When a customer's principal engineer, staff engineer, software architect, head of platform, head of infrastructure, head of data, head of security, or independent operator delivers a conference talk and names your product as part of their architecture, references your product on a presentation slide that lists the stack, demonstrates your product live on stage as part of the talk, or attributes a measurable outcome to your product during the conclusion of a session, they are delivering a category of endorsement that no marketing-elicited testimonial can replicate. The talk has been delivered in front of a peer audience whose attendance is itself a credibility signal, recorded for the conference archive, slide-anchored to a visual artifact that survives independently of the recording, attended by industry peers who could publicly challenge inaccurate claims during Q&A, and archived in multiple distribution channels (the conference's official video platform, the speaker's personal slide-deck publication, the conference's program archive, and the speaker's company blog summary). The recording carries the speaker's verbal endorsement, the slide deck carries the visual attribution, and the surrounding context establishes that the talk was selected from a competitive call-for-proposals process by a program committee of peer reviewers.
Almost no B2B software-tooling, developer-platform, or operator-facing-product marketing team systematically extracts product mentions from conference talk recordings and presentation slide archives. The omission is the natural extension of the same blind spots we documented in our SEC filing extraction guide, our quarterly earnings call extraction guide, our academic paper extraction guide, our patent filing extraction guide, our YouTube content extraction guide, our Reddit content extraction guide, our open-source repository extraction guide, our Stack Overflow extraction guide, and our podcast extraction guide. Financial disclosures cover business-context written mentions. Earnings calls cover analyst-pressured spoken mentions. Academic papers cover research-context written mentions. Patent filings cover legally pressured engineering mentions. YouTube content covers face-attached video mentions. Reddit content covers peer-scrutinized text mentions. Open-source content covers cryptographically signed engineering mentions. Stack Overflow content covers reputation-attached Q&A mentions. Podcast content covers conversational spoken mentions. Conference talk and presentation slide content covers peer-reviewed-selected, peer-audience-attended, slide-anchored, multi-archive-preserved talk mentions made under the social pressure of public live presentation where Q&A could surface contradictions in real time — the tenth pillar of the structurally durable public corpus, and the only one where the customer's testimony has been ratified by a program committee of peer reviewers before it was even delivered.
This guide describes the extraction workflow for the conference talk and presentation slide corpus.
Why a conference talk mention beats almost every marketing-elicited testimonial
A conference talk mention is a category of endorsement that has passed through filters no marketing-elicited testimonial encounters. Six properties stack to make it one of the most adversarially credible presentation-format endorsement formats in modern B2B operator-facing marketing.
First, the talk has been selected through a competitive call-for-proposals process by a program committee of peer reviewers. Most reputable industry conferences (KubeCon, QCon, Strange Loop, ICML, USENIX, RSA, RailsConf, PyCon, JSConf) accept talks through a competitive call-for-proposals process where a program committee of peer experts reviews submitted abstracts and selects a small fraction for inclusion in the program. A talk that names a product as part of its content has been implicitly ratified by the program committee as having a useful and accurate technical story to tell. The peer-review-selection property is what makes conference talk mentions more credible than mentions in any format that does not pass through a comparable selection filter.
Second, the talk has been delivered in front of a peer audience whose attendance is itself a credibility signal. Conference attendees pay registration fees, take time away from work, and select sessions based on the program description. A talk attended by hundreds of peer engineers who selected the session over competing concurrent sessions is being heard by an audience of self-selected peer experts whose presence ratifies the topic's relevance and whose engagement during Q&A could surface contradictions in real time. The peer-audience-attendance property is materially stronger than the equivalent on any format without comparable attendance discipline.
Third, the talk is contextualized by the slide deck, which survives independently of the video recording. Every conference talk is accompanied by a slide deck that the speaker frequently publishes separately on SpeakerDeck, SlideShare, the speaker's personal site, or the conference's official slide archive. The slide deck carries the visual attribution of the product (logo, name, integration diagram, code snippet) in a format that is independently archivable and citable without requiring the viewer to watch the full video. The slide-deck-independence property is unique to the conference talk corpus and supports a deployment format (slide screenshot with caption) that no other corpus supports natively.
Fourth, the talk is recorded on multiple archival channels simultaneously. Most reputable conferences record talks and publish them to the conference's official video platform (YouTube, conference-hosted Vimeo, dedicated conference site) within weeks of the event. The recordings are mirrored on the speaker's personal channel, summarized on the speaker's company blog, and frequently captioned by third-party services. The multi-archive property is what makes conference talk mentions more durable than mentions in any format without comparable archival redundancy.
