According to Gartner, fewer than 25% of B2B sales organisations rate their pipeline forecasts as accurate. That number has barely moved in a decade, despite significant investment in CRM technology, AI-powered forecasting tools, and revenue intelligence platforms.
The reason is not the tools. The reason is the data inside them.
The GIGO Problem in Pipeline Forecasting
Garbage in, garbage out. It is the oldest principle in computing and the most ignored one in sales operations. A forecasting model - no matter how sophisticated - can only be as accurate as the data it reads. When that data includes stale contacts, inflated deal values, optimistic close dates, and duplicate records, the forecast inherits every one of those errors.
The compounding effect: Pipeline data errors do not cancel each other out. They compound. An inflated deal value attached to a stale contact at a company that was acquired six months ago does not appear as one error - it appears as a valid £50,000 opportunity in your Q2 forecast. Multiply that by 20-30 similarly compromised records and your forecast is off by hundreds of thousands of pounds.
Most sales leaders know their forecast is inaccurate. They apply a mental discount - "take the forecast and multiply by 0.6" is a common heuristic. But this blanket adjustment does not solve the problem. It masks it. The 0.6 multiplier might be 0.4 in one quarter and 0.8 in the next, depending on which errors happen to dominate.
The Four Sources of Pipeline Inaccuracy
1. Pipeline inflation
Sales reps add deals to the pipeline earlier than warranted and remove them later than appropriate. The incentive structure encourages this: a healthy-looking pipeline pleases management, and removing a deal feels like admitting defeat.
The result is a pipeline containing deals that are not real opportunities. A prospect who took an introductory call three months ago but has not responded since is not a pipeline opportunity - but in many CRMs, they sit there occupying a forecast slot until someone manually removes them.
Fix: Implement stage-gate criteria that require specific evidence before a deal advances. "Had discovery call" is not enough for Stage 2. "Identified specific pain point, confirmed budget authority, and scheduled follow-up" is a gate that filters out padding.
2. Stale contact data
A deal is only as real as the people associated with it. When the contact on a deal has changed jobs, retired, or been promoted to a role where your solution is no longer relevant, the deal is effectively dead - but your CRM does not know that.
In B2B, contacts change roles at a rate of 3-5% per month. For a pipeline with 200 active contacts, that means 6-10 are no longer in the role your deal depends on - every month. Over a quarter, 20-30 contacts may have gone stale without anyone noticing.
Fix: Verify contact data for all deals in Stage 2 or later on a monthly basis. A quick LinkedIn check confirms whether the contact is still in the role. Build this into the sales cadence - it takes 30 seconds per contact and prevents months of chasing a dead opportunity.
3. Missing or incorrect deal data
Close dates in the past for "open" deals. Amount fields set to zero or to a placeholder value. Stage fields that have not been updated in months. Each of these data quality issues distorts the forecast in a different way, and together they make accurate forecasting impossible.
Fix: Run a weekly pipeline hygiene report. Flag deals where: close date is in the past, amount is zero or unchanged since creation, stage has not been updated in 30+ days, or no activity has been logged in 14+ days. Make this report visible to sales managers - not as punishment, but as a tool for accurate forecasting.
4. Duplicate records
Duplicate company or contact records create phantom pipeline. The same opportunity appears twice under slightly different company names, or two reps are working the same account without knowing it. The forecast double-counts the revenue, and when the deal closes once (or not at all), the variance is significant.
Fix: Run duplicate detection monthly and merge confirmed duplicates immediately. Configure your CRM's built-in duplicate detection to alert on creation, not just on demand. For a deeper treatment of deduplication, see the VP Sales resource page.
Pipeline accuracy starts with data accuracy. Our pipeline build service delivers verified, enriched prospect data - so every deal in your pipeline starts with accurate, current contact information.
Enforcing Data Discipline
Identifying the sources of inaccuracy is the easy part. Changing the behaviour that creates inaccuracy is harder - because it requires changing daily habits across an entire sales team.
Three principles make data discipline sustainable:
Automate what you can. Mandatory fields prevent incomplete deal records. Validation rules prevent obviously wrong data (close dates in the past, negative deal amounts). Automated alerts flag stale deals. Every rule you encode in the CRM is a rule that does not depend on human memory or motivation.
Make it visible. A weekly pipeline quality dashboard - showing completeness rates, stale deal counts, and duplicate alerts by rep - creates natural accountability. Most people will fix their data when they know others can see the gaps. This is not about blame; it is about transparency.
Connect it to outcomes. The most effective motivation is demonstrating the link between data quality and results. Show the team: deals with complete data close at X% rate, while deals with incomplete data close at Y% rate. When reps see that data discipline correlates with commission, behaviour changes.
What Good Looks Like
A pipeline with accurate data has these characteristics:
- Every deal has a contact verified within the last 60 days
- Close dates are in the future and reflect genuine timelines
- Deal amounts are based on actual conversations, not estimates
- Stages reflect documented evidence, not optimism
- No deal has been inactive for more than 14 days without a reason
- Duplicate rate is below 2%
Teams that achieve this level of pipeline hygiene typically see forecast accuracy improve from 40-50% to 70-80%. The improvement does not come from better prediction algorithms. It comes from having accurate inputs.
For a related analysis of how CRM data quality affects the broader revenue engine, see our piece on how CRM data kills your pipeline.
What would change in your business if your pipeline forecast was accurate to within 15%?
