I sell link building, so read the next sentence with that in mind.

If I launched a SaaS tomorrow, I would not build links myself. I would pay someone else to do it. And I would be paying them for their relationships, not their labour.

I know how that sounds coming from me. It is the single most self-serving thing I could say. But I have watched enough founders try the other route to know how it usually ends, and the honest version of this argument includes the parts that make agencies look worse, not better. So here is the whole thing, including the numbers we do not normally publish.

What you are actually buying

The mistake is thinking you are buying labour. You are not. Sending emails is the cheap part, and you could hire someone to do it next week.

What an agency has that you do not is a web of existing relationships. Other companies it already trades with. Link slots it has already bought in bulk, often twenty, fifty or a hundred at a time. Partners who owe it placements and partners it owes placements to. That web took years to build and it is the entire product.

The agency margin is the price of not spending those years yourself. That is the whole pitch, and if a provider cannot explain their inventory to you in those terms, you are probably paying agency rates for someone doing manual outreach badly.

Why any of this still matters in 2026

Worth addressing the obvious objection before going further, because plenty of founders now assume links stopped mattering the moment buyers started asking AI instead of Google.

The opposite is closer to the truth, and there are two mechanisms.

The first is training data. When a model is trained, it ingests the corpus of the web. Being mentioned across articles teaches it which companies are associated with which problems, and models can parse the HTML, see where links point, and see which companies get referred to consistently.

The second is live retrieval. When a model does not already know the answer, it searches, and then cites what it finds. Ahrefs has a clear explainer on how retrieval-augmented generation decides what to pull in if you want the mechanics.

What makes this concrete is where the retrieved sources actually come from. Ahrefs analysed 1.9 million citations across 1 million AI Overviews and found that 76.10% of AI Overview-cited pages rank in the top 10 organically. Our own SaaS AI Citation Gap Report found the same relationship from the other direction: across 150 SaaS companies, ChatGPT citations correlated with Google keyword rankings at r = 0.76, while the correlation with organic traffic was only 0.23.

So links are not buying you rankings as an end in themselves. They are buying the ranking position that makes you retrievable.

The asset problem, which is where in house usually dies

Here is what kills most in house programmes in the first quarter.

Every link is a trade. Not always a cash trade, but always a trade. So before you send a single email you need to answer one question: what do I have that the other side wants?

Most SaaS companies starting out have nothing. The website has a domain rating of ten or twenty. It might be getting a few hundred clicks a month in Search Console, which Ahrefs will show as somewhere between zero and two hundred, because Ahrefs consistently underestimates real traffic at that end of the scale. There is no newsletter with an audience. There is no blog anyone reads.

You are asking strangers for something valuable and offering nothing in return. That does not work any more, and it is worth being precise about why.

The old free tactics are dead, and the reason is arithmetic

You have read the advice. Find a broken link on someone's resource page and offer your article as a replacement. Find an unlinked brand mention and ask them to make it a hyperlink. Point out that you published something better than the page they currently cite.

Every one of these now fails for the same reason. The person on the other end knows exactly what a link is worth. Somewhere between one hundred and four hundred dollars in perceived value, and they are being emailed about it constantly. We go into the full range in our breakdown of what SaaS link building costs across retainer tiers.

So the actual proposition you are making is: please do me a favour worth several hundred dollars, for free, because I found a typo on your website. We are all busy. Nobody is doing that for a stranger.

Templated outreach is finished as a strategy, and I would feel fairly confident saying almost nobody is running it as a genuine approach any more. What replaced it is trade.

What a real trade looks like

Once you accept that every link has a price, the question becomes what currency you pay in.

The cleanest version is something that costs you little and is worth a lot to them. A slot in your newsletter. A quote from your founder for their article. A guest appearance. A feature in a roundup you were publishing anyway. We have written separately about which assets are actually worth building for this.

Guest posts used to be the default currency here and are much harder to come by now, because SaaS companies stopped outsourcing their content. That shift is worth understanding on its own terms, and we covered it in our piece on where guest posting stands.

This podcast is an example of the mechanism. George and I are talking because the conversation is worth having. If we each add a link afterwards, no money changed hands and both sides got something they wanted. That is not a hack, it is just what a relationship produces.

The problem is that it does not scale. You can do a handful of these a quarter. You cannot do thirty a month.

How the scaled version actually works

At volume, the trade becomes structural, and this is the part that rarely gets written down.

