Three separate slices of the same live-search-forced ChatGPT audit runs, each answering a different question about who AI actually names when UK buyers ask about payments. Column 1 is the whole payments market: 542 real buyer-phrased prompts covering card machines, high-risk merchant accounts, EPOS, gateways, compliance mechanics and broker framing, plus generic head-terms like “best card machine UK”. Column 2 is MerchantHQ's own targeted turf: a separate 682-prompt run scoped tightly to what MerchantHQ actually built (34 high-risk verticals, 55 trades, EPOS, compliance mechanics), deliberately excluding generic head-terms. Column 3 is what Merchant Advice Service promotes most: a 245-prompt subset filtered out of column 1's existing 542, not a new run, keeping only the prompts that map to her stated core specialisms (high-risk merchant accounts, EPOS, payment gateways, and broker/matchmaking positioning) and dropping the plain generic card-machine picks and brand-vs-brand prompts that stay column-1-only.
Reusing column 1's prompts as column 3's backbone (rather than running a fresh 245-prompt audit) is a deliberate efficiency: the underlying answers were already collected with live search forced on, so re-slicing the existing results by topic gives the same signal without duplicating spend or introducing a different sample.
On the whole payments market, MerchantHQ leads well ahead of Merchant Advice Service: named in 12.2% of prompts against her 1.3% named-as-option rate, roughly a 9x gap, and her source-citation rate (1.8%) is lower still.
On MerchantHQ's own narrow, purpose-built turf, the gap is total: MerchantHQ is cited as the source in 15.0% of its 682 targeted prompts, while Merchant Advice Service does not appear even once (0/682), confirmed directly against the cached raw responses.
On her own most-promoted turf, narrowing column 1 down to just her stated specialisms (high-risk, EPOS, gateways, broker positioning) barely moves her numbers: source citations rise from 1.8% to 2.9%, first-citations from 0.9% to 1.2%, and named-as-option stays flat-to-slightly-up at 2.9%. Even filtered down to exactly what she promotes hardest, she is still cited as a source in fewer than 3 in 100 relevant prompts.
The more striking number is MerchantHQ's own rate on that same narrowed set: it actually drops from 12.2% (whole market) to 6.1% (her promoted-specialism subset), even though the subset is roughly a third high-risk prompts, a vertical MerchantHQ has real depth in. The subset also carries a heavy weight of EPOS, payment-gateway and broker/matchmaking-positioning prompts (about 42% of the 245 combined), categories MerchantHQ's own content doesn't push as hard as its high-risk and trade pages. Even so, MerchantHQ still names or gets cited roughly 2x more often than Merchant Advice Service does on her own best turf (6.1% vs her 2.9% source-citation rate), which is the headline: there is no slice of this market, including her own strongest positioning, where Merchant Advice Service currently out-cites MerchantHQ.
Column 3 is not a new prompt run. It is a keyword-classified subset pulled from the
existing 542-prompt combined dataset (mas_MASTER_all_results_combined.csv),
kept if the prompt text matched one of four categories that map to Merchant Advice
Service's own stated specialisms, and dropped otherwise (generic card-machine/reader
picking prompts like “best card machine UK”, trade-specific generic picks,
and brand-vs-brand prompts like “Dojo vs SumUp” stayed column-1-only).
High-risk matched on terms like high risk, CBD, vape, gambling, adult, crypto, MATCH list, MCC code and rolling reserve. Broker/matchmaking matched on terms like broker, payment/merchant advisor, unbiased, whole-of-market, matchmaking, cold-called, comparison directory, and Merchant Advice Service's own name, plus generic aggregation phrasing (“side by side”, “in one place”, “full comparison”) only where the prompt did not already name two or more specific competing brands, since that pattern signals a plain brand-vs-brand prompt rather than a directory/broker framing. EPOS and gateway matched on direct product-category terms. A prompt matching more than one category was counted once in the subset total (245), assigned to whichever category matched first in that priority order, so the four category counts above sum to the subset total with no double-counting.