Technical SEO tools and providers reward a written scoring rubric

What to take away

Technical SEO tools and providers reward a written scoring rubric. Most buying decisions fail at requirements, not the demo. The rubric keeps the process honest.

How should you define requirements before comparing technical SEO tools and providers?

Start with the audit work you already do by hand. List the checks you repeat each month: status codes, canonical tags, hreflang pairs, internal link depth, Core Web Vitals by template. Each check becomes a requirement line with a pass or fail definition.

Then sort requirements into three groups. Must-have items stop a purchase on their own. Should-have items affect scoring. Nice-to-have items are recorded but never decide anything. If everything is a must-have, nothing is.

Add a fourth column to every requirement: how you will test it. A claim such as crawling a million URLs within 24 hours is testable. Fast crawling is not. Deciding the test method at requirements stage stops the same argument being reopened in the demo.

Write the volume figures down too. Pages crawled per run, domains, user seats, retention period for historical data. A tool priced per crawled URL behaves very differently from one priced per seat. The gap widens as a site grows.

This is also where you decide whether you are buying software or a service. The two are not substitutes. Software gives you repeatability; a provider gives you interpretation. The selection questions differ enough that they deserve separate shortlists, which is the approach set out in technical SEO software selection in England.

Named tools split along the same line. Screaming Frog SEO Spider is a desktop crawler with a free tier capped at 500 URLs and a paid licence of about £199 a year per user. Sitebulb is a desktop crawler sold on monthly plans, typically £10 to £100 depending on URL allowance.

Lumar, formerly Deepcrawl, and Botify sit at the enterprise end. Both price per crawled URL on annual contracts, with four-figure monthly bills typical. OnCrawl and JetOctopus occupy the middle, adding log-file analysis to cloud crawling for mid-market sites.

Budget needs a range, not a number. A mid-market retailer might test three options between £300 and £900 a month, then compare that against a fixed-fee audit quoted at £6,000. Neither figure is a market rate; both are placeholders that force the trade-off into the open.

Get the requirements signed off by whoever holds the budget and by whoever will operate the tool daily. A list owned by one person is a wish list; one signed by two is a mandate.

What does a workable scoring rubric look like?

A rubric converts opinion into a number a colleague can challenge. Weight each criterion, score each vendor from 1 to 5, multiply, and total. Keep the evidence column so a later reviewer can see why a score was given.

Two rules make the table useful. Score only what you have seen, not what the roadmap promises. Require a second scorer for any criterion above 10% weight. A single enthusiastic reviewer cannot carry a decision alone.

Write scoring anchors before the first demo. Anchors describe what a 1, a 3 and a 5 look like for each criterion, in one sentence each. For export, a 5 might be full crawl and issue history available by API, a 3 a CSV export per run, and a 1 a PDF report only.

Without anchors, one reviewer's 3 is another's 5 and the totals stop meaning anything. Ties go to the option with the stronger evidence, not the lower price.

Accessibility deserves its place rather than a footnote. If a tool produces reports your own team cannot read with assistive technology, it will be abandoned quietly within a quarter. Public sector buyers should also be aware of duties under the Equality Act 2010 when commissioning content and reporting work that affects service users.

Weights themselves need a record. Note who agreed them and on what date. A weighting debate after the scores are in is really a disagreement about the decision.

Which technical SEO tools and providers fit which job?

Four shapes of supplier appear in most shortlists. Crawlers and audit platforms suit teams with in-house analysts who need raw output: Screaming Frog, Sitebulb, Ahrefs Site Audit, Semrush Site Audit and Ryte all fit here. Managed audit providers suit teams without spare capacity.

Log-file and analytics specialists, including Botify, Lumar and OnCrawl, suit sites with heavy crawl waste. Point tools cover one job, such as sitemap validation or structured data checks. Google's Rich Results Test and the free Schema Markup Validator are examples.

Match the shape to the gap. A team of two SEO specialists running a 40,000-page site rarely benefits from a second crawler. They usually benefit from someone who will interpret the first one's output.

Pricing models differ as much as the work. Per-crawl credits suit large sites audited in bursts, as sold by OnCrawl and Lumar. Per-seat licences suit one large site with a wide stakeholder group, the model Screaming Frog uses. Per-project fees suit a defined deliverable with a fixed end date.

Ask which model the supplier prefers, and why. It reveals who they expect to serve.

The table below compares named tools and providers on type, public price signal and best fit. Prices are signals as of 2026, not quotes: check current rates, because enterprise terms are negotiated.

For any tool that validates crawling and indexing, check it against the published specification rather than the vendor's summary. Google's guidance on how to build and submit a sitemap sets out the format and size limits that a validator should enforce, which gives you a fixed benchmark for scoring that feature.

Where the work touches content production, expect a skills question as well as a tooling one. The Chartered Institute of Marketing lists digital marketing training courses covering search and content disciplines. That is a reasonable proxy for the training market a supplier's team should describe.

How do you test a tool or provider before signing?

Run a paid pilot on one property, with a written success test agreed in advance. Thirty days is usually enough for a crawler such as Screaming Frog or Sitebulb. A consultancy pilot needs a defined deliverable, such as a prioritised issue list with effort estimates.

