Algorithm recovery that starts at the date you dropped

What algorithm recovery is, and what a drop looks like from your chair

Algorithm recovery is a fixed-scope engagement that starts at a date instead of at your homepage. Something happened in one week, the site stopped bringing in work, and the job is to establish what happened and repair what can be repaired. It ends in a written verdict, not a retainer.

Owners never arrive describing an algorithm. They describe the week:

• The phone went quiet, and nothing else about the business changed.
• The chart shows a cliff, not a slope. One week is normal, the next is half.
• You still come up first for your own business name, but the searches that brought strangers are gone.
• A page that ranked for six years sits on page three now, and what replaced it is no better.
• You held position in one town and lost it in the next one over.

Two different things get called the same thing. One is that the ranking system reassessed your pages and moved them down. The other is that the results page changed shape underneath you: a map pack where ten links used to sit, an AI answer above everything, more ads at the top. Then your position may not have moved at all while your clicks still fell. A traffic chart cannot tell you which one you have.

The cliff is what matters. A long slow slide over a year is a different problem and belongs to the standing SEO audit, which starts wherever the site is today rather than at a date. Recovery is for the drop you can put a finger on.

The first question is whether the update did it at all

Drops tend to arrive already labeled. The owner has read three forum threads, the date roughly matches something Google announced, and the diagnosis is settled before anyone opens the site. We try to disprove it first: the dull causes are cheap to check and expensive to miss.

What gets ruled out before an update takes the blame:

• Something changed on the site. A redesign, a theme swap, a plugin update, a page renamed by somebody who did not know it ranked.
• A rule started blocking the crawler. A noindex tag that survived staging, a robots line, a bot filter refusing search engines.
• Redirects broke. Chains, loops, or a rule set that stopped applying after an update.
• A link went away. A directory closed, a supplier rebuilt their site, a partner dropped the page that pointed at you.
• The season did it. On a short enough chart, February looks like a penalty in a business that sells in July.
• Somebody else moved. A competitor rebuilt, or a national brand walked into your market.
• The traffic never left. A tag got dropped in the redesign, or a filter changed in the report. The measurement fell and the business did not.

That last one gets checked first. On the sites we manage we run first-party lead dashboards, more than twenty of them, so calls and form fills sit beside the traffic line. If the leads held, the problem is smaller than it looks.

One more split. A manual action is a human decision at Google against your site: it appears in Search Console in plain words, names the reason, and carries a reconsideration request you can file once the cause is fixed. An algorithmic drop has none of that. No notice, no form, no appeal. Anyone offering to call their contact at Google is selling something that does not exist.

The deliverables, and how to check you got them

Every item below is something you can hold up against the invoice afterward:

• A dated timeline with two lines on it. Everything that changed on your website and everything that changed in search, from weeks before the drop through today. Built from Search Console, CMS revision history, hosting and plugin records, archived copies of the pages, and your own memory.
• A loss map at page and query level. Which URLs lost, which searches they lost, and where those clicks went: to a competitor, to another page of yours, or off the results page.
• A verdict with the confidence stated out loud. Sometimes it reads that this is your redirects and we are certain. Sometimes that we cannot pin it on one cause, and here is what to rule out next.
• A manual action check, and the reconsideration request if there is one. Written after the cleanup is done, not in place of it.
• A ranked repair plan, split by who does what. What we handle, what needs your developer or host, and what nobody should touch because the risk outweighs the upside.
• The repairs themselves. Recovery is not a document. The pages and settings that lost get fixed, worst loss first.
• A dated recheck, in writing. A watch window with a date on the calendar, then a plain call: recovered, partly recovered, or not, and what that means for the next dollar.

If the verdict is that the results page changed shape rather than that your site got worse, the deliverable changes with it: a decision about whether to compete for the page's new shape. Better said in week one than after three months of fixing a site that was never broken.

Six steps from the date to the recheck

Six steps, in this order, because attribution falls apart when one gets skipped.

• Freeze the site. For a couple of weeks: no redesigns, no bulk edits, no new plugins. If the site keeps moving while we reconstruct it, nothing afterward can be attributed to anything.
• Get access. Search Console is the whole game, and it has to reach back past the drop. If the property was created after the traffic fell, that history does not exist, and you hear that on day one rather than after the invoice. Then analytics, CMS admin, and hosting access.
• Take your statement. Twenty minutes on a call, often the most useful part of the job. What changed, who touched it, when, which agency had the site before, and whether anybody ever bought links. Nobody enjoys that last question. It saves more hours than any tool we own.
• Reconstruct and attribute. Rebuild what the site and the results looked like before, then lay the two timelines against each other and see what sits at the date.
• Repair. Largest loss first, and inside that, the cheapest thing to verify first.
• Wait on purpose, then recheck. Pages are re-crawled and reassessed on the search engine's schedule, not ours, and some movement only appears once a later broad update runs. There is no way to hurry it. That is the part everybody hates.

Along the way you see the timeline as it is built, a short weekly note in plain English, and the lead numbers the whole time. On the recheck date you get a straight answer, including the one that says we were wrong and here is what to test next.

When a recovery is worth buying, and when it is not

Two things decide it: whether there is still a trail to follow, and whether there was enough traffic before the drop to be worth getting back.

It is usually worth it if:

• The drop has a date you can point at, and Search Console history that reaches back past it.
• Search was a real share of where your work came from, so being wrong is expensive.
• You are willing to change the pages, not only the plumbing. Some causes live in the writing, where no configuration fix reaches.

Keep your money if:

• There was never much to lose. A nine page site that never ranked did not get hit by an update. It was never in the running. That budget belongs in keyword research, content writing, and local SEO.
• It is a slope, not a cliff. A gradual decline over eighteen months is a general condition, and the right purchase is an SEO audit.
• The drop starts the week the new site launched. That is nearly always lost URLs and redirects, which is website migration work with a different playbook.
• You want a guarantee. A shop offering pay-on-recovery is betting your money on something it does not control.
• You bought links and plan to keep buying them. We will not clean a link profile that gets dirty again next quarter.
• You need the phone ringing this month. Recovery moves on the search engine's clock. Google Ads can put you in front of buyers now while the slower work happens.

Before hiring anybody for this, us included, run the site through our free website report card. A clean result rules out a family of causes for nothing, with no email wall.

Which part of the program owns each cause

Recovery is one deliverable inside the larger SEO program, and it borrows from most of the others. The division of labor:

• An SEO audit has no starting point. Recovery has a date. If recovery turns up problems older than the drop, the audit is the follow-on.
• Technical SEO owns the machinery for good. When the cause is crawling, indexing, or speed, that is the repair we run, and keeping it right afterward lives there.
• Website migration owns drops that begin at a launch. New platform, new domain, or a URL scheme replaced without a redirect map.
• On-page SEO and content writing own the pages. If you lost to a result that answers the question better, the repair is the page, not a setting.
• Link building owns the link profile, both rebuilding what disappeared and dealing honestly with what was bought. Competitor analysis covers the case where somebody else moved and you did not.
• SEO reporting is how you catch the next one in week one, and analytics and CRO owns the tracking underneath it. A drop caught in week one is a cheaper job than the same drop in month four.
• Local SEO owns the map. A profile suspension, a category change, or a proximity shift in the map pack is a different animal from a classic ranking drop.
• AI search owns the answer box. If positions held while clicks fell, the click may be ending inside an AI answer. The free AI visibility checker shows whether those systems name you.

Recovery ends. Holding the position once you have it back is the ongoing work on the SEO page, which covers what a month of work includes.