Programmatic SEO without the spam

Programmatic SEO has a reputation problem: earned by sites that turned keyword lists into thousands of empty pages. We do the opposite: fewer, better pages, each one built for a person with a question.

Our one test: Would we build this page if Google didn’t exist? If the answer is no, it doesn’t ship.

What Google has done about site quality

Fifteen years of updates point the same way: reward pages made for people, demote pages made for rankings, and judge the whole site, not just the page.

  1. Feb 2011Panda

    Demotes thin, low-quality and copied content, and judges quality across the whole site, so weak pages drag down strong ones.

  2. Mar 2015Doorway pages

    Targets many similar pages built to rank for variations (cities, keywords) that funnel users to the same place.

  3. Jan 2016Panda goes core

    Quality assessment becomes part of the core ranking system: always on, not an occasional update.

  4. Aug 2022Helpful content

    A site-wide signal against content made primarily for search engines rather than people.

  5. Dec 2022E-E-A-T

    Rater guidelines add Experience to Expertise, Authoritativeness and Trust.

  6. Feb 2023AI guidance

    Google rewards helpful content however it’s produced, but using automation to manipulate rankings is spam.

  7. Mar 2024Scaled content abuse

    Helpful content folds into core ranking. New spam policies: scaled content abuse, site reputation abuse, expired domain abuse.

  8. Jan 2025Rater guidelines

    Raters told to rate lowest the pages whose main content is auto- or AI-generated with little effort, originality or added value.

The problems we design out

Every common programmatic SEO failure, the risk it carries, and the guardrail we build in from day one.

Thin content

RiskPages with too little unique substance to deserve a ranking.

Our guardrailA minimum-data rule per template. Combinations below it are never generated.

Low quality site-wide

RiskQuality is judged across the site. A mass of weak pages pulls down your best ones.

Our guardrailLaunch in waves, measure each, prune or improve what doesn’t earn its place.

Doorway pages

RiskNear-identical pages for every city or keyword, all funnelling to one destination.

Our guardrailA page exists only if it has a distinct answer and a real offer for that query.

Duplicate content

RiskThe same text with a swapped keyword. Google indexes one and ignores the rest.

Our guardrailOne URL per search intent, similarity checks across pages, canonicals where needed.

Built for SEO, not people

RiskKeyword-first pages nobody would bookmark, share or return to.

Our guardrailEvery template is built around a user task: compare, calculate, choose, find.

AI content at scale

RiskMass-produced pages made mainly to rank: what Google calls scaled content abuse.

Our guardrailFacts come from data. AI assists; people own strategy, copy and final review.

Index bloat

RiskThousands of URLs crawled, few indexed: wasting crawl budget and trust.

Our guardrailSitemaps by page type, indexing tracked weekly, low performers merged or noindexed.

Borrowed authority

RiskParasite pages on someone else’s domain or expired domains bought to rank.

Our guardrailWe don’t do it. Pages live on your site, about your business, under your control.

Good vs bad programmatic SEO

The template and the tools are the same. The difference is where you start and what you’re willing to not publish.

Bad programmatic SEO Good programmatic SEO
Starts fromA keyword listA customer need and data you own
What changes per pageA swapped city or product nameThe data, the answer and the recommendation
How many pagesEvery combination possibleOnly combinations with demand and enough data
SubstanceAI paragraphs as paddingTables, calculators, comparisons, expert notes
Launch50,000 pages on day oneWaves of a few hundred, measured
Internal linksOrphans found only via sitemapHubs, related pages and breadcrumbs
Success metricPages indexed, raw trafficLeads, sales, sign-ups, tasks completed
After launchSet and forgetRefresh data, improve, merge, prune

Same keyword pattern, two very different pages

The doorway page

Best Plumber in {City}
Same 400 words on 3,000 city pages
Call now
Same phone number, same offer, no local facts

The useful page

Water heater replacement cost in {City}
Typical local range$: $
Local permit rules, prices and options from real data
Estimate your cost
Get quotes from installers who serve {City}
Updated monthly · only for cities you serve

Most pages never get built, on purpose

Before anything goes live, the keyword universe goes through demand, data and quality gates. What’s left is worth indexing.

