When I write a SaaS homepage, I start with the smaller sections and leave the H1 until those are in place.
That first line gets endless debate because everybody can have an opinion about six words without having to understand the product. A visitor gives it maybe two seconds. It needs to make sense, but it is not going to carry the buying decision by itself.
I’d work out the rest of the page first. What is the product, where does it fit, what can a buyer see or try, and what would make them trust the answer? The smaller sections have to walk them through that in an order that makes sense. If they don’t, a clever H1 is a fancy hat on a confused page.
Once the page has a shape, the H1 becomes a smaller job. Say what the product is in language a buyer will recognise, then spend the saved afternoon on a section that explains why anyone should care.
Here’s my Halloween horror story. It has a new CMO, a £20k website rebuild and an industry of about 400 people in the UK.
Most deals for this enterprise SaaS company came through relationships. Rebuilding the website was the new CMO’s first move, and the shiny new site achieved nothing. Nobody grew the market beyond those 400 people because the navigation changed.
I can understand wanting a better site. I cannot understand spending that money before asking how this particular market buys. With about 400 people in the industry, I’d ask whether any of them was struggling to find what they needed on the existing site.
A rebuild gives a new CMO something visible to show in a board update, but it doesn’t enlarge the market or replace the relationships that bring in deals. If nobody can name the problem with the existing site, I wouldn’t approve £20k to replace it, even if the old one looked like it had been designed in a crypt.
I don’t want a content score telling me whether my opinion is good.
Surfer gives you a content score, Jasper tries to model your brand voice, and Semrush suggests topics. They can organise a workflow, but I can’t see how any of that tells me which argument your buyers are tired of hearing or which bit of conventional advice is rubbish for your market.
For the kind of short B2B SaaS writing I care about, I’d open Claude or Codex with a rough opinion, the audience, and the claims I need to check. Then I’d push against whatever came back until it kept the awkward detail instead of smoothing it away. A score-led workflow doesn’t give me that kind of control.
I’d build a few reusable skills of my own for content research, SEM research and voice rules. If a post needs an image or video, I’d make one for that too. Then I’d send the finished piece to a CMS I can stand using and have it publish on schedule. I’d expect better posts from that than from a standalone writing tool because I can change the research or the rules as soon as they’re getting in the way.
That does require having an opinion in the first place. The assistant cannot supply a client story you haven’t told it or decide which product annoyed you in real use. Feed it generic notes and you’ll get generic copy, however nice the interface is.
I can see why the packaged tools were attractive when AI writing was new. Now, for a small team, they look increasingly like the Grammarly of B2B content marketing. They tidy the surface while the interesting decision stays with the writer. For a piece of writing, I want control over the argument, not another dial to turn green.
If I have a developer to help build a B2B marketing site, I want Payload. I’ve used every tool in this list, and what I care about most is who can change the site without breaking it.
Payload lets a developer set up the content structure and give editors an actual admin login. Permissions can be limited by role, so a content contractor can work on the words without getting the keys to everything else. It still needs a developer to build it properly, which is fine with me. I’d rather give them that work than ask a marketer to become one on a Friday afternoon.
For a personal blog, I’d use Ghost or Hugo. Ghost gives me a publishing interface, while Hugo lets me keep posts in Markdown and publish a static site. I use Hugo for this blog because I just want somewhere to publish writing, and I don’t need a conventional CMS for that.
Webflow is my fallback if I cannot get developer time but still need to build and change a marketing site myself. Its CMS collections give you room to structure content. I’d use it in that situation and hate myself a little while maintaining it. Give me a developer and Payload instead.
WordPress is near the bottom. An intern with an unrestricted plugin drawer and a page builder can wreck your page load stats before anyone notices. A careful team can keep WordPress fast, but I don’t want every landing-page change to become a discussion about what got installed this week.
HubSpot is last. The CMS product is now called Content Hub, separate from Marketing Hub, and both sit around a CRM I find mediocre. I find the whole stack parasitic because another content need becomes another reason to buy further into the suite. HubSpot itself explains how to put its forms on an external site. I can keep the integration without giving it my website too.
This post isn’t sponsored by anyone, I just can’t stand HubSpot Content Hub. It sucks.
If somebody has agreed to discuss whether your software could fit their company, putting a bot in front of them first feels like a strange way to start a relationship.
