1001 SEO Media
All posts
By 1001 SEO MediaAthenaHQMigrationPromptwatchGEOClient reporting

How we migrate a client's AI visibility program off AthenaHQ

Our runbook for moving a brand from AthenaHQ to Promptwatch after the Bloomreach deal: the intake questions, how each AthenaHQ object maps across, how we keep the client's reporting continuous, and what we never promise during a switch.

Earlier this week we wrote about what we're telling clients about the AthenaHQ and Bloomreach deal. The short version was: don't panic, export your prompt set, and run a second tool before renewal. That last step is where the work is, so this post spells it out. It's the runbook we use for a tool switch, written down so you can run it yourself if you'd rather.

It assumes the destination is Promptwatch, the platform we run client programs on. The steps work with any tool, but the details are Promptwatch's, because that's where we do the work.

The intake call

Before anyone touches a CSV, we spend 30 minutes with the client on four questions. The answers decide the plan.

  1. When does the AthenaHQ contract renew? We count back six weeks from that date. That's the latest day the parallel run can start.
  2. Who reads the reports, and what do they look at? A CMO who looks at one trend line needs a different handover than a content team that works the task list every day.
  3. Which markets matter? Our AthenaHQ notes, taken from its plans page on November 16, have Starter at country level, with multi-region, multi-language and persona targeting on Enterprise. Brands that sell in several countries often tracked fewer markets than they wanted. We ask which ones they would have added.
  4. What's half finished? Open tasks in AthenaHQ's Action Center, and content the client started off the back of them. Those need a home before anything is cancelled.

AthenaHQ pricing page with a free Essential plan, Starter at $295 a month and a custom Enterprise plan

Week one: export and inventory

We export everything the client's AthenaHQ plan allows. When we read the plans page on November 16, Starter listed CSV export. That means prompts, topics, competitors, countries and as much response history as the export gives us. No tool we know of imports another vendor's history, so the export is an archive. The client keeps a copy, and so do we.

Then we copy the open Action Center tasks into a plain list: page, type of fix, owner, status. This is tedious and it is the step people skip. Skip it and three months of planned work disappears with the old login.

Week one: map and rebuild

Every AthenaHQ object has a counterpart in Promptwatch. This is the mapping we work from:

In AthenaHQWhere it goes in Promptwatch
PromptsPrompts inside a monitor; the monitor's language and location are fixed once created
TopicsTopics, each with its own color and a Topics page showing visibility, citation share and share of voice
CompetitorsTracked brands, with brand merging and auto-discovered competitors
Naming rulesBrand aliases and Brand Book analysis instructions
GEO scoreVisibility Score, Share of Voice and Position, read together
Action Center tasksCustom actions on the Action Board, with owners and comments
Content optimization agentContent Refresh and AEO articles from Professional; the Content Agent on Business
Personas (Enterprise on AthenaHQ)Personas, with persona stacking

We rebuild the prompt set exactly as it was. Same wording, same competitors, same country and language. We load it through Promptwatch's REST API, which creates prompts in bulk, instead of typing it in. The temptation is to fix the prompt set while we're in there. We don't, not yet. Change both at once and the first question from the client, "why did our number move?", has no honest answer.

Promptwatch prompt tracking page headed "Track prompts that get citations and clicks", with a product illustration of a prompts view listing top competitors and a visibility-by-prompt-type chart

Day one of the new tool: connect what can't be backfilled

Two things in Promptwatch only have history from the day you connect them, so we connect them first:

  • Crawler logs. Agent Analytics reads the client's logs through Cloudflare, a CDN log drain or the WordPress plugin and records IP-verified visits from GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, PerplexityBot and the rest. Our AthenaHQ notes list no crawler-log product, so for most clients this is the first time they see which pages AI bots actually fetch. Crawler logs start on the Professional plan.
  • Visitor analytics. The script, the GTM template or the WordPress plugin. It ties AI-referred visits to conversions on every plan, and its "Include ChatGPT UTM referrers" setting counts ChatGPT visits correctly since ChatGPT stopped sending a referrer on October 8.

Weeks two to five: the parallel run

Both tools run the same prompt set for four weeks. We don't compare headline scores. AthenaHQ's GEO score and Promptwatch's Visibility Score are built differently, and a gap between them is a difference in definitions, not a result. What we check:

  • Do the same prompts flip from absent to present in both tools?
  • Do the same competitors lead the same topics?
  • Are the cited sources broadly the same?

Promptwatch's responses export lists every brand in each answer with its position, which makes the prompt-by-prompt check quick. When the two tools disagree on a prompt, we check the usual causes first: a different engine mix, a different location, or a different time of day for the run. We line those up before we draw conclusions.

Keeping the client's reporting continuous

A client shouldn't notice a tool switch in their inbox, apart from the reports getting more useful. During the parallel month we set up the reporting they'll get afterwards: a custom dashboard on the views they already know, and a scheduled PDF report on the same day of the month their AthenaHQ report used to arrive. Every brand plan includes 10 scheduled PDF reports. For clients who report in Looker Studio, the connector is available from Professional.

The report for the switch month carries one note: the trend restarts on this date, and here's why. We'd rather explain a reset once than defend a cliff in a chart for a year.

Week six: decide

If the trends agree, the client lets AthenaHQ lapse at renewal. If they don't, we find out why before anything is cancelled. Either way, nothing gets switched off before the parallel run is complete.

What we don't promise

We don't promise the new score will be higher. It may be lower on day one simply because it's measured differently. We don't promise that history moves across, because it doesn't. And we don't promise a timeline shorter than six weeks, because a switch done in a fortnight leaves the client with numbers nobody can vouch for.

The plan we'd start on

For a single brand leaving AthenaHQ Starter, we'd start the trial on Professional: $245 a month for 150 prompts and 18,000 responses, with crawler logs, against Starter's $295 for 3,600 credits. Seats are unlimited, and the 7-day free trial on monthly billing covers the rebuild and the first week of the parallel run.

Promptwatch pricing page with Essential at $95, Professional at $245, Business at $579 and a custom Enterprise plan, each listing unlimited seats

If you'd like us to run the migration for you, from intake to the week-six decision, email hello@1001seomedia.com.