The tool question is the easy one. An in-house SEO stack covers six jobs, four of those jobs have a free option that is honestly good enough for a site under a few hundred pages, and the whole thing lands somewhere between nothing and roughly one contractor month. Teams still spend three weeks on the shortlist. Then they buy Ahrefs and Semrush, pay twice for the same backlink index, and still have no answer to the only question that matters on Monday morning, which is what to publish.
This is the stack we would build for a one-to-three person B2B SaaS team, organized by job rather than by brand, with the free option and the upgrade trigger for each. Then the part the shortlist never covers: the three jobs no software does, and the point where the stack genuinely is enough.
6
jobs an in-house stack has to cover
4
of those six have a free option that works at small scale
$150K+
loaded annual cost of the person using it, salary, benefits and overhead
Buy against the six jobs, not the brand names
Every SEO tool sells itself as a suite, which is how you end up with three products answering the same question and nothing answering the others. Write the six jobs down first, then fill each one exactly once. Anything a tool does that another tool already does is not a feature you are buying, it is a line item you are paying for twice.
| The job | Free or near-free | Paid options | Upgrade when |
|---|---|---|---|
| 1. Crawl & technical audit Find what is broken, blocked or orphaned | Screaming Frog free tier (500 URLs), Search Console's Pages report | Screaming Frog licence, Sitebulb, or the site-audit module inside Ahrefs and Semrush | Your site passes 500 URLs, or you need scheduled crawls and JavaScript rendering |
| 2. Keyword & competitive research What is being searched, and who owns it now | Search Console query export, Google Keyword Planner, your own sales-call notes | Ahrefs or Semrush. One of them. Not both | Day one, honestly. This is the one job where paid data earns its money |
| 3. Performance data What actually happened | Search Console plus GA4. This is the source of truth and it is free | A dedicated rank tracker, usually already bundled with job 2 | You need daily positions on a specific set of money terms. Weekly is fine for most SaaS |
| 4. Content briefs & on-page What goes in the piece before anyone writes it | A brief template in Docs, built from the top ten results and your Search Console queries | Clearscope, Surfer, Frase, MarketMuse | You publish four or more pieces a month, or briefs go to freelancers who need guardrails |
| 5. AI visibility tracking Whether ChatGPT, Perplexity, Gemini and Claude name you | A fixed prompt set run monthly by hand, logged in a sheet, plus our free AI visibility checker for the page-level signals | Profound, Peec AI, Otterly, Semrush's AI toolkit, Ahrefs Brand Radar. On WordPress, CiteTrack AI tracks it from the dashboard you already run; on any other stack, Pageoptimized. We build those two | Logging the prompt set by hand takes longer than a morning, or you need engine-by-engine share of voice for a board deck |
| 6. Reporting Getting the numbers in front of the people who fund you | Looker Studio, connected straight to Search Console and GA4 | Swydo and similar reporting platforms, worth it when several stakeholders need a report on a schedule | You are rebuilding the same view every month, or reporting eats half a day |
One disclosure on row five, because it changes how you should read that recommendation. We build two of the tools in it: CiteTrack AI, which does the tracking inside WordPress, and Pageoptimized, the platform for every other stack. So we are not neutral about the category. The advice we would give anyway is to run the prompt set by hand for a quarter first. Reading forty raw answers teaches you more about which sources build the answers in your category than any dashboard summary will, and you will know exactly what you need before you pay for it.
Three budget tiers, and what each one actually buys
List pricing moves constantly, so treat these as bands rather than quotes and check current prices before you commit. The shape of each tier matters more than the number.
- Under $200 a month. Search Console, GA4, Looker Studio, the free Screaming Frog tier, and a manual prompt log. That covers four of the six jobs. It works on a site under roughly 500 URLs with one person publishing, and plenty of B2B SaaS teams should stay here longer than they do.
- $500 to $1,500 a month. One all-in-one research platform, a Screaming Frog licence, and a content optimizer once volume justifies it. Most Series A and B teams land here, and it is enough to run a real program. The failure mode at this tier is buying the second research platform.
- $2,000 and up. Adds a dedicated AI-visibility platform and enterprise crawling. Only worth it when somebody owns the output daily. A $2,000 stack watched by someone with six hours a week produces less than a $200 stack watched by someone with twenty.
The stack tells you what happened. It does not decide what to publish, and it cannot earn you a mention on a page you do not own.
The three jobs no tool does
This is where in-house programs actually stall, and none of it shows up on a shortlist.
Deciding what to publish
Every research tool returns volume, difficulty and a gap list. None of them returns the question worth owning for your revenue segment, because that judgment needs your pricing, your win rates and the reason deals get lost. A gap list sorted by volume reliably points at traffic your sales team cannot use. The method we teach for this is prioritizing by revenue instead of volume, and it is the first module of the free SEO and AEO Academy, which is built for exactly this reader.
Earning the citations
Across the 101 buyer questions we ran through four engines, Reddit appeared in the cited sources for 96% of them and YouTube for 81%, and the engines cited 1,811 unique domains in total. Almost none of those pages are yours. A tracker can show you that a G2 comparison or a “best tools” roundup decides the answer in your category, and then it stops, because getting into that page is outreach, review generation and original data rather than software. That work is the off-page playbook, and it is the slowest part of the job.
