Comparing Marketplace Results Year Over Year Fairly
Last October is not this October. Comparing Marketplace results year over year fairly means adjusting for mix, coverage, staffing, and season so a raw lead count cannot bully the used-car meeting.
The GM pulls last October's "Marketplace leads" and this October's number, then asks why the channel is down. Nobody mentions that last year the store stocked trucks, had two reps answering Messenger, and posted most of the pack, while this year the mix is cars, one associate posts when they can, and a third of retail-available units never went live. That comparison is not analysis. It is two unlike months sharing a calendar name.
Comparing Marketplace results year over year fairly is a discipline: freeze definitions, then adjust (or at least disclose) mix, coverage, staffing, and season before you declare a trend. Facebook does not publish a year-ago benchmark for your rooftop. Your own logs are the only series you have, and they lie when the definitions move.
This article is for general managers and internet directors who already sit in used-car meetings. Build the monthly package first with a Marketplace performance scorecard. This piece is how not to abuse the year-ago column.
Why Raw Year Over Year Marketplace Results Mislead
A number without a definition is a mood. A year-ago number with a different definition is a trap.
The store changed, even if the logo did not
Over twelve months a typical independent or franchise used department may have changed:
- Pack size and body-style mix.
- Who is allowed to post, and from which accounts.
- Whether sold units were archived the same day.
- Whether inquiries were logged at all.
- Hours, rooftops, and recon backlog.
If any of those moved, "messages down 30 percent" (a shape of sentence you should not treat as a real statistic in this article) is not interpretable. You do not need a fake percentage. You need a list of what is unlike.
Season and segment are not noise
Last year's October truck mix versus this year's October car mix is not a Marketplace story. It is a pack story. Read segment calendars the way you already do for trucks and SUVs. Compare like segments to like segments, then roll up.
A blended year-ago days-to-turn has the same problem. Cut aging the way you would in days-to-turn and online exposure, then by segment.
Platform behavior changes and you will not get a memo with a number
Meta updates commerce rules, ranking, and product surfaces. You will not receive a dealer-specific "your reach coefficient." Do not invent one to explain a gap. Do check current commerce policies when listings were removed. For the year-over-year conversation, disclose: "We do not control distribution. We compare our coverage, our listing hygiene, our response, and our logged conversations."
Freeze Definitions Before You Compare Years
If January used a different meaning of "lead" than October, you do not have a series.
Definitions that must not drift
Write them once and date the page:
- Posted unit: a VIN that reached posted status, counted unique, not post clicks.
- Inquiry: a buyer-initiated message on a specific VIN, not a view count, not your reply.
- Coverage: retail-available units that were posted, by rooftop.
- Response: first human reply, with a same-day or daypart standard you already use.
- Sold-still-live: sold VIN still advertised.
- Marketplace-touched deal: only if you have a thread or a buyer-stated source. Otherwise do not use the phrase.
When you change a definition, start a new series. Do not splice.
What you are allowed to put in a year-ago table
| Line | This period | Year ago | Unlike notes |
|---|---|---|---|
| Unique VINs posted | count | count | pack size? |
| Coverage of retail-available | ratio you defined | same definition | recon backlog? |
| Inquiries logged | count | count | did we even log last year? |
| Same-day reply misses | count | count | staffing? |
| Sold-never-posted | count | count | process? |
| Segment mix (share of stock) | shares | shares | the honest context |
Leave a notes column. If year-ago inquiries were "whatever was in personal Messenger," write "not comparable" instead of a number. A blank with integrity beats a fake trend.
Do not compare vanity posts
Eighty posts on forty VINs versus forty posts on forty VINs is not twice the work quality. Unique vehicles and coverage first, post count second as a labor and pacing line. Burst weeks last year will inflate posts and hide a coverage hole.
Adjust, or Disclose, These Four Distortions
Fair comparison is mostly humility.
