Google uses automated systems and many signals to find, understand and rank pages. It does not publish a fixed checklist with weights that an agency can manipulate. A responsible agency can improve the website, its content, its reputation and the evidence available to search systems. It cannot control the algorithm, competitors, demand or a promised position.
The useful buyer question is therefore not “Which 200 factors will you optimize?” It is “Which controllable constraints will you improve, what evidence supports the work and how will you verify customer value?”
Require an agency influence contract
This SEO Companies Hub contract replaces factor-count claims with controllable releases.
| Influence area | Agency can produce | Production proof | Cannot promise |
|---|---|---|---|
| Discovery/indexing | Diagnosed URL cohort and implementable technical change | Status, canonical, render, link and index checks on the real route | Crawling, indexing or ranking on a fixed date |
| Relevance | Intent map, canonical owner and clearer task experience | Query/page evidence and editorial acceptance | Exact weight for a title, heading or term |
| Usefulness/evidence | Original support, expert review and decision-ready page | Source ledger, method and reader-task test | That “more content” earns visibility |
| Authority | Legitimate assets, relationships and internal navigation | Relevant references and policy-safe acquisition record | Link quantity, DR target or placement guarantee |
| Experience | Faster, accessible and stable task path | Field/lab diagnostics plus functional QA | One page-experience score determines ranking |
| Learning | Versioned cohort, hypothesis and business outcome | Release record, comparison and review decision | Attribution from a simultaneous change package |
Contract on verified work and decision quality, never on a secret factor list or position.
Understand systems, signals and outcomes
Google's ranking-systems guide describes notable systems including language understanding, freshness, link analysis, original content, passage understanding, reviews and reliable-information systems. It states that systems generally operate at the page level while some site-wide signals and classifiers also contribute.
That distinction matters. A named system is not a knob in an agency dashboard. A signal is not necessarily a direct instruction to change one metric. A ranking outcome is the result of the query, index, page, context, competing results and current systems.
Treat claims about exact weights or secret formulas skeptically unless Google documents them. Even genuine observations may not generalize across queries and markets.
Influence area 1: discovery and accessibility
An agency can help ensure intended pages are reachable and technically functional. Work may include:
- logical URL and navigation architecture;
- crawlable internal links;
- accurate XML sitemaps;
- reliable status codes and redirects;
- robots and index directives;
- canonical consistency;
- rendered text and links;
- mobile behavior;
- monitoring after releases.
Google's guide to how Search works explains crawling, indexing and serving. If a page cannot be accessed or indexed as intended, relevance improvements may never be evaluated.
The agency should show affected URLs, the diagnosed failure, the proposed change, implementation owner and production verification. A generic audit score is weak evidence.
Influence area 2: relevance to the search task
Agencies can research audience questions, inspect result types, map a canonical page to each distinct task and improve how the page communicates its purpose.
Controllable elements include:
- page type and search-intent fit;
- title and main heading;
- direct answer and section structure;
- terminology and entity clarity;
- descriptive internal links;
- useful media and text alternatives;
- consolidation of overlapping pages.
Exact phrase repetition is not the objective. Google's ranking guide describes systems such as BERT, neural matching and RankBrain that help understand language and concepts. Write accurately for the audience and cover the necessary decision rather than forcing every wording variation.
Influence area 3: usefulness and evidence
An agency can help the client publish information worth using: original research, product data, expert explanations, comparisons with consistent criteria, tools and complete processes.
The client must supply or approve much of the evidence. Agencies should not invent experience, customer results or private metrics.
Useful controls include:
- primary and authoritative sourcing;
- clear authorship and reviewer roles;
- methodology and limitations;
- dates on volatile claims;
- separation of facts, interpretation and forecasts;
- updates and correction ownership;
- appropriate depth without padding.
Google's systems guide notes systems intended to surface original content and high-quality reviews. That does not create a formula; it reinforces the value of distinct, attributable work.