Direct swaps do not work. If site A links to site B and site B links straight back, the pattern is obvious. Do enough of them and the only links you have are from the same sites you link to, which is plainly unnatural. Set aside how much you think Google penalises it, it is simply not scalable and it is not the best use of the links you have to give.

So the third site enters. Site A links to site C. Site B links to site A. There is never a visible relationship between the two companies actually trading. That is the ABC exchange, and it is how a large share of SaaS links are placed today, by agencies and by any in house team big enough to run one.

The economics get more interesting once your own site has weight. If you are a genuine powerhouse, a domain rating of seventy or eighty and twenty to fifty thousand visitors a month, you stop trading one for one. You can ask for two, sometimes three, in return for every link you give.

Give ten links a month from a site like that, contextually and only where it genuinely fits, and you take twenty to thirty back. Ten outbound links a month on a large content operation raises nobody's eyebrows. Twenty to thirty inbound links a month is more than competitive in most SaaS categories unless you are fighting at the very top.

That is the machine. And you can see immediately why it is closed to a company with a new website: you have nothing to feed it.

The thing nobody tells SaaS founders about their own blog

There is a corollary here that most founders discover far too late.

Once your blog has real authority, other people will pay you for links on it. Not a small amount either. Plenty of SaaS companies quietly make two, three, four thousand dollars a month in almost pure profit from placements on their own content.

I am not telling you to do that. In a lot of cases you should not, and if you do you want a written policy about who you will and will not link to. But you should at least know the asset exists, because the moment you know its price, every inbound request stops being a nuisance and starts being a negotiation. A paid placement becomes an exchange the second you have something to offer back.

So when does in house actually make sense?

It does, in two situations.

The first is scale. Companies past roughly forty or fifty employees tend to have the content operation, the domain authority and the headcount to run their own exchange programme properly. At that size the maths flips, because you have inventory.

The second is a dedicated hire. If you have an in house SEO with genuine link building experience, this is a reasonable thing for them to own, and over time you will save money on every placement.

What almost never makes sense is a founder doing it. Yes, you will eventually shave the agency margin off your link costs. But you will spend a year of your attention learning a market you do not otherwise need to understand, and founder attention is the most expensive input in the business. It is not a high return activity for you specifically.

The part where buying looks bad

Two honest caveats, because a piece that only argues one way is not worth reading.

Second, the quality range is enormous. There are sites selling placements at fifty to a hundred dollars, often out of India, and there is a reason they are cheap. Agencies buy in bulk at that end and the results are exactly what you would expect. Our guide to white hat approaches that still work covers the other end of that range.

And an agency has an incentive you do not share. Volume is easier to sell than judgement. The good ones turn business away when the fit is wrong. The rest will take anyone, which brings us to the question you should actually be asking, which is not whether to buy but who from. That is the subject of a separate piece on telling a good provider from a bad one.

The mechanics of that third site, the partner pools and the ledger you need to keep are covered in full in the piece on reciprocal links and three way trades.

How to decide

If you are a founder about to do this personally in the evenings, the first row is you. Go and publish something worth linking to, then revisit this in six months.

And if you land on buying, you should know what you are buying before you pay for it. How we structure SaaS link building engagements is built on exactly the exchange model described above, with the quality floor stated up front rather than after you sign.

I have a client in the HR space we have worked with for over two years, almost entirely building links to their service pages. They could have run that in house. They chose not to, because the two years of relationship building was the thing they were buying, and they wanted those two years back.

Frequently asked questions

Is buying backlinks against Google's guidelines?

Paying for links that pass ranking signals is against Google's guidelines. In practice most of the SaaS market runs on paid placements and exchanges, and enforcement targets obvious footprints rather than individual placements on real sites. Judge the risk on the quality of the site, not on whether money changed hands.

How much should I expect to pay for one backlink?

Between $300 and $800 for a negotiated placement on a relevant SaaS site is typical. Bulk supply runs $50 to $100 and is cheap for a reason. We have been quoted $3,000 for a single link and mention, and we turned it down.

At what size does in house link building make sense?

Roughly forty to fifty employees, or earlier if you hire a dedicated SEO with link building experience. Below that you usually lack the domain authority and content inventory to trade with, which is the actual barrier.

What is an ABC link exchange?

A three way trade. Site A links to site C and site B links to site A, so the two companies actually trading never link to each other directly. It avoids the obvious footprint that direct reciprocal swaps leave behind.