"What do we know about this account?" is still, for most B2B sales and customer-success teams, answered by manually stitching together CRM fields, prior emails, call transcripts, support cases, handoff notes, stakeholder maps, and Slack threads — and, when those run out, human memory. The organization already knows the answer. It's just scattered across six systems and one person's head.
The number is bigger than it looks
Salesforce's State of Sales research puts it precisely: sellers spend about 9% of the workweek researching prospects and another 9% on preparation and planning — 18% total, or 7.2 hours per rep in a 40-hour week. Scale that to a 50-rep team across a 48-week year and you get roughly 17,280 hours annually spent reconstructing knowledge the organization technically already had. Forrester's independent research lands in the same neighborhood: the average rep spends about 14 of 51 working hours a week on admin tasks generally.
Outreach's revenue-productivity research adds the per-meeting texture: manual prep typically runs 30–60 minutes per account. AI assistance has been shown to cut that by roughly 23–26 minutes per meeting — saving an estimated 4–7 hours per rep per week in aggregate. On the customer-success side, PandaDoc reported automating QBR deck preparation cut prep time by 83%, from 45–60 minutes down to about 10. Gainsight markets its Customer 360 and Success Snapshots explicitly on the same premise: saving "thousands of hours every quarter" at team scale.
| Source | Finding |
|---|---|
| Salesforce State of Sales | 18% of the week (7.2 hrs) on research + prep; AI expected to cut prospect research ~34% |
| Forrester | ~14 of 51 working hours/week on admin generally |
| Outreach | 30–60 min manual prep/account; AI saves ~23–26 min/meeting, 4–7 hrs/rep/week aggregate |
| PandaDoc | QBR deck prep cut 83% (45–60 min → ~10 min) via automation |
| Gainsight | Markets Customer 360 / Success Snapshots as saving "thousands of hours per quarter" at scale |
Vendors are already productizing the gap
The pattern is consistent enough that most major platforms have already shipped a version of "assemble this for me": Microsoft surfaces a meeting-preparation card built from CRM data, prior meetings, and inferred meeting intent. HubSpot shows pre-meeting insights, pain points, and recent activity alongside post-meeting summaries. Gainsight centralizes Customer 360, Timeline, and Success Snapshots specifically to automate QBR prep. Salesforce positions AI-generated record summaries and account research as a way to get a seller "up to speed" without manually sifting through records.
None of these count as solved, though — they're assembling the same scattered fields faster, not fixing the fact that the underlying knowledge (who the real economic buyer is, what was actually promised in the last renewal call, which commitment is still open) lives as unstructured text across systems that were never designed to answer "what do we know" as a single query.
This is exactly the gap contextfellow.ai's Action Packet is built to close: instead of assembling a meeting-prep card from whatever the CRM happened to capture, it resolves the actual grounded facts, cited decisions, and known gaps for an account on demand — the same underlying problem these vendor features are all racing to paper over, solved at the fact layer instead of the dashboard layer.