12,000
possible keyword × modifier combinations
4,100
have real search demand
1,700
have enough data for a genuinely useful page
1,150
pass our template quality checks
6 waves
launched, measured, improved or pruned

Illustrative example of how a keyword universe narrows before anything is published.

A strategy built around your business

We don’t sell a page count. Every engagement starts from five things only you can tell us, and ends with a plan that fits them.

Your customers

What they’re deciding, what they search, what they need to see to act.

Your data

Catalog, pricing, locations, usage or expertise only you have.

Your offer

What you can really deliver, where, to whom, at what price.

Your platform

CMS, engineering capacity, brand and compliance constraints.

Your economics

Value of a lead, sale or sign-up, so we build what pays.

Your strategy

  • Shortlist of page patterns worth building
  • Template spec, module by module
  • Data plan: sources, gaps, refresh schedule
  • Launch waves, internal links and KPIs

How we use AI

AI makes careful work faster. It never decides what gets published.

AI helps with

  • Cluster keywords and intents
  • Clean, match and enrich data
  • Draft variations for human editing
  • QA checks across thousands of pages

People own

  • Strategy and which pages exist
  • Facts, sources and accuracy
  • Template copy and expert insight
  • Final review, sampling and pruning

What we will never do

  • Spin or keyword-swap text across pages
  • Publish unreviewed AI text at scale
  • Create city pages where you don’t operate
  • Ship pages with empty or placeholder data
  • Buy expired domains or rent other sites’ authority
  • Invent reviews, statistics or experts
  • Launch everything at once and hope
  • Make pages for crawlers that users never see

After launch: measured, pruned, improved

Publishing is the start. Every page type earns its place with numbers, or gets improved, merged or removed.

Indexed ratio

Share of each page type Google chooses to index

Clicks per page

By template, not just site-wide

Task completion

Calculator use, comparisons, clicks to act

Conversions

Leads, sales and sign-ups per page type

Weeklyfirst 90 days

Indexing, errors, early signals

Monthlyongoing

Performance by template; improve / merge / prune list

Quarterlyongoing

Data refresh, template upgrades, new patterns

Score your own templates with the free quality checklist, read programmatic SEO and Google’s spam policies, or request a proposal.

Frequently asked questions

Is programmatic SEO against Google’s guidelines?

No. Google’s spam policies target pages made primarily to manipulate rankings and large volumes of low-value pages, not templates or data-driven pages. Travel, real estate, marketplaces and software directories have used programmatic pages for years. What matters is whether each page helps the person who lands on it.

Will AI-generated content get our site penalized?

Google says it rewards helpful content however it’s produced. Using automation to mass-produce pages with little value is what violates the scaled content abuse policy. We use AI for research, data work and drafts: the facts come from data, and people review what’s published.

How do you avoid doorway pages?

A page only exists when it has a distinct answer and a real offer for that query. No location pages where you don’t operate; no near-identical pages that all funnel to one form.

How do you prevent duplicate content?

We map one URL to each search intent, merge combinations that would answer the same question, check similarity between pages before launch, and use canonicals where variants must exist.

How many pages should we launch?

As many as pass the quality gates, usually far fewer than the keyword list suggests. We start with a pilot wave, read the data, then expand what works.

We already have thin programmatic pages. What now?

We audit them by template: improve the ones with potential, merge overlaps, noindex or remove the rest. Because quality is judged site-wide, a smaller set of strong pages usually serves you better than a large set of weak ones.

Can you guarantee rankings or traffic?

No one honestly can. We give ranges with the assumptions shown, measure from the first wave, and report on outcomes that matter to you: leads, sales and sign-ups.