Switching GEO Agencies Without Starting From Zero

Nobody writes this guide, for an obvious reason: agencies do not publish instructions for leaving them. Which leaves the buyer working out on their own what transfers, what is lost, and how much of eight months' investment survives a change of provider.

More of it transfers than people expect, and the parts that do not are usually lost through poor exit handling rather than through anything inherent to switching.

The first question is harder than the logistics, though, and worth settling before anyone drafts a notice email.

Key Takeaways

• Distinguish the agency from the timeline: quiet months six to nine are frequently normal.
• Most of the asset is yours: published content, earned placements, and source corrections stay.
• The baseline is the thing to extract: without it the new agency restarts measurement from scratch.
• Tool accounts matter more than they seem: historical data lives in whoever's name they were opened.
• Switch between cycles, not mid-placement: outreach in flight is the easiest thing to lose.

First: Is It the Agency or the Stage?

The most common trigger for switching is a quiet stretch somewhere between months six and nine, and that stretch is frequently the programme working as expected rather than failing.

Progress runs through three stages. Description accuracy improves first. Then you appear in answers as one option among several without being the recommendation. Only afterwards does recommendation follow, because it depends on third-party corroboration accumulating over months.

Stage two looks exactly like a plateau. Mentions rise, nothing converts obviously, and someone reasonably asks what the retainer is buying.

The bottom three rows account for a large share of cancellations and none of them indicate a provider failure. If your reasons sit entirely in that half of the table, changing agencies restarts the clock on something that was progressing, and the distance between mention and recommendation is exactly what our AI Recommendation Gap report measures.

What Actually Transfers

The valuable part of a GEO programme lives outside the agency's systems. Published content, earned placements, and corrected sources belong to you and keep working regardless of who manages them next.

The pattern is clear: outputs survive, working knowledge does not. Everything in the at-risk column is recoverable by asking for it before you give notice, and largely unrecoverable afterwards.

What to Extract Before You Give Notice

Order matters here. Requests made after notice are handled with less enthusiasm, and some agencies treat measurement data as their intellectual property regardless of who paid for it.

• The full baseline, with method. Prompt set, platforms, locations, dates, run frequency, and the raw answers if they were stored. This is the single most valuable item and the one most often lost.
• The historical measurement series. Every subsequent run, not just the summary charts. A new agency inheriting a time series can pick up where you were; one starting fresh cannot tell you whether anything improved.
• The source map. Which specific pages, threads, and profiles were producing your answers, and which were corrected or displaced.
• Placement records. Which publications and platforms were approached, which responded, and which relationships exist. Re-approaching an editor who already declined wastes the new agency's first month.
• Technical documentation. Schema implemented, on which templates, and any crawler or configuration changes made.
• Content inventory. What was published, when, and which pieces were built for which prompts.

If any of this cannot be produced, that is itself diagnostic. An agency unable to hand over its own baseline and method probably was not working from one, which retroactively answers the question of whether switching is warranted.

The Contract Details Worth Checking

Four clauses cause most of the friction, and all four are worth reading before the conversation rather than during it.

Notice period. Thirty to ninety days is typical. The practical question is what happens to work in progress during that window, and whether outreach continues or stops on the day notice is given.

Content ownership. Usually straightforward for work published on your domain, occasionally less so for research, datasets, or tools produced during the engagement. Check whether anything was published under the agency's name.

Tool accounts. If visibility tracking was set up under the agency's account, your historical data lives there. Accounts opened in your name with the agency granted access are considerably easier to retain, and this is worth changing before you leave rather than after.

Data export. Whether measurement data must be handed over, in what format, and by when. Many contracts are silent on this, which means it depends on goodwill.

When to Time It

The worst moment to switch is mid-placement-cycle, when outreach has been sent and responses are pending. Those conversations do not transfer, and restarting them costs the new agency a month and costs you credibility with the same editors.

The best moment is after a reporting cycle closes and before the next outreach round begins. You leave with a clean measurement point, and the incoming agency starts from a defined position rather than mid-conversation.

Give yourself a two to four week overlap if the contract permits it. Not for handover meetings, which rarely produce much, but so the new agency can run their own baseline against yours before the old data goes cold, which makes continuity of measurement possible rather than theoretical.

How to Brief the New Agency

The main risk in switching is paying twice for the same foundation work. Avoiding it requires being specific about what already exists.

Provide the baseline and historical series upfront, and state plainly that you do not want a fresh discovery phase covering ground already covered. A competent agency will validate rather than replace your existing measurement, which takes days rather than weeks.

Then be explicit about what was already tried and what failed. Publications that declined, sources that could not be corrected, and content that did not move anything are all useful information, and none of it appears in a summary report.

Ask what they would do differently, specifically, before agreeing a scope. If the answer is a generic programme description, you are buying the same thing from a different vendor, and the third-party work that drives citations is the part to examine most closely.

What the Restart Actually Costs

With clean handover, expect roughly four to six weeks before the new agency is producing rather than orienting. Without it, closer to three months, most of which is rebuilding measurement that already existed.

The published content, earned coverage, and corrected sources continue working throughout, since they are not dependent on anyone managing them. This is why switching is less costly than it feels: the asset is external, and the agency was managing it rather than holding it.

Where switching genuinely costs you is in momentum on relationships and in the sequencing knowledge that never got written down. Both are reduced substantially by extracting the right material before giving notice.

If You Are Not Sure Which Situation You Are In

The question underneath all of this is whether the programme is stalled or progressing quietly, and it is answerable independently of whoever is running it.

Run your core category questions, record which sources produce the answers, and compare against where you started. If third-party sources are entering the mix and description quality has improved, the work is landing whatever the headline number says. If neither has changed in six months, it is not a timeline problem.

That read is what Lureon runs at the start of an engagement, and it is equally useful as a second opinion on one already underway.

FAQs

1. How do I know if my GEO agency is underperforming?

Check whether they can name the specific sources producing your current answers, whether any third-party placements have occurred, and whether reporting includes a stated method. Failures on those three points indicate a provider problem. Rising mentions without recommendation, or uneven results across platforms, are normal stages rather than failures.

2. What do I lose when I switch GEO agencies?

Less than expected. Published content, earned coverage, corrected sources, and technical work all remain yours and keep working. What is commonly lost is baseline measurement, the prompt set, the source map, and outreach conversations in progress, all of which can be preserved by requesting them before giving notice.

3. When is the best time to change providers?

After a reporting cycle closes and before the next outreach round begins. Switching mid-placement-cycle loses conversations already in flight and forces the incoming agency to re-approach contacts who have already been contacted.

4. How long before a new agency is productive?

Roughly four to six weeks with a clean handover including baseline data and source documentation. Closer to three months without it, since most of that period goes into rebuilding measurement that already existed.

5. Should I switch if results have plateaued at month six?

Not on that basis alone. A plateau where mentions are rising but recommendation is flat is the expected stage before recommendation, which depends on third-party corroboration accumulating. Switch if your description is still inaccurate, no earned placements have occurred, or reporting has no stated method.

Contractual arrangements vary and this is not legal advice. Notice periods, data ownership, and content rights should be verified against your specific agreement before acting on any of the above.