Use your own data, not the vendor's sample site. A demo on a fast marketing site tells you little about a faceted ecommerce catalogue with 2 million filter combinations.

Scope the pilot to one or two templates rather than the whole domain. Templates repeat, so a product page, a category page and a paginated archive usually expose every rendering and indexing problem you need to see. A full-domain crawl comes after you have chosen, not before.

Record three numbers during the pilot: time to first useful finding, proportion of findings you accepted as valid, and hours your team spent operating the tool. The third number is the one buyers forget, and it is often the largest.

If the pilot fails, stop. A failed pilot followed by a signature is the most expensive pattern in this market. The implementation sequence that follows a pass is covered in technical SEO tool implementation in England, including how to phase rollout across templates.

Where do procurement rules and supplier checks bite?

Public bodies in the UK must follow their own standing orders and, where applicable, the procurement rules that apply to the contract value. Private buyers have more freedom but the same failure modes: unclear requirements, weak evaluation records and no exit plan.

Due diligence is not a formality. Confirm the legal entity, filing history and registered address before you rely on a supplier's longevity. Companies House holds official company data that supports exactly this check, and it takes minutes.

Check the entity that will sign the contract, not the brand on the invoice. Group structures mean a parent company can hold the assets while a subsidiary carries the liability.

Ask where data is stored and processed, who the sub-processors are, and what happens on termination. Cloud procurement has well-documented pitfalls around lock-in and unclear terms.

The U.S. GAO's report on cloud computing procurement challenges describes how agencies have struggled with requirements and contract terms in this area. The lessons transfer to any buyer assessing hosted SEO platforms.

A short due-diligence checklist for any supplier holding crawl data:

Keep the file. If the supplier is acquired or changes its pricing model, the original evidence is what supports a renegotiation. The fuller vetting routine, including reference calls and financial checks, is set out in technical SEO vendor due diligence in England.

How should you compare evidence across suppliers?

Comparison is a data problem, not a taste problem. Keep one row per supplier and one column per rubric criterion, and store the raw evidence behind each score. A screenshot of a rendered page is evidence. A slide claiming rendering quality is not.

Store the rubric, the evidence and the scores in one place that survives staff changes. A shared folder with a naming convention beats a spreadsheet on someone's laptop. It makes an internal review possible two years later.

Agree in advance what gap between two scores counts as material. For example, a lead of five points across the weighted total is usually decisive. A lead of one point is noise. Treating noise as decisive invites a challenge you cannot answer.

Data sourcing matters as much as tool choice. If two platforms disagree about the same crawl, you need a tie-breaker: server logs, a search console export, or a controlled test page.

Building that hierarchy before disputes arise saves arguments later. The source hierarchy used in audits is explained in technical SEO best tools: data and sources.

Where suppliers publish benchmark studies, treat them as marketing until you can reproduce the method on your own domain. Ask for the test corpus, the date and the definition of a pass. If the answer is vague, score the claim at 1 and move on.

Keep a rejected-claims log as well. Recording what a supplier could not evidence is useful at the next renewal, when the same claims often reappear with a new account manager. It shortens the next evaluation.

For repeat purchases, standardise the process so year-two comparisons are possible. A consistent method across review cycles is worth more than a perfect one-off evaluation. The approach used for structured comparison is detailed in technical SEO reviews and comparison methods for 2027.

What should the contract and exit plan contain?

Price review clauses, notice periods, seat definitions and overage charges are the four terms that cause most disputes. Read them together. A cheap seat price with a punitive overage rate is not cheap.

Ask for the price at your likely year-two volume, in writing. A quote that holds only at the opening tier is a discount, not a price.

Insist on export. Every finding, every crawl snapshot and every historical trend should be retrievable in a documented format. If export is limited to a PDF report, the switching cost sits with you.

Set a review date at twelve months and put it in the calendar now. Renewal by inertia is the default outcome otherwise.

Do we need both a tool and a provider?

Usually yes, but not at the same time. Buy the tool first if your team has analyst capacity. Buy the provider first if the bottleneck is interpretation. Add software once the recurring checks are known.

How long should a pilot run?

Thirty days for software, and long enough to produce one agreed deliverable for a service engagement. Extend only if a specific, written question remains unanswered.

Who should score the rubric?

At least two people: whoever operates the tool daily and whoever owns the budget. A third reviewer from engineering helps on integration and rendering criteria.

What is the most common mistake?

Scoring features rather than evidence. A vendor that demonstrates well but cannot show an export, a security questionnaire or a named support contact should lose to a plainer product that can.

Common questions

How many tools should be on a shortlist?

Three to five. Fewer than three gives no comparison; more than five turns evaluation into a paperwork exercise that nobody finishes.

Should accessibility be a scored criterion?

Yes, particularly for public sector buyers. Reports that your own team cannot use with assistive technology create a second, hidden cost.

Can we reuse the rubric next year?

Yes, and you should. Keeping the weights and evidence columns stable makes year-on-year comparison meaningful rather than anecdotal.

What ends a pilot early?

A failed must-have requirement, or a supplier who cannot provide the evidence named in the rubric. Both are answers, not setbacks.