AutoSEO or FullSEO: Four Questions a Procurement Review Asks

A subscription of a hundred and forty-nine dollars a month does not survive a Zurich procurement review because it is cheap. It survives because someone can answer four questions about it: what exactly is being bought, what happens at twelve domains instead of one, what is deliberately excluded, and what it costs to walk away. Most proposals fail on the third and fourth.

This piece works through those four questions with the published list prices and a set of clearly labelled assumptions. It is written for the person who has to defend the line item rather than the person who will use the tool — the two need different documents, and the first one is usually missing.

One framing point before the numbers. The comparison is not automation against expertise. It is the repeatable part of the work against the part that is decided case by case. The platform covers the first; the second remains a human cost regardless of which tier appears on the invoice.

Two tiers, priced per domain, plus a small add-on menu

The structure is short enough to fit in a single slide, and one detail in it drives every subsequent calculation: the price attaches to the domain, not to the account. A group with twelve domains does not buy one licence with twelve seats.

The analytics are identical in both tiers: eight views over Search Console data, six over rank tracking, six in the generated market research, plus the indexing area and the project stream. That matters for the review, because the cheaper tier is not a cut-down dataset. The difference is who decides what the campaign does.

Add-ons are bought in slots rather than tiers: Wikipedia placements at ten dollars a slot in steps of zero, one, five or ten; partner-network placements at one dollar a slot in steps of zero, twenty, one hundred or five hundred. The network is stated at over 230 000 sites. Five Wikipedia slots come to fifty dollars, five hundred network slots to five hundred.

What happens when the domain count is not one

Per-domain pricing is linear, which makes the arithmetic trivial and the conclusions non-obvious. The line item that looks negligible at one domain becomes a real budget position at twelve, and the tier mix matters more than the tier choice.

The fifth row is the configuration most groups actually end up with, and it is worth putting in front of a reviewer next to the last row. Moving eleven of twelve domains to the cheaper tier saves more than fifty thousand dollars a year, and the question of whether those eleven need manual keyword selection is answerable in one meeting.

The line items that are deliberately not in the price

This is the section that decides whether a proposal is credible. A submission that presents the licence as the total cost of the programme will be treated, correctly, as incomplete. Naming the exclusions before anyone else does converts a weakness into evidence of preparation.

Two further exclusions belong on the list because they are the ones that surface late. The first is data handling: connecting Search Console, analytics and mailbox runs through a single Google consent, which is convenient operationally and needs a named owner internally, not a shared login. The second is reporting format. Automated reports are generated with your own logo and colours, but the PDF renders up to 250 rows; anything larger has to leave as a raw export and be formatted by whoever presents it. Neither costs much, and both cost a fortnight of confusion if discovered after signature rather than before.

Where the group has more than a handful of domains, the review will also ask who can see what. Access is granted per domain, so an external supplier sees only the properties it works on, and a single property can be handed to another person's address without touching the rest. That is worth a line in the submission, because it converts a data-protection question into an already-answered one. Groups that need per-project control over targets as well tend to place the main domain on the managed tier and leave the rest on the automated one.

The build-versus-buy comparison, done properly

Every review reaches the question of hiring instead. The comparison is legitimate and usually run badly, because it compares a licence against a salary rather than against a capability. The honest version compares three routes on the same set of tasks.

Route three deserves a specific caution in this market. A full-time specialist in Zurich carries a total employment cost that will exceed the licence for a twelve-domain portfolio several times over, and the role still requires tooling. That does not make hiring wrong — it makes it a decision about capability retention rather than about cost, and it should be argued on those grounds.

What happens if it is stopped after a year

Procurement asks this and rarely gets a straight answer. The straight answer here has two halves, and both are worth knowing before signing rather than after.

• The data connection runs through your own Search Console property. The historical data was never held hostage; it stays where it was.
• The exports run to 10 000 rows in CSV and JSON, so the working dataset can be taken out in a usable form.
• The pages are yours and remain published. Nothing about them depends on the subscription continuing.
• The links built exist on third-party sites; what stops is the further building, not what already stands.

Compare that with an arrangement where reporting, tracking and account access sit with a supplier. The difference shows up only at the moment of a change, which is precisely when nobody has time to deal with it. Anyone who has run a supplier transition knows what that half-day of clarity is worth.

A twelve-month schedule the reviewer can hold you to

Budget approval usually comes with an implicit question about when the result will be visible. Answering with «SEO takes time» is accurate and useless. The specific sequence is defensible and, more importantly, checkable.

The first measurable movement usually appears within four to eight weeks, and it appears in the number of terms producing impressions rather than in revenue. Presenting that as the month-two milestone, in advance, prevents the conversation in month three where somebody asks why sales have not moved.

Reading the add-on menu without over-buying

At a dollar a slot, the largest network tier looks like an obvious purchase. It usually is not. Link volume is not the constraint on most programmes; the constraint is whether there are enough worthwhile destination pages for those links to point at. Five hundred placements aimed at a domain with four service pages produces a shape that raises more questions than it answers.

A more defensible rule ties the tier to the destination count: a domain with a handful of service pages takes the smallest tier, one with a couple of dozen pages across languages takes the middle, and the largest tier makes sense only where a genuinely large estate can absorb it. Wikipedia slots invert the logic — dearer per unit, but far more useful in fields where corroboration carries weight, provided a factual occasion exists. Where no such occasion exists, that money belongs in the writing budget.

There is one more piece of arithmetic worth running before the tier discussion, and it takes a minute. Divide the annual licence for the portfolio by the number of pages the programme is expected to touch in that year — new pages plus revisions. In most realistic plans the figure lands somewhere between forty and two hundred dollars per page touched. That is a number a reviewer can weigh against what a single page costs to write, and it usually settles the tier argument faster than any feature comparison, because it puts the licence in the same unit as the work it is meant to support.

What the one-page proposal should contain

Reviewers do not read appendices. The submission that gets approved fits on a page and answers the four questions in the order a finance function asks them.

Sections two and three are the ones that get proposals approved, and both are usually absent. A submission that names its own exclusions and its own exit reads as a considered plan rather than a purchase request, which is a different conversation entirely. The implementation work behind it sits under technical services and the writing under content.

The number nobody puts in the submission

Every review compares the cost of acting against zero. The comparison that is missing is the cost of not acting, and it is not zero. In a field of forty comparable providers competing for the same enquiries, visibility does not stay level when nothing is done — it drifts towards the firms that are doing something. The drift is slow, hard to attribute and therefore systematically absent from budget discussions.

There is no honest way to quantify that drift in advance, and any figure offered for it is invented. What can be stated is the direction and the mechanism, which is enough for a reviewer who understands that competitive position is a stock rather than a flow. Presenting it as a stated risk rather than a calculated number is both more honest and, in most rooms, more persuasive.

If the review needs figures from your own estate rather than a worked example, connect a single domain in the Semalt workspace and let one month pass before deciding on tiers at all. The question of which tier belongs on which domain tends to answer itself as soon as the first analysis is on the table. Further pieces are in the blog, and the campaign mechanics under automated campaigns.