Surviving month three
The stack does not create capacity. Month one is the audit and everyone is excited. Month two ships three pieces. By month three your SEO owner is inside the website redesign, the launch and a deck for the board, and publishing drops to one piece with no off-page work behind it. Rankings do not fall off a cliff, they just quietly stop moving, and the renewal comes up a year later against a flat chart. If you protect one thing, protect the calendar, not the budget.
When the stack is enough
We sell the alternative, so read this with that in mind. The honest answer is that in-house works well under a specific set of conditions. You have one person whose SEO time is genuinely ring-fenced rather than borrowed. Your site is small enough that technical debt is a weekend, not a quarter. Someone in the building can write with real product knowledge. And your buyers still mostly search rather than prompt, which you can check in an afternoon by running your category's buying questions through the four engines and seeing whether anyone in your space gets named at all.
When those hold, the $200 tier plus discipline beats most retainers, and you should not be reading agency comparison pages. When they do not, adding tools will not fix it, because the missing input was never the software. The version of that argument with the numbers in it sits on our in-house comparison page. If the question you are really weighing is whether an AI-visibility platform can replace the work rather than measure it, that one is here. And if the plan is to hire someone to run this stack, the real requirements for that role are in how to hire a Head of SEO for B2B SaaS.
Your first 90 days with the stack
- Week 1. Connect Search Console and GA4, build one Looker Studio view, and run a full crawl. Fix nothing yet. Write down what is broken and rank it by revenue impact.
- Weeks 2 to 3. Pull the Search Console query export and read it against your sales notes. Pick the ten questions where a good answer would help a deal, not the ten with the most volume.
- Week 4. Run a fixed prompt set across ChatGPT, Perplexity, Gemini and Claude and log every cited source. That sheet is your off-page target list, and it is worth more than any dashboard you could buy this quarter.
- Weeks 5 to 10. Ship against the ten questions, and start the off-page layers in parallel: review generation and real community presence. Both are slow, which is why they start now instead of later.
- Weeks 11 to 13. Re-run the same prompt set and compare the cited sources, using the method in how to measure AEO. Citation share moves here well before anything moves in sessions.
YOU DID IT RIGHT IF
By day 90 you have one Looker Studio view your team actually opens, ten questions chosen against revenue rather than volume, at least three pieces shipped against them, and two prompt-set runs you can compare.
- PRO TIP
Do the week 4 prompt run before buying anything in tier two or three. The cited-source sheet tells you which tools you actually need, and it regularly talks teams out of a purchase.
- WATCH OUT
If week 5 arrives and the ten questions are still being debated, that is the month-three stall arriving early. Pick the ten in one meeting, timebox it, and revise next quarter with data instead of opinions.
That sequence is a compressed version of what we run for clients as The Citation Engine. The order of operations is the same whether you run it or we do, and the receipts for the version we ran are in the Swydo case study, where the brand became the most-cited in its category across every major LLM with zero outreach emails.
Frequently asked
What software does an in-house SEO team need?
Six jobs, filled exactly once each: crawling and technical audit, keyword and competitive research, performance data, content briefs and on-page, AI visibility tracking, and reporting. Four of the six have a free option that works at small scale, namely Search Console, GA4, Looker Studio and the free Screaming Frog tier. The one job worth paying for from day one is research, which means a single all-in-one platform such as Ahrefs or Semrush.
How much does an in-house SEO stack cost?
Three realistic bands. Under $200 a month covers four of the six jobs with free tools and works for a site under roughly 500 URLs. Between $500 and $1,500 a month adds one research platform, a crawler licence and a content optimizer, which is where most Series A and B SaaS teams land. Above $2,000 a month adds a dedicated AI-visibility platform and enterprise crawling, which only pays back when somebody owns the output daily.
Do I need both Ahrefs and Semrush?
No, and buying both is the most common in-house overspend. They answer the same questions from different indexes, so the second licence buys a slightly different number rather than a new capability. Pick one, learn it properly, and put the difference into content production or off-page work, which is where the results actually come from.
Do I need an AI visibility tool?
Not at first. Run a fixed set of your category buying questions across ChatGPT, Perplexity, Gemini and Claude once a month and log every cited source in a sheet. That teaches you which sources build the answers in your category, which a dashboard summary will not. Buy a platform when the manual log takes longer than a morning, or when you need engine-by-engine share of voice for reporting. When you do, the split is simple: on WordPress, CiteTrack AI runs it from the dashboard you already have; on any other stack, a platform like Pageoptimized. We build those two, so weigh the recommendation accordingly.
Can one person do SEO in-house for a B2B SaaS company?
One person can run a real program if their SEO time is genuinely protected, the site is small enough that technical work is measured in days, and someone in the building can write with product knowledge. What one person cannot do is cover technical SEO, content production, answer engine optimization and off-page citation work at full depth at once. Something goes thin, and it is usually off-page, because that is the slowest to show results.
When should an in-house team bring in an agency?
When the stack is full, the calendar is not, and the chart has been flat for two quarters. That combination means the missing input is capacity or a discipline nobody on the team has shipped before, most often AI search, and neither is solved by another subscription. If your SEO owner gets pulled into launches and redesigns every month, adding tools makes the problem more expensive rather than smaller.