An AI tour is fine when someone just wants to look around without booking a call. It can show where the settings live, and nobody needs a salesperson to narrate every menu.
The useful part of a real demo starts when the buyer says, “Our setup is a bit odd.” I want somebody to ask how odd, listen to the answer and work out whether the product can handle it. An agent may produce a fluent reply, but fluency is cheap if the answer misses the one exception the buyer cares about.
A buyer who explains their constraints to a bot and then repeats them to a salesperson has been given an extra task so the company can save a meeting. I would be irritated before the human even arrived.
I’d also rather hear “I don’t know yet, but I’ll check” than get a confident answer that skips the one integration or exception that matters. The awkward question is often the point of the demo.
Use the agent for a shallow tour if you like. When somebody wants to talk about fit, let them talk to a human being. They are deciding whether they trust your company as well as your interface.
YouTube is being slopified, and B2B software comparisons are an especially grim corner of it.
The usual comparison gives me two logos, a narrated tour of each product page, a few spotless screenshots, and a conclusion in which both tools are great for different needs. It rarely gives me a reason to think the person making it has used either product. I could read the two landing pages myself and save 15 minutes.
Paid placements make the trust problem worse, and YouTube’s paid-promotion label can tell me when a commercial relationship has been disclosed. It cannot tell me whether the creator has used either product, let alone found the awkward bits. AI can now turn those same landing pages into a plausible script, so the polish tells me even less than it used to.
If you’re going to call it a comparison, show both products doing the same job. I want to see the setup, where one breaks down, and which limitation would make you stop using it. Say if you were paid or given access. Someone who hasn’t used either product can still make an overview, but calling it a review would be a stretch.
I’d rather see a verdict I can disagree with than another conclusion that both tools are powerful. Even “neither can handle this workflow, so keep the spreadsheet” gives me something useful to test.
This comparison video is one of the few I’ve seen that doesn’t leave me wondering whether I might as well have read the landing pages myself.
Video makes sense when somebody works through a real task on screen, but I have no interest in an AI summary read aloud over a recording of the happy path.
If an impressions chart goes up 20 percent, what changes on Monday?
Before putting the impressions chart in a weekly meeting, I’d ask whether a rise would make us invest more in the topic or check whether the right buyers had found it. If nobody can name the decision, we’ve brought a graph to a meeting because graphs look good in meetings.
B2B SaaS marketing gets dressed up as a measurement problem far too often. An executive asks for more numbers as though the numbers will make the decision for them. The team adds MQLs, influenced pipeline, engagement, share of voice and a dozen other things it can pull from a dashboard. Soon the deck is 40 slides long and nobody has said what they plan to do on Monday.
If prospects keep describing your product the wrong way on calls, I’d pay more attention to that than to a rise in impressions. I’d listen for where they picked up the idea, fix the copy that’s giving it to them, and see whether the confusion stops. You won’t get a pretty chart out of the conversation, but you may find the problem.
In a weekly meeting I’d rather argue over which channel gets more effort or why buyers keep getting the wrong message. We can pull up the numbers that help settle it and leave the rest in the reporting tool. Otherwise we’re doing a group reading of a dashboard, and I would prefer almost any other book.
Sometimes the team hasn’t agreed on a decision, and another dashboard won’t settle that argument.
AEO is SEO. I realise that is a disappointing answer if you’ve just put AEO in a pitch deck.
From what I’ve seen, Google’s AI summaries for broad B2B searches read like a bad listicle glued to the top of the results. A definition, a few safe points, and links to pages saying much the same thing. Summaries of summaries of summaries. They might help someone learn the vocabulary. I wouldn’t trust one to tell me which software to buy or what to change in a marketing plan.
Even Google’s guidance for its AI search features says the SEO fundamentals still apply. It recommends accessible pages and useful, original material that gives people a reason to visit the source. The new acronym hasn’t revealed a secret second search engine.
An AI answer above the results might cost you clicks, and I’d track whether search referrals change. I wouldn’t approve a separate AEO budget without knowing what work it buys. If someone proposes making the site accessible and publishing material worth reading, they can call it SEO.
My bet is that practitioners will get bored with the generic summaries. When the decision matters, they’ll look for someone willing to make a call. A page that says when a tool is the wrong choice has something the summary cannot manufacture by compressing five pages of feature lists.