankings?

No, and you should treat any ranking guarantee as a red flag, because nobody controls Google's results. What an agency can commit to is process and transparency: which pages they're targeting, where links will come from, how progress maps to pipeline, and clear exit terms if the fit isn't right.

What is the 80/20 rule in SEO?

Applied to SEO, the 80/20 rule says a small share of the work drives most of the result. For SaaS, that usually means a handful of commercial pages targeting buyer-intent keywords, backed by relevant authority signals, will outproduce dozens of informational posts chasing search volume. Prioritise money pages first.

Matt Emgi is the founder of EMGI Group, a SaaS link building and AI visibility agency. This piece is drawn from a two hour conversation with George on the Rank Up podcast about SaaS off page SEO, and from the seven questions video above.

Matt Emgi, founder of EMGI Group

Written by Matt Emgi

Founder of EMGI Group, a SaaS link building and GEO agency helping B2B SaaS companies doing $1M to $30M ARR become the company buyers find, trust and shortlist: on Google, in ChatGPT, and everywhere in between.

Matt runs strategy on every account personally. There is no account manager layer, and every engagement carries a 90-day firing clause, because the work should earn the retainer. EMGI builds editorial links and brand mentions to the pages that drive revenue, with placements across sites like HubSpot, Zoom, Monday.com and BigCommerce at an average of DR 60+, and tracks AI visibility and share of voice for every client over time. Client retention runs above 90% past the first cycle.

He also publishes EMGI’s original research on how AI search picks brands, including a 137-company, 1,486-query study of AI citations, and his work has been published on VentureBeat. When he is not building links he is usually filming, training Muay Thai, or surfing badly.

Watch Matt on YouTube, follow him on LinkedIn, or read more about Matt and EMGI.