Fifth, the talk frequently includes a live demonstration that proves the product is in production use. Many conference talks include a live demonstration where the speaker shows the product operating in the speaker's actual production environment, runs a query against the speaker's actual data, or steps through a code path in the speaker's actual repository. The live-demonstration property is materially stronger than the equivalent on any format without comparable proof-of-deployment signaling.
Sixth, the talk is associated with structured metadata that supports automated discovery at scale. Every conference talk has structured metadata (speaker name, speaker company, talk title, talk abstract, conference name, conference year, conference track, talk duration, recording URL, slide deck URL) that the conference's program page exposes in a structured format and that conference aggregators (InfoQ, ChannelNine, the conference's own archive site) expose through search interfaces. The structured-metadata property is what makes the conference talk corpus one of the most discoverable presentation-format corpora for testimonial extraction at scale.
The six conference talk content locations where customer mentions appear
The conference talk ecosystem has six primary content locations where a product mention can surface, and each carries a different credibility weight and a different downstream usability.
Location 1 — The keynote talk where your customer's senior engineer names your product as part of their architecture
A keynote talk where the customer's senior engineer names the product as part of their architecture is the highest credibility-dense location because the keynote format is the conference's most prominent stage and is reserved for talks with the broadest expected appeal. A keynote mention is delivered in front of the largest audience the conference produces and carries the strongest implicit endorsement from the program committee. The keynote-talk format is the highest-weight format for conference extraction.
Location 2 — The breakout session talk where your customer presents a deep-dive on their architecture
A breakout session talk where the customer presents a deep-dive on their architecture is the second-highest credibility-dense location because the breakout format encourages detailed technical content where product mentions are embedded in the explanation of specific architectural decisions. A breakout session mention is delivered in front of a self-selected technical audience whose presence ratifies the topic's depth and whose engagement during Q&A could surface contradictions in real time.
Location 3 — The lightning talk where your customer demonstrates a focused use case of your product
A lightning talk where the customer demonstrates a focused use case of the product is the third-highest credibility-dense location because the lightning-talk format encourages concentrated demonstrations of a single tool's value. A lightning-talk mention is concise, focused, and delivered with the speaker's full attention on the demonstration of the product's value.
Location 4 — The panel discussion where your customer references your product in response to a moderator question
A panel discussion where the customer references the product in response to a moderator question is a moderate credibility-dense location because the panel format encourages comparative discussion of multiple tools in the same category. A panel mention is implicitly compared against the perspectives of other panelists who may be using different tools, and the survival of the mention without contradiction by the other panelists is itself a signal of consensus.
Location 5 — The workshop or tutorial where your customer teaches participants to use your product
A workshop or tutorial where the customer teaches participants to use the product is a moderate credibility-dense location because the workshop format encourages hands-on demonstration of the product as a working tool. A workshop-format mention is the strongest possible proof-of-deployment signal because the speaker is using the product to deliver instructional content rather than merely describing the product.
Location 6 — The poster session or BoF (Birds-of-a-Feather) session where your customer informally references your product
A poster session or Birds-of-a-Feather (BoF) session where the customer informally references the product is the lowest credibility-dense location because the format is the most informal and the audience is the smallest. However, poster and BoF mentions frequently surface the most candid product references because the informal format encourages unguarded discussion.
The extraction workflow — eight steps from query to deployable testimonial
The conference talk corpus rewards a workflow that distinguishes between the program-archive search (which uses talk metadata) and the slide-deck and transcript search (which uses verbatim content from the talk itself). The eight-step workflow below converts a query into a deployable testimonial in a way that survives downstream review and remains attributable to the original customer.
Step 1 — Construct the product-name query and the conference-target set
The first step is to construct the product-name query and the conference-target set the workflow will use across all six content locations. A query for ProofShow would include the product name itself, the company name, and the canonical category terms (testimonial collection, social proof, customer evidence) that conference talk titles and abstracts are likely to use. The conference-target set should enumerate the conferences whose tracks are most likely to surface relevant content (frontend conferences, growth-marketing conferences, B2B SaaS conferences) and should be saved as a structured artifact for reuse across all subsequent extraction sessions.
Step 2 — Run the program-archive search and the SpeakerDeck/SlideShare slide-content search
The second step is to run the program-archive search across the target conferences (using each conference's program search interface) and the SpeakerDeck and SlideShare slide-content search for the product-name query. The output is a corpus of candidate talk-level mentions that include the program abstract context and the slide-deck context where the product is named.