Free interactive tool

Enrichment Waterfall Calculator: Is Each Source Worth It?

In an enrichment waterfall each record goes to one source at a time and stops at the first match, so the second source only ever sees what the first could not fill, and the third sees even less. A source’s own advertised match rate says nothing about that leftover. This calculator measures every step on the records it actually received: its hit rate on the leftovers, the coverage it adds, what each correct match costs once wrong matches are taken out, and whether that is more than a correct match is worth to you.

The formula

records reaching step i      a_i = N − (new matches from steps before i)
hit rate on leftovers        h_i = m_i ÷ a_i
step cost                    c_i = fixed fee + price × (a_i if billed per lookup, m_i if billed per match)
correct matches              r_i = m_i × share correct
cost per correct match       k_i = c_i ÷ r_i
coverage added               m_i ÷ N

keep step i when k_i ≤ V     (V = what one correct match is worth to you)

Two things change the answer more than the price. A source billed per lookup pays for every leftover record it is sent, matched or not, so its cost per match rises as the steps before it get better. And a match that is wrong costs the same as one that is right, then costs again when a rep acts on it; that is why the verdict uses correct matches, from your own spot check, not all matches.

Run your numbers

In one period, e.g. last month: accounts or contacts sent to the first source.

The most you would pay for one correct field. Leave blank to see costs without a keep/drop call.

Step 1, in the order the waterfall runs

Subscription share for the same period; 0 if none.

Only records this step filled that earlier steps had not.

From a spot check of this step's matches; 100 if not checked yet.

Step 2, in the order the waterfall runs

Subscription share for the same period; 0 if none.

Only records this step filled that earlier steps had not.

From a spot check of this step's matches; 100 if not checked yet.

Step 3, in the order the waterfall runs

Subscription share for the same period; 0 if none.

Only records this step filled that earlier steps had not.

From a spot check of this step's matches; 100 if not checked yet.

1 step costs more than a correct match is worthSource C: it costs more than the correct matches it returns are worth at your value. Removing it changes coverage from 81.0% to at least 77.0% and cost from 114.50 to 80.00.
Coverage (any match)
810 of 1,000 · 81.0%
Coverage (correct matches)
756 · 75.6%
Total cost for the period
114.50
Blended cost per correct match
0.15
Each step, measured on the records it actually received
StepRecords reaching itHit rate on leftoversCoverage addedCostPer correct matchWrong matches addedVerdict
Source A1,00062.0%62.0 pts50.000.0831Keep
Source B38039.5%15.0 pts30.000.2215Keep
Source C23017.4%4.0 pts34.501.088Costs more than it earns

Costs are in whatever currency you entered. “At least” coverage without a dropped step assumes the later steps match no more than they do now, although they would receive more records.

Worked example (hypothetical numbers)

A RevOps team fills company size for the 1,000 identified accounts a month that arrive without it (hypothetical numbers, the tool’s default input). A correct size is worth about $1.00 to them: it replaces two minutes of a rep’s manual lookup. Source A, billed at $0.05 per lookup, fills 620; Source B, billed at $0.20 per match, fills 150 of the 380 left; Source C, billed at $0.15 per lookup, fills 40 of the 230 left. A spot check of 40 records per source found 38, 36 and 32 correct: 95%, 90% and 80%.

Source A costs $50 for 589 correct matches, about $0.08 each. Source B costs $30 for 135, about $0.22 each. Source C costs $34.50 for 32 correct matches, $1.08 each, more than the $1.00 they are worth, and it adds 8 wrong company sizes a month. Its hit rate on its leftovers is 17.4%, and it adds 4 points of coverage (77% to 81%). The team drops Source C: cost falls from $114.50 to $80, and correct coverage goes from 75.6% to 72.4%. The 230 records still unfilled go to the reps’ manual queue, which the value already priced in.

Where do the inputs come from?

  • New matches per step. From the waterfall’s own run log, or a CRM report grouped by the field that records which source filled the value (the waterfall guide calls it enrichment_source_used). Count only records a step filled that no earlier step had.
  • Price and billing basis. From the order form or the vendor’s price page. Check which it is: some sources charge for every lookup, others only when they return a value. A subscription with an included allowance is a fixed fee for the period.
  • Share correct. Pick 50 of one step’s matches at random (all of them if it has fewer) and check each against the company’s own site or a record you trust, as in the match-quality audit. Check each step on its own: accuracy on the leftovers can differ from what the same source shows on easy records.
  • What a correct match is worth. Price the work it saves or the decision it enables: the minutes of manual research it replaces times the cost of a rep’s minute, or the routing it makes possible. If you cannot price it, leave the field blank and compare the steps’ cost per correct match with each other.

What should you change when a step fails?

  • Drop it when the records it fills can wait for a rep or go without the field. The incomplete-data guide helps decide which gaps block a decision and which do not.
  • Move it earlier when it is billed per match and accurate: it still pays only for what it fills, and the per-lookup sources after it are sent fewer records. Moving a per-lookup source earlier does the opposite and makes it pay for more lookups.
  • Narrow what it receives when the field only matters for some records, for example only target-segment accounts. Fewer lookups cut a per-lookup cost directly.
  • Renegotiate or swap it when its accuracy is the problem. A cheaper source with more wrong matches can cost more per correct match than a dearer one.

When is the calculator too simple?

  • Reordering. The calculator scores the order you ran. It does not know how many matches a source would make on records an earlier step used to take, so a new order needs a new test month, or a sample sent to each source on its own.
  • Several fields per call. If one lookup returns company size, industry and region, count a match as the field you needed, and value the extra fields only if a decision uses them.
  • Sources that disagree. A waterfall that stops at the first match never sees a later source contradict an earlier one. If accuracy is low early, a cross-check sample through two sources is worth more than a longer waterfall.
  • Personal data. Each extra source is one more party receiving your records. Before you add one, the data-minimization guide and your vendor review apply, whatever the cost per match.