Google's goto redirect is repricing rank tracking

Most enterprise SEO reporting rests on an assumption nobody has examined in a decade: that SERP data is cheap, abundant, and refreshed daily by default. That assumption held because scraping Google was economically trivial relative to what the data was worth. As of this month it does not hold, and the teams who notice first get to make a deliberate choice instead of absorbing a price increase they did not plan for.

What Google actually changed

Until recently, a Google result linked to the destination. You could read the URL out of the markup and you were done. Now the result links are rewritten to point at google.com/goto with an encoded parameter, and that parameter cannot be reverse-engineered locally. Following it returns a 302 to the real destination. If you want to know where a result points, you have to ask Google, one link at a time.

Two details make this sharper than it first sounds. Perkins reported that Google does not let HEAD requests through, which is the cheap way to resolve a redirect without downloading the body. That forces a full GET for every single link. And the rollout is effectively complete: Perkins described near total coverage across several residential IP providers by August 26. A Google spokesperson confirmed the change without describing the mechanism, saying the company has a long history of deploying technical measures against evolving forms of abuse and regularly takes steps to protect its services and users.

Read the last row alongside the rest, because that is the part most coverage treats as old news. The num=100 removal multiplied how many result pages a tracker has to fetch to see a top 100. The goto change multiplies how many requests each of those pages costs to resolve. These are not two separate annoyances. They compound, and the product of two multipliers is what actually lands on a pricing page.

Why the cost multiplies instead of adds

Here is the arithmetic that matters. Before num=100 went away, one request could return a hundred results with readable URLs. After it went away, seeing a hundred results took roughly ten pages. Now each of those pages carries results whose destinations each need their own resolving request. Perkins put a real number on the combined effect, reported by PPC Land: 500 to 1,000 requests for the links in a single five-page ranking. That is per keyword, per check.

Multiply by a keyword set. A mid-market client tracking 500 keywords daily was, in the old model, a rounding error. In the new model that same client is somewhere in the range of a quarter million to half a million requests a day, and the binding constraint stops being bandwidth or storage. It becomes rate limiting, which is exactly the pressure point Google controls.

Vendors will absorb some of this and pass on the rest, which is what Semrush did after num=100 when it confirmed a tenfold operational cost increase. Expect the pass-through to arrive as quieter changes rather than a headline price rise: lower daily quotas on existing tiers, slower refresh on anything but your top keywords, deeper result pages moving to a premium add-on, and historical backfill getting expensive. Watch your plan's fine print more carefully than its sticker price this renewal.

“Google did not make rank tracking impossible. It made rank tracking expensive, which is a far more effective way to reduce how much of it happens.”

What breaks in your rank tracking stack

The failures here are quiet. Nothing throws an error. Numbers keep arriving in the dashboard and they keep looking like the numbers that arrived last month, which is the dangerous part. Four specific things degrade first, and each of them looks like a market change rather than an instrumentation change.

The fourth one causes the most expensive mistakes. A team sees a drop across a keyword set, reads it as a penalty or an update, and spends a month and a content budget responding to a change in how their vendor samples results. The tell is that the drop appears in one tool and not another, and it appears everywhere at once rather than clustering by page type or intent.

The reporting setup that survives this

The fix is not a better scraper. It is deciding, on purpose, which numbers you are willing to pay premium rates for, and moving everything else onto data Google gives you directly. First-party data got structurally cheaper this month relative to third-party data, and that shift should show up in how your reporting is built.

Start by splitting your keyword set into two tiers. Tier one is the set where a position change triggers an actual decision: your money pages, your commercial head terms, the twenty to eighty phrases a category leader must own. Track those daily, at full depth, and pay properly for them. Tier two is everything else, and it belongs on Search Console impressions and average position, weekly, at no marginal cost. Most enterprise programs we audit are paying third-party rates for a tier two set that nobody has looked at in a year.

1. Export your tracked keyword list with last-90-day position history
2. Flag every keyword whose movement triggered a decision this year
3. That flagged set is tier one. It is usually under 15% of the list.
4. Move the rest to Search Console reporting, weekly, position plus impressions
5. Re-price tier one at full depth and daily refresh, and check the quota
6. Add a monthly manual spot check of 10 keywords to catch vendor drift

That last step is the one teams skip and then regret. A small manual check, run the same way every month, is your only independent read on whether your vendor's methodology moved. It costs an analyst half an hour. It is the difference between catching a sampling change in September and discovering it in January when the annual review does not reconcile.

Rank tracking is not the same as knowing what happened

There is a larger point here that the pricing conversation tends to bury. Position data was always a proxy. It told you where you sat in a list, and you inferred traffic, visibility and demand from that position. The inference has been getting weaker for two years as results filled with AI Overviews, AI Mode and the rest, and the citation slots inside AI Mode link carousels do not map onto a blue-link position at all. A tool that tells you that you rank third is describing a surface that decides less of your traffic every quarter.

This is also part of a broader pattern of Google closing off programmatic access to its own results, which we covered when the SerpApi ruling landed. Read together, the direction is unambiguous: the open, cheap, machine-readable SERP is being withdrawn, and the data that remains free is the data Google chooses to give you in Search Console. That makes Search Console reliability a load-bearing dependency rather than a nice-to-have, which is uncomfortable given that its generative AI logging had a documented bug this month. Build your reporting so a single source going wrong is visible rather than invisible.

Do this before your next reporting cycle

Three concrete moves, in order. Open your rank tracking contract and find the quota, the refresh rate and the result depth you are actually entitled to, not the ones you assume. Then run the tier one and tier two split above and cut the keyword count to what you will genuinely act on. Then add one line to your monthly report that states the data source and refresh rate for every number in it, so that when a figure moves, the first question is answered before it is asked.

Expect pushback from two directions, and both are worth taking seriously. Your agency or in-house team will point out that a smaller tracked set makes it harder to spot broad, site-wide movement early, and that is true: wide shallow tracking is a genuinely useful smoke alarm. Keep it, but move it to Search Console and accept weekly resolution, because a smoke alarm does not need to be a high-resolution camera. Finance will ask why the tracking bill did not fall when the keyword count did. It often will not fall immediately, because tier one at full depth and daily refresh is the expensive part and you just committed to paying for it properly. The saving shows up as the thing you stop buying next year, not as a credit this quarter.

For teams where organic is a primary revenue channel, this is a good moment to move reporting off vendor defaults entirely and onto something you control, which is the case we make for dashboards built on first-party data. It matters most in categories with long buying committees and heavy comparison research, B2B SaaS especially, where a single head term genuinely does move pipeline and deserves the premium refresh. Google spent the summer making SERP data expensive. The right response is to be deliberate about which of it you still need, not to pay more for all of it out of habit, and not to pretend the number in the dashboard means exactly what it meant in June.

See where you are cited today

A free snapshot audit of your rankings and AI citations before we ever talk.