We audited every citation we have ever submitted

Citations are the one part of the stack we cannot see working from the outside. There is no traffic signal and no ranking line that isolates them, so the only way to know they landed is to open every destination and look.

So we did. We pulled every citation campaign in the account, took the per network URLs Search Atlas reports, and fetched each one to confirm the page exists and carries the right business. This document is what came back.

Most of it is good news, and we have said so plainly below. The parts that are not, we would rather ask about than assume. If any of our method is wrong, that is the most useful thing you could tell us.

21 campaigns, 14 distinct client businesses, 2,680 Hyperdrive credits in total. Each campaign buys the same five network bundle, which resolves to seven destination rows.

This is Search Atlas's own status field, not our interpretation. Each campaign produces seven destination rows: three under YP Network, and one each for Neustar, Data Axle, GPS Network and Foursquare.

A status field is a claim. We treated it as one and went to look. For every destination Search Atlas marked Live, we requested the page and checked that the client's business name actually appears on it.

This is the part we want to credit properly. When Search Atlas says a YP Network, Neustar or Data Axle listing is live, we could nearly always open it and find the client's real name, address and phone on the page. That is the product working.

Two caveats we want to be explicit about, because they cut against us. Several of these destinations block automated requests and the Neustar pages render client side, so our first pass produced false negatives purely from our own rate limiting and from reading pages before they rendered. We re ran those with pacing and a render wait, and every one resolved. Nothing in this section should be read as us catching a dead listing, because we did not find one. Second, this check proves the page exists and carries the business name; it does not prove the address and phone are correct on all 99, which is a deeper audit we have not run.

GPS Network and Foursquare are part of every bundle and are billed on every campaign. Across all 21 campaigns, going back to 2025-03-27, neither has ever moved past Submitted, and neither has ever returned a URL.

These are small in number but they matter, because the entire value of a citation is that the details are right everywhere.

One location's citation points at a different location

Campaign 1867 is Curated Events Chesapeake, Virginia. Its YellowPages, DexKnows and Superpages URLs all resolve to curated-events-raleigh-577185364, the same listing as campaign 1869, which is Curated Events Raleigh, North Carolina. Only the Neustar record carries the correct Chesapeake address. Two campaigns were billed, and on three of the destinations only one location appears.

A business name is misspelled in the live listing

Campaign 1481 is Custom Builder Connection. All three YP Network URLs read stom-builder-connection-578687752. The leading C is missing. For a product whose purpose is exact name consistency, this is the kind of error that works against the goal.

Nearly every listing is filed nationwide rather than local

Of 21 campaigns, all but one carry YellowPages URLs under /nationwide/ rather than a city path. We checked what that means in practice: a search on YellowPages for one client in its own city returned no result, while the nationwide URL for that same client was live and correct. A second client filed under a city path did surface in its local search.

If the nationwide path still feeds local results through the data pipeline, that is fine and we will stop worrying about it. If it does not, then the listings exist but are not doing the local job they were bought for, and that would be worth fixing across the whole portfolio.

We would like help tying the charge to the line items. Three numbers are in play and none of them agree.

Per destination rows returned by the API   7 rows x 15 credits = 105
Pricing preview shown before submission    5 networks at 30, less 20% bundle = 120
Actually charged on recent campaigns       160

Our own submission records note the 120 versus 160 gap and describe it as consistent across every client, so this is systematic rather than a one off. Across the portfolio the difference between the previewed price and the charged price is real money, and we would like to understand which figure is correct.

The related question is whether the two networks that have never gone live are billed at the same rate as the three that do. If GPS Network and Foursquare carry the same 15 credits each while never producing a listing, that is 630 credits across the portfolio for an outcome we cannot observe.

We pulled every traffic source for 10 client properties since the first campaign and matched them against the exact destinations the citations landed on.

We do not read this as a failure. Citations are not a referral traffic product. Their job is consistent name, address and phone data feeding the local pack and the wider entity graph, and judging them on click throughs would be the wrong test. We include the number only so it is on the record and so nobody later mistakes it for a hidden win.

It does raise one honest question though, which is question four below: if clicks are not the point and the listings themselves are not visible in local directory search, what is the observable signal that a citation is doing its job?

We looked, and we want to report the limits of what we found rather than dress it up.

Only one of these has enough history on both sides to compare, and its movement sits inside the range it was already bouncing around before submission. The other two had no Search Console property connected far enough back. On top of that, citations were never running alone: profile work, content, technical fixes and reviews were all in flight in the same windows.

So this table is not evidence for or against citations. It is us showing our work and saying that organic clicks are not a measurement that can answer the question.

1. 1Are GPS Network and Foursquare expected to reach LiveAcross 21 campaigns and 2025 to now, neither has ever returned a URL. Is Submitted their normal terminal state, or should we treat this as a failure and ask for the credits back?
2. 2Does the nationwide path still deliver local relevanceNearly all our listings are filed under a nationwide URL rather than a city one. We showed one client not appearing in its own city search on YellowPages. Does the underlying data still carry the local address to downstream consumers?
3. 3Can the two wrong records be correctedCurated Events Chesapeake resolving to the Raleigh listing, and Custom Builder Connection filed as stom builder connection. What is the correction path, and will it consume credits again?
4. 4What is the observable signal that a citation workedIf it is not clicks and not local directory visibility, tell us the metric and where it surfaces, and we will track it there and report it to clients honestly.
5. 5Which credit figure is correctThe API rows total 105, the preview quotes 120, the charge is 160. We would like the three reconciled.
6. 6Is there a re push or refresh pathWhen a client changes address, phone or hours, what is the supported way to update every downstream listing, and does it cost a full bundle again?
7. 7Can the aggregator detail expose a last checked dateRight now a status reads Live with no indication of when that was last confirmed. A timestamp would let us audit continuously instead of manually, and would have made this document unnecessary.

Every figure on this page is computed from the raw pulls, not typed by hand. The page is generated from the data files, so what you read is what the API returned.

Campaign inventory
  gbp_list_citation_submissions(page_size=50, detail=true)   21 campaigns, 2,680 credits

Per network status and URLs
  gbp_get_aggregator_details(citation_id) x 21              7 destination rows each

Independent verification
  Direct HTTP GET on each claimed live URL, browser user agent
  Rendering service fallback for hosts that refuse automated requests
  Pass condition: HTTP 200 and the client business name present in page text
  99 URLs checked, 99 verified, 0 mismatch, 0 no response

Referral traffic
  GA4 Data API runReport, 10 properties, dimension sessionSource
  Window from the first campaign to today, matched against citation destination hosts

Organic performance
  Search Console API searchAnalytics/query, dimension date, aggregated monthly
  Split at each client's own citation submission date