Demand Stream Digital
Names 4 answer engines: ChatGPT, Google AI Overviews, Perplexity, Gemini.
Read 12 September 2026 · demandstreamdigital.com/plumbing-and-hvac-ai-search-optimization
The scores
| Criterion | Score | Weighted of its weight | Contribution |
|---|---|---|---|
| Named engine coverage | 4 / 5 | 20 / 25 | |
| Measurement method shown | 4 / 5 | 16 / 20 | |
| Citation change over time | 0 / 5 | 0 / 10 | |
| Entity and schema competence | 4 / 5 | 24 / 30 | |
| Falsifiability | 1 / 5 | 3 / 15 |
What the page actually says
- Engines
- Four named engines, used consistently across the audit and monitoring phases.
- Method
- A prompt set built for the client's services and cities, a citation audit across the four named engines, competitor citation gap analysis, and monthly prompt scoring. A named prompt set plus a stated cadence is a real method.
- Change over time
- Nothing published, and no sample artefact.
- Entity and schema
- Service, LocalBusiness, FAQ and HowTo schema, brand and location entity disambiguation, crawlability.
- Falsifiability
- One honest timing statement about compounding from month two. No stated limit and no failure condition.
What we would ask them about
One thing on Demand Stream Digital's own page, worth raising before you sign. Not a judgement about the work.
- AskLists HowTo schema among its deliverables. HowTo rich results were withdrawn for commercial pages in 2023, so it buys nothing a reader should pay for.
The check we would run on this one
Demand Stream Digital is one of the few here that shows a method, so the useful check is a level deeper than usual. Ask whether the prompt set is fixed in writing before the work starts, whether it stays identical between reports, and whether you can read the raw answers rather than a summary score. A company already publishing this much will have those answers to hand, and the one that matters is the second.
Disagree with any of this? The corrections log is where it gets settled, in public, with a date.