Influence area 4: internal and external authority
Internally, an agency can design links that help readers and search systems understand relationships between pages. Externally, it can create and distribute assets that relevant publishers, partners and communities choose to reference.
Legitimate work includes original research, useful public tools, expert commentary, partnerships and accurate media resources. The agency cannot guarantee editorial links from independent sites.
Avoid purchased ranking links, excessive exchanges, mass directory placement and manufactured authority metrics. A third-party domain score is a benchmark from that tool, not a Google ranking factor.
Evaluate outreach by relevance, editorial legitimacy, referral quality and durable awareness—not a link quota.
Influence area 5: page experience
Agencies can diagnose and improve loading performance, responsiveness, layout stability, mobile usability, intrusive overlays, security and clarity of the main content.
Google says Core Web Vitals are used by ranking systems, but achieving strong scores does not guarantee top rankings. Its page-experience guidance advises assessing the overall experience rather than optimizing one or two metrics in isolation.
Prioritize problems that affect real users and important templates. Chasing a perfect laboratory score after the page is already fast and stable may be less valuable than fixing an unusable form.
Influence area 6: structured information
Where Google supports a structured-data type and the visible page qualifies, an agency can implement accurate markup. It can also improve product fields, organization identity, author records, breadcrumbs and other data relationships.
Structured data can make content eligible for certain search features. It does not guarantee display or ranking. Markup must represent visible, truthful content and remain synchronized with the page.
Require validation, rendered inspection and an update owner. Do not buy “secret schema” or unsupported AI markup.
Influence area 7: measurement and learning
An agency can create a disciplined loop that connects deployment cohorts to search, behavior and business outcomes.
For each initiative, record:
- hypothesis;
- affected URLs;
- release date;
- leading technical or search signal;
- customer progress action;
- qualified business outcome;
- risks and comparison group;
- review and scale decision.
This does not control rankings. It improves resource allocation and makes agency claims auditable.
What an agency cannot control
No agency controls:
- Google's algorithms or update timing;
- whether a page is indexed or selected for a feature;
- competitors' investments;
- seasonal or macroeconomic demand;
- the searcher's location, device or context;
- news and result-layout changes;
- third-party editorial decisions;
- the client's product quality, price or capacity;
- an exact ranking date or position.
Google explicitly says it does not accept payment to rank pages higher. Any provider claiming special paid access to organic rankings is misrepresenting the system.
Evaluate factor claims with an evidence ladder
Classify recommendations:
Documented guidance
Supported by current Google documentation, such as crawl controls, canonicalization or Core Web Vitals.
Reproducible site evidence
Observed in a bounded client cohort with recorded changes and outcomes.
Industry research
Useful for hypotheses but subject to sampling, correlation and market limitations.
Expert hypothesis
A reasoned idea that still needs testing.
Unsupported assertion
An exact weight, guarantee or secret mechanism without verifiable support.
Agencies should label the level instead of presenting every tactic as settled fact.
Ask an agency for controllable deliverables
During selection, request:
- the priority page cohort and diagnosed bottleneck;
- proposed changes and why they fit the audience task;
- client and agency responsibilities;
- implementation access and rollback plan;
- evidence and policy boundaries;
- acceptance criteria;
- measurement definitions and reporting cadence;
- risks, dependencies and stop rules.
A strong answer may acknowledge uncertainty. A weak answer lists factors, tool scores and guaranteed gains without a release plan.
A practical buyer verdict
An agency can materially improve the conditions under which pages are discovered, understood, trusted and used. It can build better architecture, content, evidence, page experience and learning systems. Those are significant levers.
It cannot sell control over Google's rankings. Evaluate the provider on the quality of its diagnosis, the changes it can implement, the evidence it preserves and the business decisions its reporting enables.
Related decisions
- B2B SaaS Marketing Channels Compared: Where SEO Fits — the adjacent saas marketing decision.
- Customer Acquisition Channels Compared: Where SEO Wins and Loses — the adjacent channel strategy decision.
- Demand Generation vs Inbound Marketing: Differences and SEO — the adjacent demand generation decision.