1. Mix
Reweight, even crudely. If you cannot statistically reweight, show two pictures: overall, plus the three largest segments then vs. now. A truck-heavy year ago will "win" on messages in many markets without anyone being a genius. A van-heavy now will "lose" without anyone being a failure.
If you added a segment this year, it has no year-ago Marketplace history. Do not punish it with a blank comparison. Put it on a test scorecard instead.
2. Coverage and unposted days
A year with high coverage and a year with a recon pile are different businesses. Report unposted retail-available units. If last year's team listed almost everything and this year's team listed the easy trucks only, the inquiry gap is partly a choice.
3. Staffing and response
Inquiries that nobody answered are not demand you "had." They are demand you burned. Compare first-reply performance before you compare raw message counts. A quieter inbox with faster, cleaner appointments can be a better operation than a louder inbox that died in personal chat.
4. Logging quality
The most common year-ago lie is "we started tracking properly in March." Then October looks like a miracle. Mark the date logging became real. Compare only post-habit months, or show a break in the chart (even a table footnote: "logging began [date]; prior months excluded").
Source capture has the same break. If you only started asking where they found you this summer, do not compare "Marketplace deals" to a year-ago jacket that said internet for everything.
How to Compare Marketplace Results Year Over Year in the Meeting
Keep it short enough that people stay honest.
A fair GM script
- Are the definitions the same?
- What is unlike (mix, coverage, people, lots, logging)?
- On the lines that are comparable, what moved?
- Is the movement a process we own (coverage, hygiene, reply, sold-still-live) or a mix/season issue?
- What one change do we make this month? Not five.
If the room wants a single verdict ("Marketplace is up/down"), refuse. Give the table.
Illustrative, not a benchmark
Consider a store that posted more unique VINs this October than last, with a weaker truck mix and a new associate. Logged inquiries are similar, same-day misses are down, sold-never-posted is down. A raw "leads" comparison might look flat. The operational story is better coverage and better response on a harder mix. That is a fair year-over-year read. Inventing a conversion rate to make it dramatic would make it false.
What never enters the comparison
- Testimonials and named customers.
- Claims that PostToMarket or any tool guaranteed reach.
- Exact Meta limits or quoted policy text.
- Annual pricing of software (there is none to discuss here).
- A blended close rate you did not measure with frozen funnel stages. If you want deals, use the message-to-sale baseline with the same stages both years.
How PostToMarket Automates This
Comparing Marketplace results year over year fairly requires a durable record, not a folder of screenshots from whoever still had the login. If statuses, unique VINs, and inquiries were never stored, you cannot reconstruct last October, and you should not pretend you can.
PostToMarket keeps unlimited vehicles in inventory with statuses (draft, posting, posted, sold, failed), which is how unique posted VINs and sold-still-live incidents become countable later. Import history and row-level error logs show whether a year's volume started from clean files. Team analytics by rep and source give you staffing and feed context for the unlike-notes column. The leads inbox is a place inquiries can live beyond a personal phone. Per-rep queues, manager assignment, and the team queue view explain who was supposed to cover the pack. Mark as sold and archive so year-ago "live" is not polluted by dead ads. AutoPost's one-vehicle-at-a-time cadence with human-like delays and server-enforced daily limits makes posting volume a paced series instead of undocumented bursts. PostToMarket is not affiliated with or endorsed by Meta and does not provide Facebook's year-ago distribution numbers. It stores your operation.
Compare Years Like an Operator, Not Like a Headline
Last October only helps if it was measured the same way, on a disclosed mix, with coverage and logging you can defend. Comparing Marketplace results year over year fairly means freezing the definitions, filling the unlike column, and changing one process you own.
PostToMarket includes inventory, bulk CSV and Excel import, AI listing copy, queues, analytics, a leads inbox, and AutoPost. Use Starter at $59 per month if you are a solo seller or independent lot with one login. Use Dealership at $299 per month when you need up to 13 team logins.
Start your 10-day free trial and check out through secure Stripe checkout. Build a record this month that next October can actually compare. Cancel anytime.