What a CI engine is, and why most teams never build one
Most companies think competitive intelligence means reading a competitor's blog, checking their G2 page once a month, or running a quarterly win/loss review. That is not an engine. It is a hobby.
A competitive intelligence engine is a systematic, always-on, multi-source system that turns raw signals into intelligence and routes it to the people who need it at the moment of decision. The difference comes down to three capabilities.
- Coverage. How many signal sources you watch, external and internal.
- Speed. How fast a signal becomes usable intelligence.
- Activation. How reliably that intelligence reaches the decision-maker in time to change the outcome of a deal or a renewal.
If a rep in a live competitive deal cannot get a current read on that competitor in under a minute, you do not have an engine yet.
The two-layer signal architecture
An engine processes two fundamentally different classes of signal. External signals are what the market and competitors broadcast publicly. Internal signals are what your own customers tell you about competitors through their behaviour, conversations and requests. Most companies build only for external. The differentiated intelligence comes from combining both.
Four layers turn signals into outcomes.
- 01Collection. Continuous ingestion from every mapped source, external and internal. No manual scraping; agents pull and classify without human involvement.
- 02Processing. Every signal is classified by type (pricing, feature, messaging, people, financial), by competitor and by urgency. Noise is filtered; relevant signals are tagged, timestamped and queued for scoring.
- 03Intelligence. Processed signals are correlated with your own revenue events, closed-lost deals, churned accounts, blocked expansions, so each signal is weighted by business impact rather than by volume.
- 04Activation. Intelligence is packaged per role: battlecard updates for sales, churn-risk alerts for CS, feature-gap briefs for product, positioning notes for marketing, delivered through Slack, the CRM and email.
Recon is the collection and processing layer for competitor moves. Chorus measures the customer reaction. Atlas prices the exposure per account. Cause & Effect draws the link between a move and the reaction, with the measured lag.
External signals: what the market broadcasts
External signals are theoretically available to everyone, which is exactly why most companies watch only two or three of them. Cover the surface, then let scoring decide what matters.
| Source | What it tells you | How to watch it |
|---|---|---|
| Pricing pages | Tier changes, free tiers, packaging moves | Diff the page daily; a changed line is a signal |
| Changelogs and release notes | What shipped and how it is positioned | Subscribe to the feed; classify by theme |
| Job postings | Where the roadmap is going before it ships | Track titles and team names over time |
| Funding and M&A news | Capacity to compete and change in strategy | Alert on the company name |
| Review sites (G2, Capterra, Trustpilot) | How customers compare you, in their words | Read new reviews weekly, especially the comparative ones |
| App stores | Sentiment shifts and release reactions | Watch rating trends after their releases |
| Social and community | Positioning, launches, customer reactions | Track the company and its product names |
| Webinars, events, ads | The message they are choosing to lead with | Sample monthly; note the claims |
| Partner and marketplace listings | Ecosystem bets and integrations | Watch new listings |
| Analyst and press coverage | How the category is being framed | Quarterly read |
| Website and messaging changes | Repositioning, new segments, new claims | Diff the homepage and solution pages |
| Leadership changes | Strategy shifts ahead | Track executive moves |
| Patents and technical posts | Where they are investing | Occasional scan |
| Customer case studies | Which segments they are winning | Read each new one |
| Support and status pages | Reliability, outages, policy changes | Watch for incidents customers will mention |
No team reads all of this by hand. That is the point of the collection layer.
Internal signals: the underused half
Internal signals are the most underused source of competitive intelligence in SaaS. They come from your own customers, and they are frequently more timely and more reliable than anything on the open market. A customer naming a competitor in a call is richer than any review: it tells you which account, which competitor and which feature.
- Sales and CS call transcripts: competitor mentions, decision criteria, objections.
- Support tickets: comparisons, “your competitor does this” requests.
- Win/loss notes in the CRM, structured or not.
- Opportunity fields: the competitor logged on a deal.
- Renewal conversations: pricing pressure, alternatives being evaluated.
- Feature requests that reference a competitor's capability.
- Usage patterns: data-export requests, integration-doc downloads.
- Survey verbatims that compare you to someone.
- Community and Slack Connect conversations.
- Churn reasons captured at cancellation.
- Solutions and implementation notes from evaluations.
Every competitor can read your pricing page. None of them can read your support tickets. Internal signal is the moat.
Mapping signals to revenue
The critical mistake most CI programmes make is treating every signal as equally important. The signals that matter are the ones correlated with revenue outcomes: closed-lost deals, churn, blocked expansion, stalled pipeline. Until you have run that correlation on your own history, you have a news aggregator, not an engine.
Build a revenue-weighted signal stack with five weights.
- ARR at stake. The same signal on a large account outranks it on a small one, whatever the signal type.
- Renewal proximity. Any competitive signal inside the renewal window escalates, because the intervention window is short.
- Recency. A launch from yesterday is actionable; the same launch from six months ago is history.
- Historical correlation. Signals that have preceded losses in your own data escalate faster than ones that have not.
- Pattern density. One mention is a signal. Several from the same account inside a month is a confirmed pattern.
A scoring and routing model
Not every signal deserves the same response. A scoring model stops the team drowning in undifferentiated alerts while making sure nothing high-stakes goes unhandled. Score each signal across six dimensions, then route by band.
| Dimension | 0 | 1 | 2 | 3 |
|---|---|---|---|---|
| ARR at stake | Under threshold | Small account | Mid-market | Enterprise or strategic |
| Renewal proximity | No renewal in sight | Inside a year | Inside 90 days | Inside 30 days |
| Recency | Older than a quarter | This quarter | This month | This week |
| Historical correlation | None in your data | Weak | Moderate | Preceded losses before |
| Pattern density | Single mention | Two mentions | Three in a month | Multiple accounts, same move |
| Competitor tier | Rarely met | Occasional | Frequent | Primary rival |
Route by total: low band goes to the weekly digest, mid band creates a task for the account owner, high band alerts sales and CS in real time and refreshes the battlecard. Publish the thresholds so nobody argues with the routing.
Show the score's working. A brief that says “68% confidence, three sources” gets acted on. A bare score gets argued with.
Activating intelligence across teams
The last failure mode is not collection or processing. It is activation. Intelligence that does not reach the right person in the right format at the right moment is expensive storage. Each team needs a different package of the same underlying signals.
| Team | Package | Cadence |
|---|---|---|
| Sales | Current battlecard per competitor, surfaced in the CRM record | On demand, refreshed by signal |
| Customer success | Accounts naming a competitor, with renewal window and exposure | Real time, for high band |
| Product | Feature gaps behind lost deals, ranked by revenue | Weekly brief |
| Product marketing | Positioning shifts, messaging that resonates, review language | Weekly brief plus alerts |
| Leadership | Competitor pressure by segment, wins and losses by cause | Monthly read |
The auto-battlecard system
Battlecards are sales' primary CI tool, and the endemic problem is that they are written once and never updated. By the time a rep pulls one in a live deal, it is months stale. An auto-battlecard system treats every significant competitive signal as a trigger to refresh the relevant card.
- 01Trigger. Any signal in the high band against a tracked competitor starts a refresh of that competitor's card.
- 02Data pull. The latest reviews, changelog entries, pricing-page changes and relevant call excerpts are pulled into the template automatically.
- 03Draft and hold. The counter-positioning is drafted with the evidence attached and a confidence score, and held for review unless you run it on Autopilot.
- 04CRM surfacing. When a rep logs the competitor on an opportunity, the current card appears in the record.
- 05Versioning. Every update is versioned and timestamped so reps know how fresh it is and what changed.
- 06Distribution. High-band updates go to the account team channel with a summary of what changed and why it matters.
The 90-day build roadmap
Building an engine is a sequenced process. Deploying everything at once produces an overbuilt system nobody uses. Four phases take you from zero to a revenue-connected engine in a quarter.
- 01Days 1 to 14: signal audit and source mapping. Audit current CI practice. Map the external and internal sources against your stack. Run the correlation on the last year of closed-lost and churn data to find which signals actually preceded losses. Pick the top five sources by revenue correlation.
- 02Days 15 to 30: collection and processing. Connect the priority external sources (review sites, pricing-page diffs, job posts, changelogs) and the primary internal ones (call recordings, helpdesk, CRM notes). Turn on classification by type, competitor and urgency.
- 03Days 31 to 60: scoring and activation. Implement the six-dimension score and the routing bands. Ship the first battlecards from real signals. Put the CS alert and the product gap brief into the tools those teams already use.
- 04Days 61 to 90: measure and tune. Track lead time, precision and the deals where the card was used. Tune thresholds by segment. Retire sources that never move the score and add the ones reps keep asking for.
Phases one and two collapse into connecting sources. Recon sweeps daily, scores moves and drafts the battlecard; Cause & Effect attaches the customer reaction and the lag. The roadmap becomes a tuning exercise rather than a build.
Templates
Two templates to start with. The battlecard is the shape Recon drafts; the scoring sheet is the model from chapter six.
COMPETITOR 8×8 · primary rival · last refreshed 04 Sep 2026 (v14) LATEST MOVE Free small-team tier (22 May) · score 15 · high band WHO IT HITS SMB segment · 12 of our accounts mentioned it · 4 renew inside 60 days THEIR CLAIM “Enterprise features at team pricing” OUR COUNTER Seat maths over 12 months; migration and admin controls they lack; two customer quotes PROOF G2 comparison (3 reviews), call excerpt (Northwind, 28 May), pricing page diff OBJECTIONS Price per seat → total cost of ownership; feature parity → admin, SSO, audit log WHAT CHANGED Added free tier; removed annual discount line; new “teams” landing page OWNER PMM · sales enablement note sent to #competitive-8x8
SIGNAL Competitor named in renewal call · Northwind Labs · 28 May ARR AT STAKE 3 (enterprise, $412K) RENEWAL 2 (inside 90 days) RECENCY 3 (this week) CORRELATION 3 (preceded losses before) PATTERN 2 (two mentions this month) COMPETITOR TIER 3 (primary rival) TOTAL 16 of 18 → high band ROUTE Alert CS + sales now · refresh 8×8 battlecard · open play in the Signal Desk